Method for constructing wind power plant with alignment constraint

Abstract

The invention relates to a method for constructing a wind farm in a predetermined space, in which at least the following consecutive steps are carried out: a) forming (GR) individual grids in the predetermined space, b) determining, for each grid, the average annual energy production (AEP-mf) of a small wind farm consisting of wind turbines at the intersections of the cells, c) selecting (Ch) a number of grids capable of maximizing the energy production, and d) selecting (Ch) a number of grids capable of maximizing the energy production. D) for each grid c in step c), determining a first layout (Alg1) of a predefined number of wind turbines on the grid, e) modifying the positions (Alg2) of the wind turbines on the grid, f) determining a deterministic layout (DispF) of the wind turbines within a predetermined space, and constructing (Const) the wind farm.

Description

Technical Field

[0001] The invention relates to a method for constructing a wind farm in a predetermined space. Background Art

[0002] In order to meet environmental challenges, wind farms have emerged. These wind farms consist of a number of wind turbines that are spaced apart from each other in a defined space. The defined space can be on land or at sea. Therefore, a distinction is made between onshore wind farms and offshore wind farms (i.e. at sea).

[0003] The wind turbines of these wind farms are typically turbines with a horizontal axis of rotation provided with a system for orienting the horizontal axis of rotation in the direction of the wind in order to maximize the energy recovered by the turbine.Sometimes, wind turbines are designed to automatically orient themselves in the direction of the wind.

[0004] Wind turbines with a vertical axis of rotation are also known, which have the advantage that they do not need to be oriented in the direction of the wind.

[0005] In a wind farm, the wakes generated by wind turbines may cause the wind speed to decrease downstream of them, thereby reducing the energy recovered by other wind turbines, especially those located downstream of the wind turbines that generated these wakes. Therefore, in order to maximize the energy recovered by the wind farm, the positioning of wind turbines in the wind farm is crucial.

[0006] Furthermore, in the site chosen for the installation of a wind farm, the local characteristics of the wind may vary. In fact, in particular, wind direction and wind speed are parameters that may change over time at the site under consideration. These characteristics can be acquired in particular by sensors positioned at the defined site and kept in this position for months or years in order to obtain sufficient statistical data to characterize the wind resource of the selected site. These sensors can in particular be anemometers positioned at a sufficient height (approximately 100 meters from the ground) to characterize the wind to which the wind turbine is subjected, i.e., for example, a wind that is substantially aligned with the rotor axis. Wind data can also be acquired by light detection and ranging (LiDAR).

[0007] Knowledge of the wind statistics for the site under consideration for a wind park makes it possible in particular to obtain the wind speed distribution, the wind direction distribution and the joint probability of occurrence of a wind speed in a given wind direction.

[0008] In order to maximize the annual power generation of the wind farm, it is therefore necessary to optimally position the wind turbines on the planned site of the wind farm.

[0009] The term "annual electricity production" or "annual energy production" is to be understood as the average total energy produced by all wind turbines in a wind farm, i.e. a wind park. This average energy is estimated taking into account statistical wind data (wind speed distribution, wind direction distribution and probability of occurrence) and is based on the duration of one year (hence the term "annual") in order to avoid seasonal effects that could distort the results. In fact, both wind speed and wind direction can vary significantly according to the season.

[0010] The average energy is obtained by knowing the wind direction distribution, wind speed distribution and the probability of occurrence of wind speed in a given wind direction.

[0011] Of course, annual energy production also depends on the wind turbines chosen.

[0012] Prior art

[0013] Many methods have been developed to determine the correct positioning of wind turbines on a planned site for a wind farm.

[0014] Patent application CN105119320 relates to a method based on evolutionary algorithms. This type of algorithm is random and uses a random process. This type of algorithm requires a large number of tests to obtain results, resulting in long calculation times and requiring a large amount of computer memory.

[0015] Patent applications CN102142103A, CN105139269 and US2016171401 relate to wind turbine positioning methods based on genetic algorithms. Genetic algorithms are a type of evolutionary algorithm. They require a large number of evaluations and cross-references of these evaluations based on real possible (continuous) data. As a result, genetic algorithms are very complex and therefore require a lot of computing time and a lot of computer memory.

[0016] The methods described in the following documents are also known:

[0017] -Tao S., Xu Q., Feijoo A., Zheng G., Zhou J., 2020. Wind farm layout optimization with a three-dimensional Gaussian Wake model. Renewable Energy, Vol. 159, October 2020, pp. 553-569: This method requires that the planned site of the wind farm is bounded by a rectangle. In other words, this method is not applicable to wind farms that contain non-convex spaces or even non-connected spaces.

[0018] -Antonini EGA, Romerp DA, Amon, CH, 2018. Continuous adjoint formulation for wind farm layout optimization. Applied Energy, Vol. 228, 15 October 2018, pp. 2333-2345: This method is based on analytical methods for evaluating wake effects. It is therefore a continuous optimization method based on CFD (Computational Fluid Dynamics) calculations that requires a lot of computing time and a lot of computer memory.

[0019] - Wagner M., Day J., Neuman F., 2012. A fast and effective local search algorithm for optimizing the placement of wind turbines. Renewable Energy, Vol. 51, March 2013, pp. 64-70: Application of this method requires that the site is bounded by a rectangle. As a result, this method is not applicable to sites with non-convex or non-connected shapes.

[0020] - Feng J., Shen WZ, 2015. Solving the wind farm layout optimization problem using random search algorithm. Renewable Energy, Vol. 78, June 2015, pp. 182-192: This method requires the site to be defined by a polyhedron. As a result, it is not applicable to sites with non-connected shapes.

[0021] -Quan N., and Kim HM, 2018. Greedy robust wind farm layout optimization with feasibility guarantee. Engineering Optimization, 6 September 2018, pp. 1152-1167. This method is based on a greedy algorithm that considers all locations within a minimum distance of already located wind turbines in the site to determine the location of the next wind turbine. As a result, this method requires a lot of computation time and a lot of computer memory, especially when the discretization accuracy of the wind farm planning site is high, the space to be placed wind turbines is large, and / or the number of wind turbines to be located is large.

[0022] Patent application WO 2022 / 028 847 from the applicant is also known, which relates to a method for constructing a wind farm. In this method, wind turbine installations are at discrete locations in a predetermined space. Although this method is satisfactory, wind turbine alignment constraints are not implemented, thereby not facilitating the installation and maintenance of wind turbines. In addition, for offshore wind turbines, the alignment of wind turbines is one of the constraints for wind farm design in order to facilitate ship navigation.

[0023] The technical problem solved by the present invention includes developing a method for constructing a wind farm, which allows wind turbines to be optimally positioned in a site with a complex shape (such as non-convex and / or non-connected areas) to maximize the total power generation of the wind farm while minimizing the required computing time and computer memory and satisfying the alignment constraints of the wind turbines in two non-zero angles to each other (i.e., two non-parallel directions). Summary of the invention

[0024] In order to solve the technical problem, the present invention relates to a method for constructing a wind farm in a predetermined space, the wind farm comprising a predefined number of wind turbines, the construction method comprising a first discrete wind speed distribution, a second discrete wind direction distribution and a probability of occurrence of each discrete wind speed value in each discrete wind direction value in the first and second discrete distributions, wherein at least the following consecutive steps are performed:

[0025] a) for each pair of predefined first and second spacings, and each pair of predefined first and second directions, forming a grid within the predetermined space, each grid being defined by a plurality of intersections between first straight lines oriented along the predefined first direction and second straight lines oriented along the predefined second direction, the first straight lines being spaced apart at the predefined second spacings along the predefined second direction, the second straight lines being spaced apart at the predefined first spacings along the predefined first direction, each grid comprising at least one cell defined by the intersections;

[0026] b) for each grid in the predetermined space, determining, based on the first discrete wind speed distribution, the second discrete wind direction distribution and the occurrence probability, an average annual power generation of a small wind farm consisting of wind turbines located at all intersections of a cell or a plurality of connected cells of the grid;

[0027] c) for each pair of predefined first and second spacings, associating at least one pair of predefined first and second directions that maximizes the annual energy production of the small wind farm;

[0028] d) for each grid corresponding to a pair of predefined first and second spacings and a pair of predefined first and second directions associated in step c), determining a first layout of the predefined number of wind turbines, each wind turbine being positioned at one of said intersections of the grid by a first positioning algorithm;

[0029] e) Next, for each grid used in step d), modify the location of the wind turbines to improve the annual energy production, and obtain a final layout of a predefined number of wind turbines on each considered grid and the annual energy production of each final layout;

[0030] f) determining a deterministic layout of the wind turbines within the predetermined space, the deterministic layout corresponding to the final layout of the grid obtained in step e) that can maximize annual power generation, and constructing the wind farm by building the wind turbines at each position of the deterministic layout within the predetermined space to generate electricity using wind energy.

[0031] Preferably, when modifying the position of the wind turbine in step e) for each grid used in step d), at least the following steps are performed:

[0032] e1) for each grid used in step d), defining a sequential order of modification of the position of the wind turbine determined by the first positioning algorithm, and repositioning each wind turbine one by one in the defined sequential order by finding the intersection of the grids that maximizes the annual energy production based on the first discrete wind speed distribution, the second discrete wind direction distribution and the occurrence probability;

[0033] e2) Repeating step e1) as needed until at the end of a certain step e1) no wind turbine is repositioned, and obtaining a final layout of a predefined number of wind turbines on each considered grid and an annual energy production of each final layout.

[0034] Advantageously, said sequential order is obtained randomly.

[0035] Preferably, the sequential order is modified in each iteration of step e1).

[0036] According to one configuration of the present invention, before step a), statistical wind data is measured in the predetermined space using a measuring device (preferably a LIDAR sensor) to determine the first and second discrete distributions and the probability of occurrence of each wind speed of the first and second discrete distributions under each wind direction.

[0037] According to an advantageous embodiment of the present invention, the predetermined space includes a non-connected area and / or a non-convex area.

[0038] According to a variant of the present invention, for each considered grid, the first positioning algorithm performs at least the following steps:

[0039] - arbitrarily define the position of the first wind turbine (P_E1) at one of the intersections of the grid;

[0040] - Next, for each wind turbine to be located, successively:

[0041] defining potential locations (PE_Ej) of wind turbines to be positioned, said potential locations (PE_Ej) being formed by the intersection points of the grid that are between a minimum distance and a maximum distance of all positioned wind turbines;

[0042] Calculating the annual energy production (Eval_AEP) of the positioned wind turbines and the wind turbines to be positioned for the defined potential locations (PE_Ej) based on the first discrete wind speed distribution (RD1), the second discrete wind direction distribution (RD2) and the occurrence probability (Prob);

[0043] Select the position of the wind turbine to be located (Pos_j) corresponding to the calculated maximum annual energy production value:

[0044] - determining said first layout (Disp_Ej) corresponding to the positions of a predefined number of wind turbines in the intersections of the considered grids within said predetermined space (Esp).

[0045] Preferably, in step c), the difference between the annual power generation of the small wind farm obtained in step b) and the annual power generation of a single wind turbine multiplied by the number of wind turbines in the small wind farm is calculated.

[0046] The present invention also relates to a wind farm obtained by using the above method for constructing a wind farm. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Other characteristics and advantages of the method and the wind farm according to the invention will become apparent on reading the following description of embodiments given by way of non-limiting examples with reference to the attached drawings.

[0048] Figure 1 An example of a construction method according to the present invention is shown.

[0049] Figure 2 An exemplary embodiment of a wind turbine repositioning step in a construction method according to the invention is shown.

[0050] Figure 3 An example of a complex site for positioning a wind turbine according to the invention is shown.

[0051] Figure 4 A variation of the first positioning algorithm according to the present invention is shown.

[0052] Figure 5 The definition of wind turbine alignment constraints in the method according to the invention is shown.

[0053] Figure 6 The discontinuity of the intersection point as a function of the grid line spacing in the method according to the invention is shown.

[0054] Figure 7 The definition of a grid in a predetermined space and the intersection points of the grid in the method according to the invention are shown.

[0055] Figure 8 The layout of wind turbines on a grid in a predetermined space is shown, which layout is obtained according to one of the steps of the method of the present invention and which is not a deterministic layout.

[0056] Fig. 9 The method according to the present invention obtains the Figure 7 and Figure 8 Deterministic layout of wind turbines on the same grid.

[0057] Fig.10 An example of searching for optimal positioning of twelve wind turbines in a predetermined non-convex space with alignment constraints in order to construct the wind turbines while maximizing annual energy production according to the present invention is shown. Specific implementation plan

[0058] To facilitate reading of this specification, some definitions are explained below.

[0059] A "greedy algorithm" is understood to mean an algorithm that involves gradually building up a local optimum solution. In the case of a wind farm, this involves positioning each wind turbine one by one until the positioning of all wind turbines is obtained in a predetermined space.

[0060] "Evolutionary algorithms" should be understood to refer to a biologically inspired algorithm, the idea of ​​which is to evolve a set of solutions to obtain better results. Therefore, these algorithms are stochastic and use random processes iteratively.

[0061] A "genetic algorithm" is understood to mean an evolutionary algorithm that uses the concept of natural selection. This type of algorithm can inter alia cross-reference or modify certain parameters of previous solutions in order to improve the result.

[0062] A "non-connected region" is a region in which there are at least two points that cannot be connected by a continuous path that is completely contained in the region. Conversely, a "connected region" is a region in which any pair of points in the region can be connected by a continuous path that is completely contained in the region.

[0063] A "convex region" is a region in which a line segment between any two points in the region is completely contained within the region. For example, a circle, square, or rectangle defines a convex region.

[0064] On the contrary, a "non-convex region" refers to a region in which there are at least two points whose connecting line segments are not completely contained in the region. For example, a region defined by a concentric outer circle and an inner circle is a non-convex region.

[0065] In this specification, "grid" should be understood to refer to a plurality of intersections located within a predetermined space (or at the boundary of the space). The intersection of a grid refers to a point located at the intersection of the following straight lines: a first straight line, which is parallel to each other and oriented along a predefined first direction; and a second straight line, which is also parallel to each other and oriented along a predefined second direction. The predefined first and second directions are defined relative to a global reference system; they are usually different from the wind direction (but may be collinear with the wind direction under certain conditions). The first straight line is spaced apart along the second straight line at a predefined second spacing. In other words, the first straight line is evenly spaced apart along the second straight line at a step size corresponding to the second spacing (they are equidistant from each other). The second straight line is spaced apart along the first straight line at a predefined first spacing. In other words, the second straight line is evenly spaced apart along the first straight line at a step size corresponding to the first spacing (they are equidistant from each other). Therefore, each grid contains at least one unit cell, which is in the shape of a parallelogram, defined by the intersection of the first straight line and the second straight line (the first straight line and the second straight line intersect each other).

[0066] The invention relates to a method for constructing (or installing) a wind farm in a predetermined space, which may be an onshore area or an offshore area.

[0067] A wind farm consists of a predefined number of wind turbines. In other words, the (predefined) number of wind turbines to be installed in a predetermined space has been defined in advance.

[0068] Advantageously, the predetermined space may include non-connected areas. The predetermined space may thus correspond to an actually planned site for installing a wind turbine, such as a wind farm planning site comprising two areas separated by a road of relatively large width (several meters or even tens of meters, such as a highway) or a river.

[0069] Preferably, in addition to or instead of the non-connected areas, the predetermined space may also include non-convex areas. The predetermined space may thus correspond to an actual site with a complex shape, such as a space defined by steep hills or cliffs, a coastline, a river channel or any other body of water. In an offshore configuration, the predetermined space may include non-convex areas, which may be determined in particular based on factors such as bathymetry, ground properties, borders with other countries, waterways, passages for cables or pipelines, etc.

[0070] Figure 3 An example of a predetermined space suitable for implementing the positioning method of the present invention (or the wind farm construction method of the present invention) is shown in a schematic and non-limiting manner.

[0071] The predetermined space may include in particular a first area Z1 and a second area Z2, which are indicated by vertical hatching. These areas Z1 and Z2 are not connected. In fact, there is a non-zero minimum distance D between the two areas Z1 and Z2. In addition, the area Z1 is rectangular. Therefore, it is convex. The area Z2 has a complex non-convex shape. In fact, if two points A and B are considered, it can be seen that part of the line segment Seg connecting the points A and B is located outside the area Z2.

[0072] In order to identify the regions Z1 and Z2 of the predetermined space in a larger third region ZE (which region ZE contains the two regions Z1 and Z2), a Boolean matrix can be used. The third region ZE has a rectangular shape, which is easier to process by computer than non-connected regions and / or non-convex regions (such as Z1 and Z2). The Boolean matrix associates each discrete value (discrete position) in the region ZE with a value that is 1 if the discrete position is in the region Z1 or Z2; if the position is outside Z1 and Z2, the value is 0. This Boolean matrix makes it possible to define the predetermined space for the method. Based on the Boolean matrix, one or more boundaries of the predetermined space can be determined. In fact, if a point has a value of 1 in the Boolean matrix and if at least one of its directly adjacent points has a value of 0 in the Boolean matrix, the point will be considered to constitute part of the boundary.

[0073] The construction method includes a first discrete wind speed distribution, a second discrete wind direction distribution, and the probability of occurrence of each discrete wind speed value in each discrete wind direction value in the first and second discrete distributions. These data can be obtained in particular by a wind data collection device, such as a LIDAR (Light Detection and Ranging) sensor, a wind pole or anemometer. To this end, the construction method can advantageously include a preliminary collection step, which is performed before step a), during which a wind data measurement device (or a collection device) is installed at a physical location (a planned onshore installation area or a planned offshore area) in a predetermined space and lasts for a predetermined time to collect wind data at the site, and these measuring devices thereby measure statistical wind data to determine the first and second discrete wind speed and wind direction distributions, and the probability of occurrence of each wind speed in each wind direction in the first and second discrete distributions. The wind data can be measured within a predetermined duration to collect the measurement values, and the predetermined duration can be at least one year to obtain data related to the four seasons. Therefore, a step of collecting statistical wind data using at least one collection device in a predetermined space can be provided.

[0074] The measuring (collecting) device may advantageously be a LIDAR sensor.

[0075] In the method, at least the following consecutive steps are performed:

[0076] a) for each (first) pair of predefined first and second spacings, and for each (second) pair of predefined first and second directions, a grid is formed within the predetermined space, each grid being defined by a plurality of intersections between a first straight line oriented along the predefined first direction and a second straight line oriented along the predefined second direction, the first straight lines being spaced apart at the predefined second spacings along the predefined second direction, the second straight lines being spaced apart at the predefined first spacings along the predefined first direction, the grid comprising at least one cell defined by the intersections (each cell being defined by four intersections, and some intersections may not be associated with any cell). The intersections of the grid thus form a discrete net. A discrete net is understood to mean that the intersections of the net formed by the grid are discrete values ​​(as opposed to continuous values);

[0077] b) for each grid within the predetermined space, determining, based on the first discrete wind speed distribution, the second discrete wind direction distribution and the occurrence probability, an average annual power generation of a small wind farm consisting of wind turbines located at all intersections of a cell or a plurality of connected cells of the grid;

[0078] c) for each (first) pair of predefined first and second spacings, associating at least one (second) pair of predefined first and second directions that maximizes the annual energy production of the small wind farm;

[0079] d) for the (first) pair of predefined first and second distances associated with step c) and

[0080] (second) for each grid corresponding to the predefined first and second directions, determining a first layout of the predefined number of wind turbines, each wind turbine being positioned at one of the intersections of the grid by a first positioning algorithm;

[0081] e) Next, for each grid used in step d), the location of the wind turbines is modified to improve the annual energy production, and a final layout of a predefined number of wind turbines on each considered grid and the annual energy production of each final layout are obtained;

[0082] f) determining a deterministic layout of the wind turbines within the predetermined space, the deterministic layout corresponding to the final layout of the grid obtained in step e) that maximizes annual power generation, and constructing the wind farm by building the wind turbines at each position of the deterministic layout within the predetermined space to generate electricity using wind energy.

[0083] The use of discrete values ​​for wind speed, wind direction and for discrete intersections of each grid in the predetermined space allows this method to simplify the process, speed up the calculation time and limit the required computer memory compared to the method using continuous real data. In fact, the discrete values ​​limit the number of possible combinations (see combinatorial method), while continuous real data (see continuous method) provide an unlimited number of solutions. The combination of the first and second discrete distributions with the grid intersections allows to obtain a high accuracy of the location of each wind turbine in the predetermined space while taking into account the wind turbine alignment constraints and limits the calculation time required to determine these positions.

[0084] The probability of occurrence of each wind speed in each wind direction is used in particular to calculate the average annual energy production. This probability may in particular come from a wind rose diagram corresponding to a predetermined space, which is well known to the person skilled in the art. To establish this wind rose diagram, use may in particular be made of a collection device as described above, such as an anemometer mounted on a mast at a sufficient height (e.g. 80 to 120 m, so as to be substantially flush with the hub of the wind turbine), or by means of a LiDAR (acronym for Light Detection and Ranging) sensor positioned near the ground and oriented vertically. The collection device is kept in place for a relatively long time, several months, ideally more than a year, in order to be able to take into account the variation of wind characteristics with the seasons.

[0085] According to one embodiment, the annual power generation can be estimated using the following formula:

[0086] [Mathematics 1]

[0087]

[0088] Where aep is the annual power generation of the wind farm.

[0089] and

[0090] is the total power generated by a wind farm (consisting of a predefined number of wind turbines in a predetermined space) with respect to wind speed w s and wind direction p The expected value of the joint probability law of . Therefore, the total power generated by the wind farm takes into account the wind speed w s In each wind direction w p The statistical distribution of , for example, can be represented by a Weibull distribution.

[0091] If the rotors of the wind turbines are systematically arranged in a direction opposite to the wind direction w p On an orthogonal plane, the total power generated by the wind farm can be written as:

[0092] [Mathematics 2]

[0093]

[0094] Where N is the predetermined number of wind turbines in the wind farm in the predetermined space, Pf is the power of each wind turbine f in the wind farm in each wind direction w p For each wind speed w s The instantaneous power provided.

[0095] If some of the wind turbine's rotors are in a position relative to the wind direction w p If the wind turbine is not facing the wind direction, but is off-axis relative to the wind direction, a correction factor may be taken into account to take into account the off-axis effect. This correction factor may be obtained in particular from a CFD (Computational Fluid Dynamics) simulation.

[0096] The instantaneous power P of each wind turbine f in the wind farm f It can be expressed as:

[0097] [Mathematics 3]

[0098]

[0099] Where ρ is the air density, S is the rotor swept area of ​​the wind turbine f, and v f (w s ) is the wind speed at the wind turbine rotor, C pf is the power coefficient of wind turbine f, which depends on the wind speed v at the rotor of wind turbine f f (w s ), and the power coefficient C pf is a characteristic of the wind turbine f.

[0100] In fact, the wake effects of wind turbines located upstream and / or to the side of wind turbine f may affect the power generation of wind turbine f. The wake may cause a drop in wind speed and / or wind turbulence at wind turbine f. The impact of these wake effects is that the wind speed v at the wind turbine rotor f No longer related to wind speed w s Correspondingly, this leads to a power coefficient C pf Also affected.

[0101] The influence of the wake effect considered in equation [mathematical 3] may be based in particular on wake models. These wake models may reflect in particular:

[0102] - a reduction in the wind speed upstream of wind turbine f due to the wake of the upstream wind turbine. Such a model is the Jensen wake model, which is well known to the person skilled in the art (the model is described in the publication “A simple model for Cluster Efficiency” by Katic, Hojstrup and Jensen, published in EWEC 1986, in particular in section 2.1 of the publication),

[0103] - the turbulence intensity of the wind increases,

[0104] - and / or the superposition of the wakes of multiple upstream wind turbines on the same wind turbine f, as described in the publication "A note on wind generator interaction" Jensen, DTU, 1983. This superposition of wakes may combine the effects of reduced wind speed mentioned in the above publication and / or the turbulence intensity of the wind from multiple wakes.

[0105] The wake model may also be determined based on CFD (Computational Fluid Dynamics) calculations.

[0106] The above formula can be used to calculate the annual energy production of a wind farm consisting of a predefined number of wind turbines, a small wind farm, or a wind farm consisting of a single wind turbine.

[0107] Preferably, when modifying the position of the wind turbine in step e) for each grid used in step d), at least the following steps may be performed:

[0108] e1) for each grid used in step d), defining a sequential order of modification of the position of the wind turbines determined by the first positioning algorithm, and repositioning each wind turbine one by one in the defined sequential order by finding the intersection of the grids that maximizes the annual energy production based on the first discrete wind speed distribution, the second discrete wind direction distribution and the occurrence probability.

[0109] e2) Repeating step e1) as needed until at the end of a certain step e1) no wind turbine is repositioned, and obtaining a final layout of a predefined number of wind turbines on each considered grid and an annual energy production of each final layout.

[0110] Figure 1 The method of constructing a wind farm in a predetermined space Esp is generally shown in a schematic and non-limiting manner.

[0111] Based on the selected predetermined space Esp (geographical area at sea or on shore), which may include non-convex and / or non-connected areas (for example, roads or rivers passing through the space), and each first pair C1 predefined first and second spacings and each second pair C2 predefined first and second orientations, a grid (GR) is established in the predetermined space.

[0112] Each of these grids (GR) is established by a large number of first straight lines that are parallel to each other and evenly spaced, and second straight lines that are parallel to each other and intersect the first straight lines (in other words, the angle formed between the first straight lines and the second straight lines is non-zero and non-straight).

[0113] The individual grids GR are distinguished from each other by a combination of the first pair C1 and the second pair C2.

[0114] For each grid GR, a first straight line is oriented at an angle corresponding to the first direction relative to a direction of a predefined fixed reference system (e.g., a reference system with north and east directions), and a second straight line is oriented at an angle corresponding to the second direction relative to the direction of the same predefined fixed reference system (e.g., north).

[0115] The first straight lines are spaced apart at a predefined second interval along the second straight line, and the second straight lines are spaced apart at a predefined first interval along the first straight line.

[0116] The first straight line and the second straight line can be positioned arbitrarily in space, and these straight lines then serve as references for the positions of the other straight lines.

[0117] The intersection points between the first straight lines and the second straight lines that are located within a predetermined space (the boundaries of which are advantageously included or, on the contrary, may be excluded) define a grid.

[0118] Thus, the grid may comprise cells, and taking into account the parallelism of the first straight line and the second straight line, the cells of the grid form a parallelogram, defined by the straight lines and the intersection point.

[0119] For each defined grid GR, the annual power generation AEP-mf of the small wind farm is determined. The annual power generation is based on statistical wind data from, for example, a LIDAR sensor or any other wind data measurement device, which includes a first discrete wind speed distribution RD1, a second discrete wind direction distribution RD2, and the probability Prob of occurrence of each wind speed value of the first discrete distribution RD1 under each wind direction of the second discrete distribution RD2.

[0120] A small wind farm consists of a grid section containing at least one cell (or multiple connected cells) of the grid, and at all intersections of this grid section there are wind turbines located. Thus, if the grid section contains only one cell, the small wind farm will consist of four wind turbines, one at each intersection of the cells. This case of a small wind farm consisting of four wind turbines is advantageous because it allows a simple and quick assessment of the mutual influence of the wakes of the wind turbines on the grid, without requiring a lot of memory or computing time when implementing this step on a computer.

[0121] Within the meaning of the present description, a cell is understood to mean an elementary cell: this means that a parallelogram defined by four intersection points is considered to be a cell if it contains no other parallelogram defined by four intersection points within itself.

[0122] By comparing the annual energy production values ​​of different grids comprising the same first pair C1 and respective second pairs C2, the most promising pair or pairs C2 can be associated (i.e. giving the maximum annual energy production or pairs). Thus, for each first pair C1 predefined first and second spacings, one or more second pairs C2 can be associated. This makes it possible to limit the number of combinations of first pairs C1 and second pairs C2 in the rest of the method.

[0123] The annual energy production values ​​can be compared directly (directly comparing the different values ​​and then finding the second pair C2 that maximizes the annual energy production) or indirectly. In the latter case, the annual energy production of the small wind farm is compared with the annual energy production of the same number of wind turbines, assuming that each wind turbine is independent (therefore not taking into account the wake effect): the losses related to the wake effect of the small wind farm will then be estimated. In this case, of course, one or more second pairs C2 that minimize the losses related to the wake effect will be retained.

[0124] Therefore, at least one second pair C2 can be selected for each first pair C1 based on the annual power generation of the small wind farm. Therefore, a certain number of grids are selected from the previously established grids.

[0125] Next, for each selected grid, a first layout Alg1 of wind turbines on the grid (located at the intersection of the grid) is determined. The first layout aims to obtain good energy output, so the first discrete distribution RD1, the second discrete distribution RD2 and the occurrence probability Prob are considered.

[0126] Based on this first layout of wind turbines on each selected grid, it is then sought to increase energy recovery. Then, by relocating the Alg2 wind turbines on the grid, each time the wind turbine is relocated, the location that allows the highest annual energy production is sought. Of course, in order to achieve this, the first and second discrete distributions RD1 and RD2 and the probability of occurrence Prob need to be taken into account.

[0127] Once the wind turbines are relocated and no more optimal solution can be found (i.e., the location of the wind turbines on the grid cannot be changed without reducing the annual energy production), the final layout of the wind turbines on each selected grid can be determined and the annual energy production AEP_g of each grid can be compared. Then, the deterministic layout Disp_F of the wind turbines within the predetermined space Esp can be determined as the final layout of the wind turbines on the grid that maximizes the annual energy production.

[0128] Then, the Const wind turbines are built at the planned site corresponding to the deterministic layout determined within the predetermined space.

[0129] Step a) Define the grid

[0130] This step makes it possible to define a plurality of possible grids within the predetermined space. Each grid is therefore spatially limited and cannot exceed the predetermined space. The grid comprises one or more cells separated by intersections. The cells have a parallelogram shape, such as a rectangular or square shape, and the intersections are located at the boundaries between the cells. All cells and all intersections of each grid are located within the predetermined space.

[0131] Each grid cell is defined between the first line and the second line.

[0132] Each of these grids constitutes a possible option for arranging wind turbines in a wind farm in a predetermined space, taking into account the alignment constraint (along the first straight line and the second straight line). The intersection of each grid constitutes a possible position for the installation of the wind turbine. By using multiple grids, multiple alignment constraints can be tested (different predefined first and second spacing values ​​and different predefined first and second direction values). The first direction and the second direction are non-collinear and form a non-zero (and non-straight) angle, so that the two-dimensional alignment constraint in space is taken into account, so that the first straight line and the second straight line intersect.

[0133] The use of discrete values ​​limits the number of possible combinations and thus the useful memory of the computer. Furthermore, these discrete values ​​are sufficient to ensure an accuracy consistent with the construction of wind turbines on site (taking into account possible construction errors).

[0134] The predefined first and second spacings are preferably between a minimum distance and a maximum distance, these minimum and maximum distances depending on, for example, the rotor diameters of the wind turbines. The minimum distance corresponds to the minimum value by which two consecutive wind turbines must be separated in order to limit energy losses associated with wake effects, and the maximum distance corresponds to the maximum value by which two wind turbines must be separated, a larger distance not bringing any gain in energy recovery and limiting the possibility of installing enough wind turbines in a predetermined space to ensure a sufficient energy output.

[0135] In fact, it is not necessary to predefine the first and / or second spacing to be smaller than the minimum distance or larger than the maximum distance, since the wind turbines have to comply with a minimum distance between each other in order to avoid excessive energy losses related to wake effects, and the maximum distance has to be complied with in order to install a predetermined number of wind turbines in a space and obtain a beneficial energy output from the wind farm.

[0136] Preferably, the minimum distance may be greater than two times the wind turbine rotor diameter, preferably greater than four times the wind turbine rotor diameter; the maximum distance may be at least four times the wind turbine diameter, preferably at least eight times the wind turbine diameter.

[0137] Preferably, the predefined first direction may be between -90° and 90°, taking into account symmetry relative to an orthogonal reference system (eg, a terrestrial reference system with 90° representing north and -90° representing south), which is preferably a reference system for defining the wind direction.

[0138] And preferably, the predefined second direction may be between -90° and the first direction. Therefore, through the symmetry effect, the entire possible space is swept while limiting the number of combinations.

[0139] Discrete values ​​of the predefined first and second spacings may be defined in steps of 0.5 times the wind turbine rotor diameter, and discrete values ​​of the predefined first and second directions may be defined in steps of 1°, which are sufficient to obtain sufficient accuracy at the wind turbine position.

[0140] Figure 5 The transformation of the reference system between the (O1, N, E) reference system and the reference system bound to the grid is shown in a schematic and non-restrictive manner, wherein the reference system (O1, N, E) corresponds to the global reference system, O1 can be set arbitrarily, N may correspond to the north direction, and E may correspond to the east direction; in the reference system bound to the grid, O2 is a point arbitrarily fixed at the intersection of the first straight line Lig1 and the second straight line Lig2 of the grid, and the first straight line Lig1 and the second straight line Lig2 intersect each other. Here, the first straight line Lig1 forms an angle θ1 with the direction E, and the second straight line Lig2 forms an angle θ2 with the direction E (or, the first straight line Lig1 can also form an angle θ1 with the direction N, and the second straight line Lig2 forms an angle θ2 with the direction N). The reference system (O1, N, E) can be preferably used to determine the wind direction.

[0141] Figure 6 The discontinuities in position that may result from different grids of different networks are shown in a schematic and non-limiting manner.

[0142] In this figure, Ω represents a predetermined space. For ease of understanding, only one line is shown in the grid.

[0143] In figure a), the spacing between the intersection points on the lines is L. Therefore, a wind turbine (indicated by a black dot) can be placed at each end point of the predetermined space Ω: the spacing between them is 2L.

[0144] In diagram b), the intersection points on the lines are spaced L+ε apart. Therefore, wind turbines (represented by black dots) can only be installed at the intersection points shown: the spacing between them is L+ε. In fact, the points located at 2(L+ε) are outside the range of the predetermined space Ω.

[0145] Therefore, even for small values ​​of ε, significant solution discontinuities may occur between grid lines spaced L and grid lines spaced L+ε due to the constraints of the predetermined space used to determine the grid and the intersection points selected with regard to possible wind turbine locations.

[0146] Step b) Determine the annual power generation of small wind farms on each grid

[0147] Step a) enables the definition of a grid panel, where the intersection of each grid is a point at which a wind turbine can be mounted. Of course, in order to verify the alignment constraints of the wind turbines, in this construction method it is not possible to use the intersection of one grid to mount some wind turbines and the intersection of another grid to mount other wind turbines.

[0148] In order to avoid finding the best layout of wind turbines on each grid each time by calculating the annual energy production, it is preferred to perform a sorting so as to use only those grids that are likely to provide the best layout. Step b) aims at performing this first sorting of the individual grids.

[0149] To this end, for each grid formed in step a), a small wind farm is considered, which includes wind turbines at each intersection of the considered grid portion. The grid portion corresponds to several connected cells (several cells are connected to each other) and is preferably a single grid to further simplify the sorting.

[0150] With the help of all wind turbines installed on the grid portion (for example four wind turbines if the grid portion comprises a single cell; 9 wind turbines if the grid portion is scaled up by a factor of 2 compared to a grid with a single cell), the annual energy production of the small wind farm can be determined, in particular calculated by the above formula.

[0151] The annual energy production of small wind farms can be associated for each type of grid.

[0152] Step c) selects a (second) pair of predefined first and second directions according to the predefined first and second spacings.

[0153] This step is used to limit the number of combinations of the predefined number of wind turbine layouts for arranging the wind farm based on the results obtained in step b). In fact, with the small wind farm, it is possible to estimate how the annual energy production varies with the (second) pair of predefined first directions and second directions, the first and second discrete wind speed and wind direction distributions, and the probability of occurrence of each discrete wind speed value in each discrete wind direction value for each (first) pair of predefined first spacings and second spacings.

[0154] Then at least one (preferably only one) (second) pair of predefined first and second directions may be selected for each (first) pair of predefined first and second spacings, thereby limiting the number of remaining combinations. In practice, for the (first) pair of predefined first and second spacings, the selected one (or more) (second) pair of predefined first and second directions is able to maximize the annual energy production and / or limit the wake effect.

[0155] In order to determine one (or more) (second) pairs of first and second directions associated with each (first) pair of predefined first and second spacings, the annual power generation can be directly compared and the pairs of directions that can maximize the annual power generation of the small wind farm can be retained, or the recovery energy loss caused by the wake effect can be compared.

[0156] In this second case, the energy loss of the considered small wind farm is determined relative to the ideal power generation of the same number of wind turbines, without taking into account the wake effect between the wind turbines. Thus, in step c), it is possible, for example, to calculate the difference between the annual power generation of the small wind farm obtained in step b) and the annual power generation of a single wind turbine multiplied by the number of wind turbines in the small wind farm. This energy loss P loss This can be determined using the following formula:

[0157] [Mathematics 4]

[0158] P loss =n*aep(f1,ws,wp)-aep(f2,ws,wp)

[0159] Wherein, aep(f1,ws,wp) is the annual power generation of a wind farm f1 consisting of a single wind turbine in a predetermined space calculated according to formula [Math. 1], and aep(f2,ws,wp) corresponds to the annual power generation of a wind farm f2 consisting of n wind turbines in a small wind farm in a predetermined space according to formula [Math. 1], ws and wp are the statistical distributions of wind speed and wind direction, respectively, taking into account the above-mentioned probability of occurrence.

[0160] The advantage of this formula [Math 4] is that it directly provides information on the impact of the wake effect on a small wind farm, which directly depends on the chosen alignment and spacing. Solutions with excessive losses can then be discarded, leaving only one or more relevant solutions (with the lowest energy losses).

[0161] Step d) determining a first layout of wind turbines in the wind farm

[0162] This step consists in determining a first layout of wind turbines on each grid defined by each (first) pair of predefined first and second spacings and each (second) pair of associated predefined first and second directions obtained in the previous step c), so as to limit the number of combinations. Each wind turbine is positioned at an intersection of a predetermined space by a first positioning algorithm (for obvious construction reasons, only one wind turbine can be installed at each intersection). Preferably, the first positioning algorithm can be an optimization algorithm that obtains a first distribution that obtains a satisfactory annual energy production. Preferably, the first positioning algorithm can be a greedy algorithm that positions each wind turbine one by one to maximize the annual energy production generated by each newly added wind turbine. The first wind turbine can be positioned arbitrarily in the predetermined space (at a discrete value of the first discrete network). The second wind turbine will be positioned at the intersection of the considered grid so that the annual energy produced by the two wind turbines can be maximized, and the position of the nth wind turbine is chosen at the intersection of the grid so that the annual energy produced by the n wind turbines can be maximized. The use of a greedy algorithm makes it possible to easily obtain a first layout of wind turbines in a predetermined space, thereby making it possible to initialize the optimization method of step e), in particular by means of a local optimization search method of the subsequent step e1).

[0163] According to one embodiment, for each grid considered (each grid defined by a first pair of predefined first and second spacings and a second pair of predefined first and second directions associated with the first pair of predefined first and second spacings), the first positioning algorithm performs at least the following steps:

[0164] - arbitrarily defining the position of the first wind turbine at one of the intersection points of the considered grids;

[0165] - Next, for each wind turbine to be located, successively:

[0166] Define potential locations for wind turbines to be located, which are formed by the intersections of the grid. In other words, select the grid intersections where the next wind turbine will be located first.

[0167] The annual energy production of the positioned wind turbines and the wind turbines to be positioned is calculated for the defined potential locations. Thus, an annual energy production is associated with each defined potential location. Furthermore, the calculation of the annual energy production takes into account the first discrete wind speed distribution, the second discrete wind direction distribution and the probability of occurrence. The calculation also involves the characteristics of the wind turbine in a known manner, in particular the swept area of ​​the wind turbine rotor, the drag coefficient and / or the power coefficient.

[0168] The position of the wind turbine to be positioned corresponding to the calculated maximum annual energy production value is selected. The annual energy produced by the wind turbines whose positions are defined in the predetermined space is thus maximized. These defined positions will serve as a basis for determining the position of the next wind turbine, in particular for defining the potential position of the next wind turbine to be positioned.

[0169] - determining a first layout corresponding to the positions of a predefined number of wind turbines in the intersections of the considered grid within a predetermined space.

[0170] Thus, the first positioning algorithm is a greedy algorithm comprising a step of arbitrarily positioning a first wind turbine, followed by iterative positioning of further wind turbines at the considered grid intersections until all predefined number of wind turbines have been positioned on the grid. This greedy algorithm, by means of the step of selecting potential positions, makes it possible to accelerate the calculation time while conveniently positioning the wind turbines. Such an algorithm makes it possible to obtain a first layout suitable for positioning in subsequent steps to find a local optimization of each wind turbine.

[0171] Figure 4 An exemplary step of obtaining a first layout of wind turbines on each selected grid according to the invention is shown schematically and non-limitingly.

[0172] For this purpose, for example, a greedy algorithm can be used.

[0173] For each selected grid G(C1, C2) in the predetermined space, a position P_E1 of a first wind turbine is defined, for example arbitrarily.

[0174] Next, for each wind turbine j, a location is found to maximize the annual energy production.

[0175] Therefore, iteratively according to F3, potential positions PE_Ej of wind turbines j to be located are defined one by one, and the potential positions PE_Ej are determined by the available intersection points of the predetermined spatial grid (ie the positions of wind turbines that are not yet planned).

[0176] Once these potential positions PE_Ej are defined for the wind turbines j to be positioned, an evaluation Eval_AEP is performed to determine the annual energy production generated in a predetermined space by the already positioned wind turbines and the wind turbines j to be positioned at each potential position PE_Ej. In this evaluation Eval_AEP, in particular the first and second discrete wind speed and wind direction distributions RD1 and RD2 are used, as well as the probability of occurrence Prob of each wind speed in each wind direction. The characteristics of the wind turbines and the wake effect described above can be used in a known manner.

[0177] Next, the position Pos_j of wind turbine j may be selected, which corresponds to the maximum value of the annual energy production in the previous step.

[0178] A new layout Disp_Ej of the positioned wind turbines (including wind turbine j) can then be defined in the predetermined space. This new layout Disp_Ej will be used in the next iteration to determine potential locations PE_Ej for new wind turbines and to estimate the annual energy production Eval_AEP.

[0179] The loop F3 is executed with j from 1 to N-1, where N is the predetermined number of wind turbines in the predetermined space, taking into account that the first wind turbine is P_E1 which is positioned stepwise.

[0180] Once all wind turbines have been located (ie when j=N-1), the final layout Disp_Ej for each grid is found to correspond to the first layout Disp1 of wind turbines for each grid.

[0181] Step e) Improve the location of wind turbines on each grid

[0182] For each grid in step d), it is sought to increase the annual energy production by modifying the positions of the wind turbines on the grid. This can be achieved in particular by steps e1 and e2 described below.

[0183] Step e1) Local search optimization

[0184] In this step, it is sought to increase the annual energy production by modifying the positions of the wind turbines in the wind farm on each grid in step d) one by one.

[0185] To this end, a sequential order of modification of the wind turbine positions determined by the first positioning algorithm is defined: in other words, the wind turbines do not necessarily have to be repositioned in the order of their positions determined by the first algorithm, and preferably a different order is used to improve the chances of obtaining a more optimal solution.

[0186] Advantageously, the sequence order is obtained randomly. In fact, using a random function for the sequence order improves the quality of the optimization by avoiding optimization paths based on a preset sequence (these paths may bias the optimization results).

[0187] Once the sequential order has been defined, each wind turbine is repositioned one by one in accordance with the defined sequential order, by finding the available intersection of the grid that maximizes the annual energy production: if no other grid intersection can improve the annual energy production, the wind turbine in question is kept in its previously determined position. Then an attempt can be made to modify the position of the next wind turbine in the defined sequential order.

[0188] In order to determine the annual energy production, of course, the first and second discrete wind speed and direction distributions and the probability of occurrence must be taken into account, which directly depend on the geographical location of the predetermined space and its local environment (such as forests, mountains, geological features, the presence of buildings) and the characteristics of the wind turbine.

[0189] The annual power generation can be determined using in particular the above formula, in particular using formula [Math 1].

[0190] Step e2) repeats step e1)

[0191] At the end of step e1), it is not known in advance whether the layout of the wind turbines in each grid is optimal or whether it can still be improved. In order to further optimize the layout of the wind turbines in each grid, step e1) is repeated as needed, as long as at least one wind turbine is relocated to an intersection of the grids during this step. The iteration is stopped when no wind turbine is relocated in the last iteration of step e1): it is then considered that the optimal layout of a predetermined number of wind turbines has been found on the grid under consideration.

[0192] At the output of step e2), for each considered grid (corresponding to a first pair of predefined first and second spacings and a second pair of predefined first and second directions associated with the first pair of first and second spacings), the final layouts of wind turbines on each considered grid and the annual energy production associated with these final layouts are obtained (one final layout for each considered grid).

[0193] Preferably, in step e2), the sequence order is modified in each iteration of step e1) to restrict the optimization path based on a preset order.

[0194] Figure 2 An example of an optimization search according to the method of the present invention is shown schematically and non-limitingly.

[0195] For each selected grid, a first layout Disp1 of wind turbines on each of these grids is determined Alg1, taking into account as input data at least a first discrete wind speed distribution RD1, a second discrete wind direction distribution RD2 and the probability of occurrence Prob of each wind speed in each wind direction. The characteristics of the wind turbines and the wake model can also be used to determine the annual energy production, thereby establishing this first layout Disp1.

[0196] Data on wind speed, wind direction and the probability of occurrence of each wind speed value in each wind direction can be obtained in a preliminary step by a collecting device in particular. These data can be used in particular to establish a wind rose diagram, which is known to those skilled in the art.

[0197] This first layout Disp1 may be improved, but its quality is good enough to allow local optimization in subsequent steps. The step Alg1 of determining a first layout Disp1 of wind turbines on each selected grid may be a greedy algorithm.

[0198] This first layout Disp1 obtained quickly by the greedy algorithm is used as input data for the subsequent optimization step Alg2. This optimization step Alg2 modifies the positions of each wind turbine in the first layout Disp1 to other intersections of the same grid one by one, so as to increase the annual energy production of the wind farm by testing other possible positions on the grid, while thus complying with the alignment constraint.

[0199] In more detail, the optimization step Alg2 comprises the following steps:

[0200] - A sequential order OS of wind turbine position modifications is defined one by one. This order OS can in particular be obtained using a random function.

[0201] - Next, iteratively, for each selected grid, modifying the position of at least one wind turbine i, preferably each wind turbine i, by performing the following sub-steps:

[0202] * Determine the possible positions PDP_i of wind turbine i, which are the current position of wind turbine i and the grid intersections where no other wind turbines are planned (other wind turbines remain in their positions, either their initial positions from the first layout, or positions that have been relocated).

[0203] *An evaluation Eval_i is then performed to determine the annual energy production for each possible layout (for each possible position PDP_i of wind turbine i, the other wind turbines remain in their last set positions). The annual energy production takes into account the first and second discrete wind speed and wind direction distributions RD1 and RD2, as well as the probability of occurrence Prob of each wind speed in each wind direction. Thus, at the end of this evaluation step, the annual energy production corresponds to each possible position PDP_i of wind turbine i to be relocated.

[0204] * Keep as the possible position POS_i of wind turbine i the position PDP_i corresponding to the maximum value of the annual energy produced in step Eval_i.

[0205] * A new layout Disp_N of wind turbines in the determined space is thus obtained. This new layout comprises the last positions of the wind turbines that have been located and the new position Pos_i of wind turbine i.

[0206] The loop B1 then makes it possible to select the next wind turbine F1 in a defined sequential order (i becomes i+1) in order to carry out the same procedure for the following wind turbines.

[0207] Once all wind turbines have been relocated through loop B1, the steps of determining the order OS and loop B1 will be repeated multiple times for each selected grid for each wind turbine one by one. Loop B1 includes: for each wind turbine i to be relocated, determining possible positions PDP_i, evaluating the annual energy production Eval_i of each possible position, selecting the position Pos_i of the wind turbine i to be relocated, and defining a new layout Disp_N.

[0208] Repeating these steps multiple times can increase the annual power generation of the wind farm. Modifying the order of the sequence each time (for example, by random selection each time) can further increase the annual power generation.

[0209] The loop B2 ends when no wind turbine has been repositioned during the last loop B1 (all wind turbines are kept in the same position as obtained in the previous iteration).

[0210] Once loop B2 has ended, the last obtained layout Disp_N becomes the final layout DispM of the selected grid.

[0211] To select a deterministic layout of wind turbines, the final layouts of the various grids can be compared and the layout that maximizes the annual power generation can be selected. It is this final layout that maximizes the power generation that is retained and becomes the deterministic layout. Wind turbines can then be built at the physical site of the predetermined space according to the deterministic layout, thereby obtaining a wind farm.

[0212] Step f) Determine the deterministic layout of wind turbines in a predetermined space and build a wind farm

[0213] Based on each final layout of wind turbines obtained on each grid, the annual power generation of these final layouts is compared, and the final layout with the highest annual power generation is selected to maximize the output of the wind farm. The deterministic layout of the wind turbines of the wind farm is then determined to correspond to the optimal final layout.

[0214] Based on the obtained deterministic layout, the wind turbines of the wind farm can then be constructed in the predetermined space at the positions corresponding to the deterministic layout. With the use of a grid, the wind turbines will be aligned along two directions and straight lines evenly distributed in space, thereby satisfying the operational constraints and constraints dependent on the onshore operation.

[0215] In the above method, steps a) to f) (except the part of step f) involving the construction of the wind farm) can be implemented by a computer, a server or a supercomputer. Steps a) to f) then constitute a method for positioning a wind turbine in a predetermined space.

[0216] The present invention may also relate to a computer program product, which implements a method for locating a wind turbine in a predetermined space, comprising the above steps a) to f) (except the part of step f) involving the construction of a wind farm), using a computer device, such as a computer, a mobile phone or a tablet. The computer program product can be downloaded through a communication network and / or recorded on a medium that can be read by a computer and / or executed by a processor or a server, and contains program code instructions, and when the program is executed on a computer or a mobile phone, the positioning method is implemented according to one of the above characteristics. In fact, the above positioning method is particularly suitable for implementation on a computer device. Therefore, it can be easily implemented and the results can be obtained quickly.

[0217] The present invention also relates to a wind farm obtained by using the above method for constructing a wind farm, which can have good energy output by considering wind-related statistics of the considered location and considering alignment constraints of wind farm construction.

[0218] Figure 7 An example of creating a grid in a predetermined space is shown schematically and non-limitingly.

[0219] The predetermined space Esp includes two disconnected and non-convex regions Z1 and Z2.

[0220] In order to obtain a grid defined by a first pair of first and second spacings R1 and R2 and a second pair of first and second directions θ1 and θ2, the space is divided by a first straight line L1 and a second straight line L2. The first straight lines L1 are parallel to each other and are evenly spaced at the second spacing R2 along the second straight line L2, and the second straight lines L2 are parallel to each other and are evenly spaced at the first spacing R1 along the first straight line L1. In addition, the first straight line L1 is oriented along a first direction, which forms an angle θ1 with the direction E of a reference system (O1, N, E), where the reference system (O1, N, E) is an arbitrarily defined reference system, for example, O1 is a selected geographical point, N is north, and E is east. The second straight line L2 is oriented along a second direction, which forms an angle θ2 with the direction E in the same reference system (O1, N, E). The angles θ1 and θ2 are selected so that the first straight line L1 intersects the second straight line L2.

[0221] One of the first straight line L1 and the second straight line L2 can be positioned arbitrarily in space and then serve as a reference for the position of the other first straight line L1 and the second straight line L2.

[0222] Within the predetermined space Esp (or on the boundary of this space), the intersection points (indicated by crosses) form a grid. Cells, such as the illustrated cell M1 (with shading), can be formed between the intersection points. It is worth noting that the only intersection point located in the area Z2 is still part of the grid, although it does not belong to any cell because it is not connected to any other intersection point through a cell.

[0223] The intersection points of the grid are points within the predetermined space Esp that can potentially be used to locate wind turbines.

[0224] Figure 8 In a schematic and non-limiting manner, the method according to the invention is shown in a predetermined space Esp (with Figure 7 in the same space) and in Figure 7 A first layout is obtained by positioning sixteen wind turbines on the obtained grid.

[0225] In this figure, Figure 7 The same reference numerals correspond to the same elements and will not be described in detail here. Figure 7 The cross symbol indicating the intersection is not Figure 8 The black dots represent the positions of the sixteen wind turbines in the first layout in the predetermined space, and each wind turbine is positioned at Figure 7 The intersection of the grid.

[0226] This first arrangement corresponds to the method according to the invention after step d) for Figure 7 The layout obtained by the grid.

[0227] Fig. 9 In a schematic and non-limiting manner, the method of the present invention is shown in a predetermined space Esp (with Figure 7 same space) and in Figure 7 On the obtained grid, Figure 8 The final layout obtained by repositioning sixteen wind turbines after the first layout was obtained in .

[0228] In this figure, Figure 7 or Figure 8 The same reference numerals correspond to the same elements and will not be described in detail here. Figure 7 The cross symbol indicating the intersection is not Fig. 9 The black dots represent the positions of the sixteen wind turbines in the predetermined space in the final layout after steps e1) and e2). Each wind turbine is positioned at Figure 7 The intersection of the grid.

[0229] Example

[0230] Fig.10An example of searching for wind turbine locations in a predetermined space Z2 (a non-convex space) is shown.

[0231] In this example, an attempt is made to position twelve wind turbines in a predetermined space Z2 taking into account the alignment constraints of the wind turbines in two non-collinear directions.

[0232] The wind turbines considered are rated at 10 MW, with a nacelle height positioned at 119.8 m from the ground and a rotor diameter of 198 m. In the predetermined space Z2, the distance between the wind turbines is greater than 4 times the diameter of their rotors.

[0233] The wind data used in this example corresponds to the following source: Baker, NF, Thomas, JJ, Stanley, APJ, and Ning, A., IEA Task 37 Wind Farm Layout Optimization Case Studies, https: / / doi.org / 10.5281 / zenodo.5809681, 2021.

[0234] Figures a) to h) represent different positions of wind turbines in the predetermined space Z2 based on different grids in the predetermined space Z2. For this example, different grids with the following characteristics were tested:

[0235] - the direction angle of the first straight line L1 changes from 1° to 359° with a step size of 1° relative to the reference R,

[0236] - the direction angle of the second straight line L2 changes from 0° to the direction angle of the first straight line L1 minus 1° with a step size of 1° relative to the reference R,

[0237] - a first pitch R1 of the first straight line L1 along the second straight line L2 varies from 4 times the rotor diameter to 8 times the rotor diameter with a step size of 0.25 times the rotor diameter; and

[0238] A second pitch R2 of the second straight line L2 along the first straight line L1 varies from 4 times the rotor diameter to 8 times the rotor diameter, with a step length of 0.25 times the rotor diameter.

[0239] Figures a) to h) represent only a few of the grids tested.

[0240] for Fig.10 The figures in:

[0241] - The "circular" points represent possible locations on the grid where wind turbines can be placed. They are therefore located at the intersection of the first straight line L1 and the second straight line L2 of the grid in the predetermined space Z2;

[0242] - the crosses indicate the positions of the wind turbines that maximize the energy recovery on the considered grid;

[0243] - The dotted line indicates the direction of the first straight line L1 or the second straight line L2;

[0244] - The dashed line represents a reference R for the first straight line positioning. For example, R may be a straight line representing the east-west direction.

[0245] - Spacing R1 represents the spacing of the first straight line L1 arranged along the second straight line L2;

[0246] The interval R2 represents the interval of the second straight lines L2 arranged along the first straight line L1.

[0247] The first directions θ11 , θ12 and θ13 represent directions of the first straight line L1 relative to the reference R.

[0248] Therefore, in Fig.10 In each of the figures shown in , a different grid in the same predetermined space Z2 is used.

[0249] For diagrams b), c), d) and e), the first straight line L1 is parallel to the line L1 shown in diagram b).

[0250] For Figures f), g) and h), the first straight line L1 is parallel to the line L1 shown in Figure f).

[0251] Depending on the different grids used, the optimal location of the wind turbines on each grid is different.

[0252] The average annual energy recovery amount of each positioning diagram a) to h) is between 486611 MWh and 487467.6 MWh, that is, a gain of 856.6 MWh can be obtained according to the optimal configuration of the present invention.

Claims

1. A method for constructing a wind farm in a predetermined space (Esp), the wind farm comprising a predefined number of wind turbines, the method comprising a first discrete wind speed distribution (RD1), a second discrete wind direction distribution (RD2) and a probability (Prob) of occurrence of each discrete wind speed value in each discrete wind direction value in the first and second discrete distributions (RD1, RD2), wherein at least the following consecutive steps are performed: a) predefining first and second spacings (R1, R2) for each pair (C1), and predefining first and second directions (θ1, θ2) for each pair (C2), forming a grid (GR) within the predetermined space (Esp), each grid being defined by a plurality of intersections between first straight lines (L1) oriented along the predefined first direction (θ1) and second straight lines (L2) oriented along the predefined second direction (θ2), the first straight lines (L1) being spaced apart by the predefined second spacings (R2) along the predefined second direction (θ2), the second straight lines (L2) being spaced apart by the predefined first spacings (R1) along the predefined first direction (θ1), the grid comprising at least one cell (M1) defined by the intersections; b) for each grid in the predetermined space, determining an average annual energy production (AEP-mf) of a small wind farm consisting of wind turbines located at all intersections of a cell or a plurality of connected cells of the grid based on the first discrete wind speed distribution (RD1), the second discrete wind direction distribution (RD2) and the occurrence probability (Prob); c) for each pair (C1) of predefined first and second spacings (R1, R2), associating at least one pair (C2) of predefined first and second directions (θ1, θ2) that maximize the annual energy production of the small wind farm; d) determining, for each grid corresponding to a pair (C1) of predefined first and second spacings (R1, R2) and a pair (C2) of predefined first and second directions (θ1, θ2) associated in step c), a first layout (Alg1) of said predefined number of wind turbines, each wind turbine being positioned at one of said intersections of said grids by means of a first positioning algorithm; e) Next, for each grid used in step d), modifying the location of the wind turbines to increase the annual energy production, and obtaining a final layout of the predefined number of wind turbines on each considered grid and the annual energy production of each final layout; f) determining a deterministic layout (Disp_F) of the wind turbines within the predetermined space (Esp), the deterministic layout (Disp_F) corresponding to the final layout of the grid obtained in step e) that maximizes annual energy production, and constructing (Const) the wind farm by building the wind turbines at each position of the deterministic layout (Disp_F) within the predetermined space (Esp) to generate electricity using wind energy.

2. The method for constructing a wind farm according to claim 1, characterized in that: When modifying the position of the wind turbine in step e) for each grid used in step d), at least the following steps are performed: e1) defining, for each grid used in step d), a sequential order (OS) of the modification of the positions of the wind turbines determined by the first positioning algorithm, and repositioning each wind turbine one by one in the defined sequential order by finding the intersection of the grids maximizing the annual energy production based on the first discrete wind speed distribution (RD1), the second discrete wind direction distribution (RD2) and the occurrence probability (Prob); e2) Repeating step e1) as needed until at the end of a certain step e1) no wind turbine is repositioned, and obtaining a final layout of a predefined number of wind turbines on each considered grid and an annual energy production of each final layout.

3. The method for constructing a wind farm according to claim 2, characterized in that: The sequential order (OS) was obtained randomly.

4. The method for constructing a wind farm according to claim 2 or 3, characterized in that: The sequential order (OS) is modified in each iteration of step e1).

5. A method for constructing a wind farm as claimed in any one of the preceding claims, characterized in that: Prior to step a), statistical wind data is measured within the predetermined space (Esp) using a measuring device (preferably a LIDAR sensor) to determine the first and second discrete distributions (RD1, RD2) and the probability of occurrence (Prob) of each wind speed of the first and second discrete distributions (RD1, RD2) under each wind direction.

6. A method for constructing a wind farm as claimed in any one of the preceding claims, characterized in that: The predetermined space (Esp) includes a non-connected area (Z1, Z2) and / or a non-convex area (Z2).

7. A method for constructing a wind farm as claimed in any one of the preceding claims, characterized in that: For each considered grid, the first positioning algorithm performs at least the following steps: - arbitrarily defining the position of a first wind turbine (P_E1) at one of the intersections of said grid; - Next, for each wind turbine to be located, successively: defining potential locations (PE_Ej) of wind turbines to be positioned, said potential locations (PE_Ej) being constituted by the intersection points of said grids which are between a minimum distance and a maximum distance from all positioned wind turbines; Calculating the annual energy production (Eval_AEP) of the positioned wind turbines and the wind turbines to be positioned for the defined potential locations (PE_Ej) based on the first discrete wind speed distribution (RD1), the second discrete wind direction distribution (RD2) and the occurrence probability (Prob); Select the position of the wind turbine to be located (Pos_j) corresponding to the calculated maximum annual energy production value: - determining said first layout (Disp_Ej) corresponding to the positions of a predefined number of wind turbines in the intersections of the considered grid within said predetermined space (Esp).

8. A method for constructing a wind farm as claimed in any one of the preceding claims, characterized in that: In step c), the difference between the annual power generation of the small wind farm obtained in step b) and the annual power generation of a single wind turbine multiplied by the number of wind turbines in the small wind farm is calculated.

9. A wind park obtained using the method for constructing a wind park according to any one of the preceding claims 1 to 8.

Images