Multi-grid fusion method for fan wake flow numerical simulation in complex terrain

Abstract

The invention provides a multi-grid fusion method for fan wake flow numerical simulation in complex terrains. A Cartesian grid of a complex terrain, a structural or non-structural hybrid grid of a fan area and an anisotropic O-shaped structural grid of a wake area are fused, multiple grid types of different topologies are fused, and assembly and interpolation calculation are realized based on an overlapping grid technology, so that the aerodynamic load precision of the fan and automatic modeling of terrain geometry are ensured, and the accuracy of the aerodynamic load of the fan is improved. The trailing vortex analysis efficiency is remarkably improved, the calculation cost is controlled, and the precision and efficiency of wind field wake flow numerical simulation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fan engineering calculation, in particular to a multi-grid fusion method for numerical simulation of fan wake on complex terrain. BACKGROUND

[0002] As a core component of the clean energy system, the key to the large-scale development of wind energy lies in the efficient operation of the wind farm. As a core indicator directly reflecting the operation efficiency and economic benefit of the wind farm, accurate assessment of the power generation capacity is an important prerequisite for improving the overall power generation efficiency. In the assessment of the power generation capacity of the wind farm, the wind turbine wake effect is a key influencing factor that cannot be ignored.

[0003] The wind turbine wake effect specifically refers to that when the upstream wind turbine is running, the rotating blades will disturb the airflow, thereby changing the airflow state downstream, resulting in a significant reduction in downstream wind speed and a substantial increase in turbulence intensity, ultimately causing the power generation efficiency of the downstream wind turbine to decrease. As can be seen, accurate assessment of the wake effect has important value, on the one hand, it can effectively improve the prediction accuracy of the overall power generation capacity, on the other hand, it can provide a solid theoretical support for the micro-siting of wind turbines and cluster control strategies (such as yaw coordination), thus becoming one of the core technical challenges that need to be overcome in the design stage of the wind farm.

[0004] At present, the wind farm is rapidly developing towards large-scale, cluster and complex. As the key equipment for energy conversion of wind energy, the layout density and scale of wind turbines are continuously improved, which makes the influence range of the wake effect on the power generation efficiency further expand and the influence intensity significantly increase. This intensified wake effect will more profoundly change the inflow conditions of the downstream wind turbine, and thus form a greater challenge to the accurate assessment of the power generation capacity of the wind farm and the improvement of the overall power generation efficiency.

[0005] At present, the research on the wind turbine wake effect mainly relies on numerical calculation methods, including empirical models (Park model), actuation-based methods and high-precision simulation of Computational Fluid Dynamics (CFD). The high-precision simulation method of CFD is based on the grid discretization scheme of the geometric analysis of the wind turbine, which can completely capture the aerodynamic characteristics of the blade surface and the evolution process of the wake vortex, and is the best choice for accurately simulating the wind turbine wake.

[0006] The basis of the high-precision CFD method is the grid discretization technology. At present, high-precision numerical simulation of complex terrain in wind fields is usually based on overlapping grid technology to discretize different components. The specific method is as follows: (1) Complex terrain component grid discretization: due to the complex geometry of the terrain, Cartesian grid is automatically generated, and the grid in the near-ground area is encrypted to approach the real terrain profile; based on the assumption of "hub height flow dominance" (> 100m), the atmospheric boundary layer inflow condition is introduced, without generating the body-fitted boundary layer grid, simplifying the terrain discretization process; (2) Fan component grid discretization: in order to accurately capture the blade geometry details and key boundary layer flow, the fan (especially the blade area) usually uses structured grid or hybrid unstructured grid (such as hexahedral boundary layer grid + tetrahedral peripheral grid) to generate body-fitted grid, ensure the accuracy of blade geometry fitting, and accurately simulate the influence of boundary layer flow on blade surface aerodynamic load; (3) Wake area grid discretization: in order to capture the influence of fan wake on downstream flow field and fan, the above terrain Cartesian background grid is globally isotropically encrypted in the estimated fan wake area, trying to improve the resolution of the wake vortex.

[0007] In order to capture the fan wake vortex structure, the prior art has to globally and isotropically encrypt the Cartesian grid in the wake area. Although this "simple and rough" encryption method can improve the resolution to some extent, its inherent globality and isotropy are seriously mismatched with the geometric characteristics of the target physical phenomenon (axially stretched spiral slender vortex filament), which leads to the two core problems of explosive growth of grid quantity and low analysis efficiency. The calculation cost of high-precision and large-scale wind farm wake simulation becomes extremely high, and even it is difficult to realize. SUMMARY

[0008] The purpose of the present application is to solve the problem that the global and isotropic Cartesian grid encryption method used in the prior art to capture the fan wake vortex structure is seriously mismatched with the geometric characteristics of the target physical phenomenon, which further leads to the explosive growth of grid quantity, low analysis efficiency, and ultimately leads to the problem that the calculation cost of high-precision and large-scale wind farm wake simulation is extremely high or even difficult to realize. The present application significantly improves the wake analysis efficiency and controls the calculation cost, and improves the accuracy and efficiency of wind farm wake numerical simulation. Specifically, the present application provides a multi-grid fusion method for complex terrain wind turbine wake numerical simulation.

[0009] The multi-grid fusion method for complex terrain wind turbine wake numerical simulation provided by the present application comprises the following steps: Step 1: curved surface geometry modeling of complex terrain, generating a continuous terrain curved surface geometry model, discretizing the terrain curved surface geometry model into triangular facets, generating a terrain surface grid, generating an outer envelope cube of the terrain surface grid, expanding the outer envelope cube to obtain an expanded calculation domain, determining an initial grid size, generating a uniform Cartesian grid, encrypting the uniform Cartesian grid based on the position relationship between the grid cells of the uniform Cartesian grid and the triangular facets, generating a Cartesian grid of the complex terrain region of the expanded calculation domain; Step 2: constructing a geometric model of a fan blade, discretizing the geometric model of the fan blade into a fan blade grid, generating a surface grid of the fan blade, filling the surface grid to generate a volume grid, and generating an anisotropic structured boundary layer grid on the surface of the fan blade, the surface grid, volume grid, and boundary layer grid of the fan blade constituting a structured or unstructured hybrid grid of the fan region; Step 3: generating an anisotropic O-shaped structure grid of the wake region, constructing a wake volume by geometric abstraction and feature line extraction of the fan wake region, converting the complex fan wake flow field into a geometric model, constructing an O-shaped topology block based on the wake volume, the topology structure of the O-shaped topology block being a closed O-shaped grid in the cross section, and encrypting the O-shaped grid to obtain an O-shaped structure grid of the wake region; Step 4: overlapping and assembling the Cartesian grid of the complex terrain region, the structured or unstructured hybrid grid of the fan region, and the O-shaped structure grid of the wake region to obtain a unified calculation domain.

[0010] Preferably, the step 1: curved surface geometry modeling of complex terrain, generating a continuous terrain curved surface geometry model, discretizing the terrain curved surface geometry model into triangular facets, generating a terrain surface grid, generating an outer envelope cube of the terrain surface grid, expanding the outer envelope cube to obtain an expanded calculation domain, determining an initial grid size, generating a uniform Cartesian grid, encrypting the uniform Cartesian grid based on the position relationship between the grid cells of the uniform Cartesian grid and the triangular facets, generating a Cartesian grid of the complex terrain region of the expanded calculation domain, comprises the following steps: Step 1.1: geometric modeling of complex terrain curved surface, generating a continuous terrain curved surface geometry model; Step 1.2: discretizing the continuous terrain curved surface into triangular facets, determining the facet position by recording the three-dimensional coordinates of the three vertices of each triangular facet, and determining the facet orientation by recording a normal vector, generating a terrain surface grid; Step 1.3: constructing an outer envelope cube based on the terrain surface grid, generating a three-dimensional computational domain, expanding the three-dimensional computational domain based on simulation requirements, extending the range upward to 10-100 times the atmospheric boundary layer top, and extending the range horizontally to 10-20 times the fan spacing or 2-3 times the wind farm scale to obtain an expanded computational domain; Step 1.4: building an initial grid framework covering the expanded computational domain, generating a uniform Cartesian grid based on an initial grid size of 30-50 m; Step 1.5: identifying the positional relationship between the grid cells of the uniform Cartesian grid and the triangular patches in the terrain surface grid using a ray method or a scanning method, and encrypting the cells of the uniform Cartesian grid based on the positional relationship; Step 1.6: if the grid cells of the uniform Cartesian grid intersect with the triangular patches in the terrain surface grid or the distance is less than a first threshold value, then encrypt the grid cells of the uniform Cartesian grid until the size of the grid cells of the uniform Cartesian grid is less than a second threshold value to approximate the terrain geometry.

[0011] Preferably, in the step 1.6: if the grid cells of the uniform Cartesian grid intersect with the triangular patches in the terrain surface grid or the distance is less than a first threshold value, then encrypt the grid cells of the uniform Cartesian grid until the size of the grid cells of the uniform Cartesian grid is less than a second threshold value to approximate the terrain geometry, wherein: the first threshold value is 10 m, and the second threshold value is 10-20 m; the encryption rule for encrypting the grid cells of the uniform Cartesian grid is to divide the grid cells of the uniform Cartesian grid into two along the x, y, and z directions respectively, and if the encrypted grid cells do not satisfy the condition that the grid size is less than the second threshold value, then continue to split the grid cells until the size of the grid cells is less than the second threshold value; wherein x represents the horizontal east-west coordinate of the grid cell, y represents the horizontal north-south coordinate of the grid cell, and z represents the vertical height coordinate of the grid cell.

[0012] Preferably, in the step 2: constructing a geometric model of the fan blade, discretizing the geometric model of the fan blade into a fan blade grid, generating a surface grid of the fan blade, filling the surface grid to generate a volume grid, and generating an anisotropic structured boundary layer grid on the surface of the fan blade, the surface grid, volume grid, and boundary layer grid of the fan blade constitute a structured or unstructured hybrid grid of the fan region, including the following steps: Step 2.1: constructing an initial geometric model of the fan blade, performing geometric cleaning and parameterization characterization on the initial geometric model of the fan blade, and using a non-uniform rational B-spline surface to accurately describe the fan blade to obtain a geometric model of the fan blade; Step 2.2: generating a two-dimensional fan blade surface grid based on the geometric model of the fan blade, the grid unit being quadrilateral or triangular structure, filling the fan blade surface grid to generate a fan blade body grid, and generating a boundary layer grid on the fan blade surface.

[0013] Preferably, the step 2.2: generating a two-dimensional fan blade surface grid based on the geometric model of the fan blade, the grid unit being quadrilateral or triangular structure, filling the fan blade surface grid to generate a fan blade body grid, and generating a boundary layer grid on the fan blade surface, comprises the following steps: Step 2.2.1: generating an anisotropic structured blade surface boundary layer grid on the fan blade surface, the grid line being > 85° with the normal direction of the fan blade surface; Step 2.2.2: generating a three-prism-shaped connection position boundary layer grid at the connection position of each component of the fan blade; Step 2.2.3: adjusting the fineness of the blade surface boundary layer grid and the connection position boundary layer grid in steps 2.2.1 and 2.2.2, adjusting the thickness of the first layer of boundary layer grid according to the preset y⁺ value, setting the number of layers of the blade surface boundary layer grid and the connection position boundary layer grid to be ≥ 30 layers, the thickness gradient ratio to be ≤ 1.2, and the boundary layer grid of the fan blade surface, wherein y⁺ is the ratio of the distance from the center of the first grid to the fan surface to the viscous length of the airflow; Step 2.2.4: generating a tetrahedral grid, smoothly transitioning from the outer edge of the boundary layer grid of the fan blade surface to the far field, the transition range being about 10 times the blade chord length.

[0014] Preferably, the step 3: generating an anisotropic O-shaped structure grid in the wake region, constructing a wake body by geometric abstraction and feature line extraction on the fan wake region, converting the complex fan wake flow field into a geometric model, constructing an O-shaped topology block based on the wake body, the topology structure of the O-shaped topology block being a closed O-shaped grid in the cross section, encrypting the O-shaped grid to obtain an O-shaped structure grid in the wake region, comprising the following steps: Step 3.1: geometrically abstracting the fan wake region, representing the fan wake region as a cylindrical body starting from a certain distance upstream of the impeller and developing downstream, extracting the feature lines of the fan wake region, extracting the wake center line and boundary shear layer of the wake region, and constructing a wake body; Step 3.2: Based on the wake body, an O-type topology block is constructed with the rotation axis of the fan as the central axis, the topology structure of the O-type topology block is a closed O-type grid in the cross section, the diameter of the cylindrical calculation domain of the O-type grid is 10 times the diameter of the wind wheel, and the length is 150 times the diameter of the wind wheel, wherein the distance from the rotation plane where the fan is located to the inlet of the cylindrical calculation domain is 60 times the diameter of the wind wheel, and the distance from the rotation plane where the fan is located to the outlet of the cylindrical calculation domain is 90 times the diameter of the wind wheel; Step 3.3: Radially encrypting the O-type grid within a range of 1.2 times the hub radius centered on the hub and within a range of 0.8 times the impeller radius to 1.2 times the impeller radius, and axially encrypting the O-type grid within a range of 5 times the diameter of the wind wheel upstream to 90 times the diameter of the wind wheel downstream in the rotation plane where the fan is located.

[0015] Preferably, the step 3.3: radially encrypting the O-type grid within a range of 1.2 times the hub radius centered on the hub and within a range of 0.8 times the impeller radius to 1.2 times the impeller radius, and axially encrypting the O-type grid within a range of 5 times the diameter of the wind wheel upstream to 90 times the diameter of the wind wheel downstream in the rotation plane where the fan is located, comprises the following steps: Step 3.3.1: Within a range of 1.2 times the hub radius centered on the hub, the O-type grid is encrypted using a fine grid of 0.2 times the hub diameter, and within a range of 0.8 times the impeller radius to 1.2 times the impeller radius, the O-type grid is encrypted using a superfine grid of 0.1 times the tip chord length; Step 3.3.2: According to the flow direction velocity decay rate and vortex structure dissipation scale, the O-type grid is non-uniformly segmented and encrypted with the rotation plane where the fan is located as 0 point, the region from 5D upstream to 5D downstream of the rotation plane where the fan is located is encrypted using an axial step length of 0.01D, the region from 5D downstream to 20D downstream of the rotation plane where the fan is located is encrypted using an axial step length of 0.02D to 0.05D, and the region from 20D downstream to 90D downstream of the rotation plane where the fan is located is encrypted using an axial step length of 0.1D, wherein D is the diameter of the wind wheel.

[0016] Preferably, the O-type topology block can also be a C-type or H-O hybrid topology block.

[0017] Preferably, step 4: overlapping and assembling the Cartesian grid of the complex terrain region, the structured or unstructured hybrid grid of the fan region, and the O-type structured grid of the wake region to obtain a unified calculation domain, comprising the following steps: Step 4.1: analyzing the grid cell node coordinates, connectivity of the grid cells, and boundary condition markers of the Cartesian grid of the complex terrain region, the structured or unstructured hybrid grid of the fan region, and the O-type structured grid of the wake region, and evaluating the skewness, aspect ratio, and smoothness indicators of the grid cells; Step 4.2: judging the attribute of the grid unit, for the Cartesian grid of the complex terrain region, the structured or unstructured hybrid grid of the fan region, calculating the minimum distance d1 of the grid unit in the overlapping region of the Cartesian grid of the complex terrain region, the structured or unstructured hybrid grid of the fan region to the wall surface of the current component, and the minimum distance d2 of the grid unit in the overlapping region to the wall surface of other components, if d1 is less than or equal to d2, judging that the grid unit in the overlapping region is a valid unit, if d1 is greater than d2, judging that the grid unit in the overlapping region is an invalid unit, judging the valid unit close to the invalid unit as an interpolation unit, for the O-type structured grid of the wake region, setting the wall surface distance of the grid unit of the O-type structured grid of the wake region as 10 times the maximum chord length of the fan blade.

[0018] The application provides a multi-grid fusion method for numerical simulation of a fan wake in a complex terrain. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiments or prior art description will be briefly introduced below.

[0020] Figure 1 is a method flowchart of the application.

[0021] Figure 2 is a continuous curved surface geometry model of the complex terrain of the application.

[0022] Figure 3 is a local encryption diagram of the Cartesian grid of the application.

[0023] Figure 4 is a non-structured hybrid grid diagram of the fan region of the application.

[0024] Figure 5 is a blade surface boundary layer grid diagram of the application.

[0025] Figure 6 is a schematic diagram of the O-type structured grid of the wake region of the application near the fan wheel plane in the axial direction.

[0026] Figure 7 is a schematic diagram of the O-type structured grid of the wake region of the application near the fan blade root and tip in the radial direction. DETAILED DESCRIPTION

[0027] In order to better understand the technical scheme of the present application, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work should fall within the scope of protection of the present application.

[0028] It should be noted that the terms "first", "second", and the like in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to the process, method, product or device.

[0029] Figure 1 A flowchart of a multi-grid fusion method for numerical simulation of wind turbine wake over complex terrain according to the present application. The multi-grid fusion method for numerical simulation of wind turbine wake over complex terrain according to the present application comprises the following steps: Step 1: Generate a Cartesian grid of a complex terrain area.

[0030] Comprising the following steps: Step 1.1: As shown in Figure 2 , the terrain surface geometry is modeled. The complex terrain includes mountains, hills, valleys, etc. The three-dimensional surface geometry is constructed by using commonly used CAD software, through field surveying data (such as unmanned aerial vehicle aerial survey, LiDAR laser scanning data), to restore the details of the terrain such as ups and downs, slopes, gullies, etc. to generate a continuous surface geometry model.

[0031] Step 1.2: Discretization of terrain surface. Discretize the continuous surface into a finite number of triangular facets to generate a terrain surface grid. The triangular facet is the smallest unit of the terrain surface, which is a high-fidelity polygonal approximation of the original continuous surface. Each triangular facet is determined by recording the three-dimensional coordinates of the three vertices to determine the facet position, and recording a normal vector to determine the facet orientation, so as to facilitate the subsequent Cartesian grid to identify the inside and outside of the terrain. This step can be realized by exporting an STL format file through CAD software.

[0032] Step 1.3: Constructing the three-dimensional computational domain, expanding the computational domain according to simulation requirements. Based on the terrain surface grid in step 1.2, a bounding box is established, that is, the entire terrain surface grid is completely enclosed in the smallest cube, and the boundary coordinate values of the cube are the maximum or minimum x, y, z coordinate values of the three-dimensional model, where x represents the horizontal east-west coordinate, y represents the horizontal south-north coordinate, and z represents the vertical height coordinate. The boundary range of the bounding box is the initial framework of the three-dimensional computational domain.

[0033] Subsequently, the computational domain is expanded according to the simulation requirements. Since the wind turbine wake is diffused in the atmosphere, it is not limited to the terrain surface, so it is necessary to expand the range outward based on the three-dimensional model of the complex terrain. The atmospheric boundary layer is the core area of wind turbine operation, which refers to the area close to the earth's surface that is most significantly affected by terrain and surface friction. The atmospheric boundary layer extends from the ground upwards, with a height of about 100-1000m, depending on the climate, weather, and surface type of the simulated area, and its height is not a fixed value. For example, in the plains, the range of the atmospheric boundary layer is about 200-300m on sunny days, and it can reach 500m on cloudy and windy days. In mountainous and complex terrain, the atmospheric boundary layer is thicker, possibly reaching 500-1000m, due to the terrain undulations disrupting the airflow. In tropical regions, the surface heating is strong, and the boundary layer can reach more than 1000m. In cold regions, the atmospheric boundary layer is thinner, about 100-200m.

[0034] The focus of expanding the atmospheric boundary layer is to extend upwards, because the wind turbine wake will spread upwards, and the expansion range usually extends to 10-100 times the top of the atmospheric boundary layer. The top of the atmospheric boundary layer refers to the upper boundary of the atmospheric boundary layer, and in the atmospheric region above the upper boundary, the influence of surface friction is basically eliminated, and the airflow is mainly controlled by atmospheric circulation (such as upper-level jet stream), with more stable wind speed and direction, and almost no terrain interference. If only simulating the short-term wake of a single wind turbine, such as diffusion within 10 minutes, the expansion range can be selected as 10-20 times the top of the atmospheric boundary layer, for example, if the top of the atmospheric boundary layer is 300m, the computational domain can be expanded upwards to 3000-6000m, which is sufficient for the wake to spread completely within the domain; if simulating the long-term wake of a large wind farm, such as the wake of dozens of wind turbines for more than 1 hour, an expansion range of 50-100 times is needed, for example, if the top of the atmospheric boundary layer is 500m, the computational domain needs to be expanded upwards to 25000-50000m, to ensure that the wakes of multiple wind turbines can still move completely within the computational domain after superposition, without being truncated.

[0035] At the same time, the horizontal direction of the atmospheric boundary layer is appropriately expanded according to the wind turbine spacing and the scale of the wind farm, usually by 10-20 times the wind turbine spacing or 2-3 times the scale of the wind farm, to avoid the wake being truncated by hitting the boundary of the computational domain.

[0036] Step 1.4: Determine the initial grid size Δx, Δy, Δz, and generate a uniform Cartesian grid.

[0037] An initial grid framework covering the entire computational domain is built. The initial grid size Δx, Δy, Δz represents the edge length of the grid in the x, y, z direction, respectively, and the commonly used value in engineering is 30-50 m. If the initial grid size is too large, such as 100 m, small terrain undulations will be ignored; if it is too small, such as 10 m, the total number of grids will increase dramatically, causing the calculation to be sluggish.

[0038] Step 1.5: Identify the positional relationship between the Cartesian grid cells and the triangular facets in the terrain surface grid using the ray method or the scanning method.

[0039] This step is used to determine whether each Cartesian grid contacts the terrain surface triangular grid to clearly identify the areas that need to be refined. The principle of the ray method is to emit a virtual ray from the center point of a certain grid cell in any direction, and count the number of intersections between the ray and the triangular facets. An odd number of intersections indicates that the grid is inside the terrain (e.g., underground), an even number indicates that it is outside (e.g., in the atmosphere), and exactly one intersection indicates that it intersects (half of the grid is inside the terrain and half is outside). The principle of the scanning method is to scan the computational domain row by row in the x or y direction, record the overlap between each grid and the triangular facets, and distinguish the positional relationship between the Cartesian grid and the triangular facets, including internal, external, and intersection relationships.

[0040] Step 1.6: As shown in Figure 3 , if the Cartesian grid cell intersects or is closer than the first threshold value to the triangular facets of the terrain surface grid, the grid cell is refined until the grid size is smaller than the second threshold value to approximate the terrain geometry.

[0041] This step refines the key grids near the terrain to make the grid conform to the terrain undulations and improve the simulation accuracy.

[0042] There are two trigger conditions for refining the grid cells. The first trigger condition is that the Cartesian grid cell intersects the triangular facets of the terrain surface grid (Cut-cell), i.e., the grid is cut by the terrain facets and the shape is no longer a complete cube. If this type of grid is not refined, it will cause the terrain surface to appear jagged, such as a steep slope being cut into steps by a coarse grid. The second trigger condition is that the distance between the Cartesian grid cell and the triangular facets of the terrain surface grid is less than the first threshold value, which is usually 10 m. At this time, even if the Cartesian grid cell does not directly intersect the triangular facets of the terrain surface grid, the terrain still has a significant impact on the airflow within 10 m, such as the airflow around a hillside, which requires fine grids to capture such changes.

[0043] The encryption rule is to divide the qualified grid into eight sub-grids by dividing the grid along the x, y and z directions respectively by half, and the edge length of the sub-grid is half of the edge length of the grid, i.e. 25 m. If the grid size is still not less than the second threshold value after encryption, the grid is continuously split until the grid size is less than the second threshold value. The second threshold value is set according to the terrain complexity. The more complex the terrain is, the smaller the threshold value is, and the threshold value is usually 10-20 m.

[0044] Step 2: Generate a structured or unstructured hybrid grid for the fan area, as shown in Figure 4 .

[0045] This step is used to generate the grid of the fan core area, and specifically includes the following steps: Step 2.1: Fan blade geometry modeling. The complex structure of the physical fan is converted into a computer-recognizable, high-precision geometric model.

[0046] Since the original CAD model of the fan may contain redundant information (such as bolts, small holes and other non-key details), the present application is based on CAD software to perform geometry cleaning and parameterization characterization on the key component of the fan system, i.e. the fan blade, to generate a high-quality STL model based on the fan blade. Geometry cleaning refers to deleting features that have no effect on airflow simulation, avoiding unnecessary increase in the number of grids, and parameterization characterization refers to defining key dimensions with mathematical formulas, such as the chord length and twist angle of the blade varying with the spanwise variation, which facilitates subsequent modification, such as adjusting the shape of the blade, which only needs to update the model by changing the parameters.

[0047] The present application uses non-uniform rational B-spline surfaces to accurately describe the geometric profile of complex aerodynamic shapes such as blades, providing accurate geometric definitions for subsequent high-quality grid division. Non-uniform rational B-spline surfaces can accurately describe complex curves and surfaces through a small number of control points, which are smoother and have smaller errors than triangular facets.

[0048] Step 2.2: Fan blade grid discretization. The geometric model is disassembled into a finite number of grids, allowing the computer to calculate airflow movement through the grids. This step is divided into three stages: surface grid generation, volume grid generation and boundary layer grid generation.

[0049] First, surface grid generation is performed for the fan blade, generating a layer of two-dimensional grids on the surface of the blade, with the grid elements usually being quadrilateral or triangular structures.

[0050] Subsequently, volume grid generation is performed based on the surface grid. Three-dimensional grids are filled in the interior of the surface grid, and this step is the core carrier for airflow calculation. The computer can obtain flow velocity, pressure and other parameters by solving the momentum equation and energy equation in each volume grid element.

[0051] Finally, boundary layer grid generation is performed, as shown in Figure 5As shown, anisotropic structured boundary layer mesh is generated on the surface of the fan blade, ensuring that the grid orthogonality is > 85°, that is, the angle between the grid line and the normal direction of the blade surface is ≥ 85° (close to vertical). If the orthogonality is poor (such as 60°), the grid will be obliquely cut, which will amplify the error during calculation, resulting in deviation of the initial intensity of the wake. At the same time, three-prism mesh is locally used in the geometric mutation area of the part connection position, such as the connection between the blade and the hub and the transition area of different airfoil sections of the blade, to avoid distortion. At the same time, the fineness of the boundary layer mesh is controlled, the first layer thickness is dynamically adjusted according to the preset y⁺ value (about 1), the number of boundary layer mesh layers is set to ≥ 30 layers (thickness gradient ratio ≤ 1.2), and the boundary layer mesh is generated. From the outside of the boundary layer to the far field, the transition range is about 10 times the blade chord length, and tetrahedral mesh is generated for smooth transition.

[0052] Wherein, the physical meaning of y⁺ is: the ratio of the distance from the center of the first layer mesh on the surface of the fan blade to the surface of the fan to the viscous length of the airflow, which is the core index for judging whether the mesh on the surface of the fan is fine enough. For fan simulation, y⁺≈1 means that the first layer mesh can capture the area where the flow velocity near the surface of the fan starts to increase sharply from 0, and this area is the key source of friction resistance of the airflow and the initial turbulence of the wake.

[0053] The number of boundary layer mesh layers is ≥ 30 layers, which means that the boundary layer is a transition area from the surface of the fan (flow velocity is 0) to the outside free flow (flow velocity is close to the incoming flow velocity), and a sufficient number of layers are needed to capture the continuous change of flow velocity. If the number of layers is too small, such as 10 layers, the flow velocity will jump, the flow velocity difference between adjacent two layers of mesh will be too large, and the calculation of the turbulence intensity of the wake will be too small.

[0054] The thickness gradient ratio ≤ 1.2 means that the thickness ratio of adjacent two layers of mesh ≤ 1.2, such as the thickness of the first layer is 0.1 mm, the thickness of the second layer ≤ 0.12 mm, the thickness of the third layer ≤ 0.144 mm, …… If the gradient ratio is too large, such as 1.5, the mesh will suddenly become thick, and the fine flow velocity change in the boundary layer cannot be distinguished.

[0055] The blade chord length is the distance from the leading edge to the trailing edge of the largest section of the blade, usually 2-5 m, and 10 times the chord length is 20-50 m. This is the interpolation boundary of the flow field near the fan blade and the background mesh. Outside the 10 times chord length, the interpolation boundary is far away from the surface of the blade, and the intensity of the flow field disturbed by the surface of the blade begins to decrease, which can improve the accuracy of the flow field interpolation.

[0056] Step 3: Generate anisotropic O-type structure mesh in the wake area.

[0057] Since the fan wake is a low-speed disturbance zone formed after the airflow passes through the blades, it contains complex vortex structures (such as tip vortex and hub vortex) and velocity gradients. These characteristics gradually evolve from the fan outlet to the downstream, resulting in clear near-field vortex structures and gradual dissipation in the far field. Ordinary grids are difficult to balance the fine structure of the wake core and the computational efficiency of large-scale diffusion, while O-type structured grids can form concentric circular grids centered on the fan, tightly wrapping the wake core and smoothly transitioning to the far field, perfectly matching the characteristics of strong disturbance in the wake center and gradual decay outward.

[0058] Step 3.1: Abstract geometry and feature line extraction in the wake area, and construct the wake body. Convert the complex wake flow field into a quantifiable geometric framework.

[0059] Based on the simulation results of the flow field or empirical models, the wake area behind the blades is abstracted as a cylindrical body with a certain distance upstream of the impeller, preferably one impeller diameter, as the starting point and developing downstream. Extract the core feature lines of this area, including the wake centerline and boundary shear layer, to define the main axis direction and lateral expansion law, providing a geometric basis for the topological framework of O-type grid construction.

[0060] Among them, the wake centerline refers to the axis of the cylindrical body, which determines the direction of the grid and ensures that the grid extends along the development direction of the wake; the boundary shear layer refers to the interface between the blade wake and the surrounding airflow, where the velocity gradient is the largest, and the radial boundary of the grid can be defined to ensure that there are enough dense grids in the shear layer area.

[0061] Step 3.2: Construct O-type topology block. The O-type topology block is the basic grid structure of the wake area, and the specific construction method is: Around the abstracted wake body, the O-type topology block is constructed with the fan shaft as the center axis, which can also be a C-type or H-O hybrid topology block. The topology structure of this topology block is a closed O-type grid in the cross section, which can surround the wake core with high quality. The diameter of the cylindrical calculation domain of the O-type grid is 10 times the diameter of the wind wheel (D), and the length is 150D; among them, the rotating plane of the fan is set at a specific position in the cylindrical calculation domain of the cylindrical O-type grid with the diameter of the wind wheel (D) as the reference, the distance between the fan position and the upstream is 60D: the distance between the fan rotating plane and the inlet of the calculation domain (the direction of the incoming flow) is 60 times the diameter of the wind wheel, the distance between the fan position and the downstream is 90D: the distance between the fan rotating plane and the outlet of the calculation domain (the direction of the wake dissipation) is 90 times the diameter of the wind wheel.

[0062] Step 3.3: Anisotropic grid refinement. To accurately capture high shear and strong vortex regions, radial refinement near the hub and tip is required, and axial refinement before and after the rotating plane is required. Among them, the radial direction is perpendicular to the fan shaft (from the center to the outside), and the axial direction is along the fan shaft (from upstream to downstream).

[0063] As Figure 7 In radial encryption, the grid is encrypted in the hub center and the tip region: a fine grid of 0.2 times the hub diameter is used within a range of 1.2 times the hub radius from the hub center, and a super-fine grid of 0.1 times the tip chord length is used within a range of 0.8 times the impeller radius to 1.2 times the impeller radius.

[0064] As Figure 6 In axial encryption, according to the physical field gradient changes such as the flow velocity decay rate and vortex structure dissipation scale, the axial direction is non-uniformly segmented, the near-field region with large gradient is encrypted, and the far-field region with small gradient is gradually sparse, so as to realize anisotropic grid distribution, which is the key to ensure the resolution of vortex core. In the near-field region, the rotating plane is taken as the 0 point, and the high-resolution grid is used in the region 5D upstream to 5D downstream (-5D to +5D) of the fan, such as the axial step length of 0.01D, to capture vortex generation; in the transition region, the grid is gradually relaxed from 5D to 20D downstream (+5D to +20D), such as the axial step length increasing from 0.02D to 0.05D; in the far-field region, the large grid is used to handle the turbulent diffusion from 20D to 90D downstream (+20D to +90D), such as the axial step length of 0.1D.

[0065] Step 4: Overlapping assembly of terrain, fan, and wake O-shaped cylindrical grid, and integration to obtain a unified grid.

[0066] Step 4.1: Read in the grid topology information and perform grid quality detection.

[0067] Read in the grid files (such as CGNS, FLUENT MSH) of each component, including the Cartesian grid of the terrain region in step 1, the fan blade grid in step 2, and the O-shaped structure grid in the wake region in step 3, analyze the node coordinates, element connectivity, and boundary condition markers, where the boundary condition markers indicate which elements belong to the wall surface (blade surface, ground, etc.) or inlet, outlet, etc. At the same time, carry out grid assembly and quality check, evaluate grid skewness, aspect ratio, and smoothness, etc. to ensure that it meets the requirements of the CFD solver.

[0068] Step 4.2: Determine the properties of the grid elements. The wall distance is used as the criterion to determine the properties of the grid elements, where the effective elements participate in the flow field calculation, the ineffective elements do not participate in the flow field calculation, and the interpolation elements are used for flow field variable interpolation between different component grids.

[0069] First, the wall distance is taken as the judgment standard to judge the attribute of the grid unit, the minimum distance d1 of the grid unit of the Cartesian grid of the complex terrain area in step 1 and the grid of the fan blade in step 2 to the wall of the part is calculated, then the minimum distance d2 of the grid unit in the overlapping area of the Cartesian grid of the complex terrain area in step 1 and the grid of the fan blade in step 2 to the wall of other parts is calculated, the size of d1 and d2 is compared, if d1 is less than or equal to d2, it is an effective unit, if d1 is greater than d2, it is an invalid unit, and the effective unit close to the invalid unit is an interpolation unit. Since the O-shaped structure grid in the wake region has no wall, the wall distance of the O-shaped structure grid unit in the wake region can be directly given as 10 times the maximum chord length of the fan blade.

[0070] Step 4 ensures that the grid itself is reliable through grid quality detection, and solves the overlapping conflict through wall distance judgment, and finally integrates the three independent grids in steps 1 to 3 into a logically unified calculation domain. This process not only takes advantage of the high precision of each grid in a specific area, but also avoids the contradiction of overlapping calculation, providing a high-quality and conflict-free calculation carrier for subsequent wake numerical simulation.

[0071] Based on the above analysis, the multi-grid fusion method for complex terrain wind turbine wake numerical simulation provided by the embodiment of the application achieves the design intention.

[0072] The multi-grid fusion method for complex terrain wind turbine wake numerical simulation provided by the application achieves the following technical effects: (1) Significantly reduce the calculation cost and improve the simulation efficiency Through the regional customized grid strategy, the redundant calculation caused by the traditional isotropic encryption in the whole wake region is avoided. In the wake region, the anisotropic structure grid is adopted, which accurately matches the geometric characteristics of the tip vortex, greatly reduces the unnecessary grid units while ensuring the resolution, and makes the calculation resources concentrated in the key physical region, thereby significantly reducing the calculation amount and shortening the simulation time.

[0073] (2) Accurately analyze the flow characteristics of the wake and improve the simulation accuracy of the wake The hybrid grid (boundary layer hexahedron / three-prism + peripheral tetrahedron) is used in the fan area to accurately capture the blade aerodynamic load; the anisotropic grid in the wake region explicitly matches the axial stretching characteristics of the vortex filament, effectively maintaining the coherence and strength of the vortex core; the complex terrain is quickly generated and locally encrypted by the Cartesian grid. This multi-grid fusion strategy can achieve higher precision in numerical analysis of key issues such as fan performance, vortex evolution and terrain wind conditions, which can significantly improve the simulation accuracy of the wake.

[0074] (3) Improve the automation degree and engineering applicability of complex wind field modeling Based on the overlapping grid technology, the grids of different topologies are integrated, which not only retains the advantage of fast generation of the Cartesian grid in the terrain modeling, but also improves the adaptability through the local optimization of the structured / unstructured grid. The method supports modular grid generation, and can dynamically adjust the grid configuration of each region according to the simulation requirements, greatly improving the automation degree and engineering applicability of complex wind farm modeling.

[0075] It should be noted that the technical features in the above embodiments can be freely combined for those skilled in the art, and the formed technical solutions also belong to the disclosed embodiments of the present application.

[0076] Further, without departing from the principles of the present application, the present application can be improved and modified in several ways, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. A multi-grid fusion method for numerical simulation of wind turbine wake over complex terrain, characterized in that, The method comprises the following steps: Step 1: curved surface geometry modeling of complex terrain is performed to generate a continuous terrain curved surface geometry model, the terrain curved surface geometry model is discretized into triangular facets, a terrain surface grid is generated, an outer envelope cube of the terrain surface grid is generated, the outer envelope cube is expanded to obtain an expanded calculation domain, an initial grid size is determined, a uniform Cartesian grid is generated, the grid cells of the uniform Cartesian grid and the positions of the triangular facets are related to each other, the uniform Cartesian grid is encrypted to generate a Cartesian grid of the complex terrain region of the expanded calculation domain; Step 2: a geometry model of a fan blade is constructed, the geometry model of the fan blade is discretized into a fan blade grid, a surface grid of the fan blade is generated, the surface grid is filled to generate a volume grid, and an anisotropic structured boundary layer grid is generated on the surface of the fan blade, and the surface grid, the volume grid and the boundary layer grid of the fan blade constitute a structured or unstructured hybrid grid of a fan region; Step 3: an anisotropic O-shaped structure grid of a wake region is generated, a wake volume is constructed by performing geometric abstraction and feature line extraction on the fan wake region, the complex fan wake flow field is converted into a geometry model, an O-shaped topological block is constructed based on the wake volume, the topological structure of the O-shaped topological block is a closed O-shaped grid in a cross section, and the O-shaped grid is encrypted to obtain an O-shaped structure grid of the wake region; Step 4: the Cartesian grid of the complex terrain region, the structured or unstructured hybrid grid of the fan region and the O-shaped structure grid of the wake region are overlapped and assembled, and a unified calculation domain is fused.

2. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 1, wherein, The step 1: curved surface geometry modeling of complex terrain is performed to generate a continuous terrain curved surface geometry model, the terrain curved surface geometry model is discretized into triangular facets, a terrain surface grid is generated, an outer envelope cube of the terrain surface grid is generated, the outer envelope cube is expanded to obtain an expanded calculation domain, an initial grid size is determined, a uniform Cartesian grid is generated, the grid cells of the uniform Cartesian grid and the positions of the triangular facets are related to each other, the uniform Cartesian grid is encrypted to generate a Cartesian grid of the complex terrain region of the expanded calculation domain, and the step 1 comprises the following steps: Step 1.1: geometric modeling of a complex terrain curved surface is performed to generate a continuous terrain curved surface geometry model; Step 1.2: the continuous terrain curved surface is discretized into triangular facets, the positions of the triangular facets are determined by recording the three-dimensional coordinates of three vertices of each triangular facet, and the orientations of the triangular facets are determined by recording a normal vector, and a terrain surface grid is generated; Step 1.3: an outer envelope cube is constructed based on the terrain surface grid to generate a three-dimensional calculation domain, the three-dimensional calculation domain is expanded based on simulation requirements, the upward expansion range is extended to 10-100 times of the atmospheric boundary layer top, and the horizontal expansion range is extended to 10-20 times of the fan spacing or 2-3 times of the wind field scale to obtain an expanded calculation domain; Step 1.4: an initial grid framework covering the expanded calculation domain is built, and a uniform Cartesian grid is generated based on an initial grid size of 30-50 m; Step 1.5: identifying the position relationship between the grid cells of the uniform Cartesian grid and the triangular facets in the terrain surface grid by ray method or scanning method, and encrypting the cells of the uniform Cartesian grid based on the position relationship; Step 1.6: if the grid cell of the uniform Cartesian grid intersects with the triangular facet in the terrain surface grid or the distance is less than a first threshold value, then encrypting the grid cell of the uniform Cartesian grid until the size of the grid cell of the uniform Cartesian grid is less than a second threshold value to approximate the terrain geometry.

3. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 2, wherein, The step 1.6: if the grid cell of the uniform Cartesian grid intersects with the triangular facet in the terrain surface grid or the distance is less than a first threshold value, then encrypting the grid cell of the uniform Cartesian grid until the size of the grid cell of the uniform Cartesian grid is less than a second threshold value to approximate the terrain geometry, wherein: The first threshold value is 10 m, and the second threshold value is 10-20 m; The encryption rule for encrypting the grid cell of the uniform Cartesian grid is to divide the grid cell of the uniform Cartesian grid into two parts along the x, y and z directions respectively, and if the encrypted grid cell does not satisfy the condition that the grid size is less than the second threshold value, then continue to split the grid cell until the size of the grid cell is less than the second threshold value. Wherein, x represents the horizontal east-west coordinate of the grid cell, y represents the horizontal south-north coordinate of the grid cell, and z represents the vertical height coordinate of the grid cell.

4. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 1, wherein, The step 2: constructing a geometric model of the fan blade, discretizing the geometric model of the fan blade into a fan blade grid, generating a surface grid of the fan blade, filling the surface grid to generate a volume grid, and generating an anisotropic structured boundary layer grid on the surface of the fan blade, wherein the surface grid, the volume grid and the boundary layer grid of the fan blade constitute a structured or unstructured hybrid grid of the fan region, and the step 2 includes the following steps: Step 2.1: constructing an initial geometric model of the fan blade, performing geometric cleaning and parameterized characterization on the initial geometric model of the fan blade, and using a non-uniform rational B-spline surface to accurately describe the fan blade to obtain a geometric model of the fan blade; Step 2.2: based on the geometric model of the fan blade, generating a two-dimensional fan blade surface grid, the grid cell being a quadrilateral or triangular structure, filling the fan blade surface grid to generate a fan blade volume grid, and generating a boundary layer grid on the surface of the fan blade.

5. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 4, wherein, The step 2.2: based on the geometric model of the fan blade, generating a two-dimensional fan blade surface grid, the grid cell being a quadrilateral or triangular structure, filling the fan blade surface grid to generate a fan blade volume grid, and generating a boundary layer grid on the surface of the fan blade, and the step 2.2 includes the following steps: Step 2.2.1: generating an anisotropic structured blade surface boundary layer grid on the surface of the fan blade, the grid line having an angle ≥ 85° with the normal direction of the surface of the fan blade; Step 2.2.2: generating a connection position boundary layer grid in the shape of a triangular prism at the connection position of each component of the fan blade; Step 2.2.3: Adjust the fineness of the blade surface boundary layer grid and the connection position boundary layer grid in steps 2.2.1 and 2.2.2, adjust the thickness of the first layer of boundary layer grid according to the preset y + value, set the number of layers of the blade surface boundary layer grid and the connection position boundary layer grid to be greater than or equal to 30 layers, and the thickness gradient ratio to be less than or equal to 1.2, and construct the boundary layer grid of the fan blade surface, wherein y + is the ratio of the distance from the center of the first grid to the fan surface to the viscous length of the airflow; Step 2.2.4: Generate a tetrahedral grid that smoothly transitions from the outer edge of the boundary layer grid of the fan blade surface to the far field, with a transition range of about 10 times the blade chord length.

6. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 1, wherein, The step 3: generate anisotropic O-shaped structure grid in the wake region, construct a wake body by geometric abstraction and feature line extraction on the fan wake region, convert the complex fan wake flow field into a geometric model, construct an O-shaped topology block based on the wake body, the topology structure of the O-shaped topology block is a closed O-shaped grid in the cross section, and the O-shaped structure grid in the wake region is obtained by encrypting the O-shaped grid, including the following steps: Step 3.1: geometrically abstract the fan wake region, and represent the fan wake region as a cylinder starting from a certain distance upstream of the impeller and developing downstream, extract the feature lines of the fan wake region, extract the wake center line and boundary shear layer of the wake region, and construct a wake body; Step 3.2: based on the wake body, an O-shaped topology block is constructed with the shaft of the fan as the center axis, the topology structure of the O-shaped topology block is a closed O-shaped grid in the cross section, the diameter of the cylindrical calculation domain of the O-shaped grid is 10 times the diameter of the wind wheel, and the length is 150 times the diameter of the wind wheel, wherein the distance from the rotation plane where the fan is located to the inlet of the cylindrical calculation domain is 60 times the diameter of the wind wheel, and the distance from the rotation plane where the fan is located to the outlet of the cylindrical calculation domain is 90 times the diameter of the wind wheel; Step 3.3: radially encrypt the O-shaped grid in the range of 1.2 times the hub radius centered on the hub and in the range of 0.8 times the impeller radius to 1.2 times the impeller radius, and axially encrypt the O-shaped grid in the range of 5 times the diameter of the wind wheel upstream to 90 times the diameter of the wind wheel downstream in the rotation plane where the fan is located.

7. The multi-grid chimera method for numerical simulation of wind turbine wake over complex terrain according to claim 6, wherein, The step 3.3: radially encrypt the O-shaped grid in the range of 1.2 times the hub radius centered on the hub and in the range of 0.8 times the impeller radius to 1.2 times the impeller radius, and axially encrypt the O-shaped grid in the range of 5 times the diameter of the wind wheel upstream to 90 times the diameter of the wind wheel downstream in the rotation plane where the fan is located, including the following steps: Step 3.3.1: in the range of 1.2 times the hub radius centered on the hub, the O-shaped grid is encrypted using a fine grid of 0.2 times the hub diameter, and in the range of 0.8 times the impeller radius to 1.2 times the impeller radius, the O-shaped grid is encrypted using a superfine grid of 0.1 times the tip chord length; Step 3.3.2: According to the flow velocity decay rate and vortex structure dissipation scale, the O-type grid is non-uniformly segmented and encrypted in the rotation plane where the fan is located as the 0 point. The grid in the rotation plane where the fan is located is encrypted with an axial step of 0.01D from 5D upstream to 5D downstream, and the grid in the rotation plane where the fan is located is encrypted with an axial step of 0.02D to 0.05D from 5D downstream to 20D downstream, and the grid in the rotation plane where the fan is located is encrypted with an axial step of 0.1D from 20D downstream to 90D downstream, wherein D is the diameter of the wind wheel.

8. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 1 or 6, wherein, The O-type topology block can also be a C-type or H-O hybrid topology block.

9. The multi-grid chimera method for complex terrain wind turbine wake numerical simulation of claim 1, wherein, Step 4: Overlapping assembly of the Cartesian grid of the complex terrain area, the structured or unstructured hybrid grid of the fan area, and the O-type structured grid of the wake area to obtain a unified calculation domain, including the following steps: Step 4.1: Analyzing the grid cell node coordinates, connectivity of the grid cells, and boundary condition markers of the Cartesian grid of the complex terrain area, the structured or unstructured hybrid grid of the fan area, and the O-type structured grid of the wake area, and evaluating the skewness, aspect ratio, and smoothness indicators of the grid cells; Step 4.2: Judging the properties of the grid cells, for the Cartesian grid of the complex terrain area and the structured or unstructured hybrid grid of the fan area, calculating the minimum distance d1 of the grid cells in the overlapping region of the Cartesian grid of the complex terrain area and the structured or unstructured hybrid grid of the fan area to the wall surface of the component, and the minimum distance d2 of the grid cells in the overlapping region to the wall surface of other components, if d1 is less than or equal to d2, the grid cells in the overlapping region are judged to be valid cells, if d1 is greater than d2, the grid cells in the overlapping region are judged to be invalid cells, and the valid cells near the invalid cells are judged to be interpolation cells, for the O-type structured grid of the wake area, setting the wall surface distance of the grid cells of the O-type structured grid of the wake area to 10 times the maximum chord length of the fan blade.

Images