Wind power plant wake flow prediction method with topographic change scene

Abstract

The wind power plant wake flow prediction method provided by the invention is high in calculation precision, good in calculation benefit and capable of adapting to complex terrains and having terrain change scenes. The method comprises the following steps: step a, determining an evaluation time period according to actual engineering conditions, and determining site terrain range frame selection, aircraft position layout and unit basic information; b, performing large eddy simulation according to modeling information of the frame selection area and given boundary conditions, and establishing a CFD high-precision background wind field with a terrain effect; c, a background wind field obtained through CFD simulation calculation is combined, a flow field is reconstructed through interpolation, and the wake flow effect of the wind power plant and the total generation power of wind power plant units are predicted through an engineering wake flow model. The method is applied to the technical field of wind power.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power, and in particular, to a method for predicting the wake of a wind farm in a scenario with terrain changes. Background Art

[0002] In practical wind engineering problems, the research on the wake effect of wind turbines generally directly applies existing engineering wake models. Although the engineering wake model has a relatively simple structure and a short calculation time, and is suitable for numerical simulation research of wind farms equipped with a large number of wind turbine units. However, because its theory is based on a simplified one-dimensional or two-dimensional flow field and largely depends on empirical parameters, there will be a serious deviation from the actual situation in some application scenarios.

[0003] In recent years, with the improvement of computer technology, if more accurate flow field simulation results are required, as Figure 1 shown, it can be carried out by the computational fluid dynamics method (CFD). The most commonly used Reynolds-averaged Navier-Stokes equations method (RANS) is based on the idea of Reynolds decomposition, decomposing the physical quantities in the flow field into an average part and a fluctuating part, and performing time averaging on the NS equations. Although this method has low requirements for calculation conditions, when solving complex problems such as wind turbine wakes, there are problems such as underestimating the wake velocity deficit and delaying the wake recovery. Nowadays, some research institutions can directly numerically solve the NS equations without any form of modeling and simplification of the NS equations using extremely fine grids. This method is also called the direct numerical simulation method (DNS). Although this method can fully resolve turbulence, due to the extremely high computational cost of the DNS method, it is difficult to directly apply it to actual engineering.

[0004] Therefore, the large eddy simulation method (LES) between the two gradually emerged. The LES method has higher accuracy than the Reynolds-averaged Navier-Stokes equations method and can be implemented on a conventional computer. It is a turbulent numerical calculation method with great development potential. However, the LES method has huge requirements for the number of grids. Therefore, the research on wakes using the LES method often combines with the actuator disk theory, that is, in numerical simulations, real blades are not simulated, but thin disks with thrust applied on the surface are used to replace the blades. Although the results obtained by using the LES method based on the actuator disk for wind farm wake prediction are relatively close to the actual situation, when predicting the wake of a wind farm with terrain changes, due to the more complex atmospheric boundary conditions, if the simplified blade model still needs to be meshed during the process, the computational cost is still very high. Therefore, in order to reduce costs and improve efficiency, when predicting the wake of a wind farm with terrain changes, a method that takes into account both computational accuracy and computational efficiency needs to be comprehensively considered. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a wake prediction method for a wind farm in a terrain change scenario with high calculation accuracy, good calculation efficiency, and the ability to adapt to complex terrains.

[0006] The technical solution adopted by the present invention is a wake prediction method for a wind farm in a terrain change scenario, which includes the following steps:

[0007] Step a: According to the actual engineering situation, determine the evaluation time period, determine the site terrain range selection, the layout of the wind turbine positions, and the information of the turbine foundations;

[0008] Step b: Perform large-eddy simulation based on the modeling information of the selected area and the given boundary conditions to establish a high-precision CFD background wind field with terrain effects;

[0009] Step c: Interpolate and reconstruct the flow field in combination with the background wind field calculated by CFD simulation, and use the engineering wake model to predict the wake effect of the wind farm and the total power generation of the wind farm turbines.

[0010] In the said step a, the site terrain range selection is carried out according to the following steps:

[0011] Step a1: Determine the layout of the wind turbines in the site to ensure that the selected area includes all the turbines;

[0012] Step a2: Determine the changes in the main and secondary wind directions to ensure that there is enough wake development area within the selected area;

[0013] Step a3: For the three-dimensional terrain geometric model, generate it by fitting the elevation point cloud, and select the spatial resolution of the data set ≤ 30m to ensure meeting the requirements of engineering calculations;

[0014] Step a4: Perform slope treatment on the terrain boundary to make the terrain around at the same elevation;

[0015] Step a5: Select the selected area to ensure that the length and width are close, which simplifies the calculation and makes the grid cells more evenly distributed in space;

[0016] Step a6: Divide the structured grid for the selected computational domain, control the number of grids, and check the grid quality to ensure meeting the engineering calculation accuracy.

[0017] In the said step b, when performing large-eddy simulation, the influence of the wind turbine model is ignored, and only a background wind field with terrain effects is simulated.

[0018] The specific steps of the said step b are as follows:

[0019] Step b1: Export the information of the selected area and the terrain elevation map, construct the computational domain, and based on the given computational domain, construct the sub-grid stress model required for large-eddy simulation;

[0020] Step b2: Set the inflow boundary condition, outflow boundary condition, and sidewall condition of the simulated atmospheric boundary layer, as well as the viscosity coefficient of air. Create a grid with appropriate resolution according to the terrain complexity and expected turbulence scale. After setting the solution mode, perform large-eddy simulation;

[0021] In step b3, when performing large-eddy simulation, first perform a steady-state calculation. When the residual reaches 10 -4 and then, based on this, perform an unsteady-state calculation using large-eddy simulation to simulate a stable wind field with terrain effects and atmospheric boundary layer characteristics;

[0022] Step b4: Configure the parameters of each wind turbine in the sub-grid stress model. Use the wind field data obtained from large-eddy simulation as the input condition, and solve it in combination with the engineering wake model to generate the wake effect diagram of each unit in the wind farm and calculate the power generation of the unit.

[0023] In step b3, when performing large-eddy simulation, use the dynamic Smagorinsky-Lilly model as the sub-grid stress model. For the sidewall and top surface of the simulated atmospheric boundary layer, adopt symmetric boundaries, and for the ground, adopt a no-slip wall boundary. Combine the boundary conditions of wind speed and wind direction, set the inhomogeneous flow probe points, simulate a stable wind field with terrain effects and atmospheric boundary layer characteristics, and obtain the wind field simulation results under the selected wind direction.

[0024] The specific steps of step c are as follows:

[0025] Step c1: Based on the CFD high-precision background wind field with terrain effects established in step b, generate a wind farm terrain file. Cut the wind farm terrain file, and the number of segmentation steps is the time step of large-eddy simulation calculation to generate vtk files of the instantaneous wind field at each step length;

[0026] Step c2: Convert the transient wind field data into steady-state data;

[0027] Step c3: Configure the wind turbine parameters and set the initial solution conditions of the engineering wake model;

[0028] Step c4: Calculate based on the engineering wake model using a solver to obtain the wake velocity field and the total power of the unit under inhomogeneous inflow conditions.

[0029] The specific steps of step c3 are as follows: Configure the wind turbine position information, the solver type is the wind turbine grid, the number of grid points is 3, set the initial atmospheric conditions of atmospheric density, turbulence intensity, wind speed, and wind direction, the wake combination model is the Sosfs model, the wake deflection model and the velocity deficit model are the Gauss model, and the wake turbulence model is the Crespo-Hernandez model.

[0030] In step c4, the engineering wake model is selected from the Gaussian rotating hybrid wake model, and it is set that the velocity deficit of the wake follows a Gaussian distribution. The flow velocity u in the flow field after passing through the turbine G has the following analytical expression:

[0031]

[0032] where C is the velocity deficit at the wake center, U ∞ is the free-stream velocity, δ is the wake deflection, and δ y is the wake deflection in the y direction, and δ z is the wake deflection in the z direction, y 0 is the spanwise position of the wind turbine, y is the calculated component of the wind turbine hub in the y direction, z is the calculated component of the wind turbine hub in the z direction, and z h is the hub height of the wind turbine, and σ y , σ z are respectively defined as the wake widths in the y direction and the z direction, and σ y0 , σ z0 refer to the initial values at the beginning of the far wake, and they depend on the ambient turbulence intensity and the thrust coefficient C T , and accurately describe the spatial distribution of the velocity drop in the wake region according to this formula.

[0033] The beneficial effects of the present invention are as follows: In the present invention, first, the information of the boxed area and the topographic elevation map are derived to construct the computational domain. Based on the obtained computational domain, the subgrid-scale stress model required for large-eddy simulation is constructed, and the dynamic Smagorinsky-Lilly model is selected. This model has better adaptability compared with the traditional Smagorinsky model and can automatically adjust the Smagorinsky constant under different flow conditions to better capture the turbulent characteristics near the wall surface, thereby improving the prediction accuracy; the inflow boundary condition, outflow boundary condition, sidewall condition of the simulated atmospheric boundary layer, and physical properties such as the viscosity coefficient of air are set, and a grid with appropriate resolution is created according to the terrain complexity and the expected turbulent scale. After setting the solution mode, large-eddy simulation is carried out. During the process, in order to accelerate the convergence speed, a steady-state calculation can be carried out first. When the residual reaches 10 -4Subsequently, based on this, unsteady calculations are carried out using large-eddy simulation to simulate a steady wind field with terrain effects and atmospheric boundary layer characteristics; configure the position coordinates and other parameters of each wind turbine in the model, use the wind field data obtained from large-eddy simulation as input conditions, and combine with the engineering wake model to solve, generate the wake effect diagrams of each unit in the wind farm, and calculate the power generation of the units. The present invention uses large-eddy simulation to only simulate the background wind field with terrain effects, combines with the engineering wake model to calculate the wind turbine wake, making the wake prediction efficiency for wind farms with terrain changes higher, significantly reducing the calculation cost while ensuring the calculation accuracy; at the same time, the engineering wake model adopts the Gaussian rotation hybrid wake model. Compared with the common Gauss wake model, the Gaussian rotation hybrid wake model (GCH) additionally considers the "wake secondary deflection" effect in the superimposed wake. This improvement makes the calculation results of the wind farm wake closer to the actual situation and improves the simulation accuracy. Description of the Drawings

[0034] Figure 1 is a flowchart of the prior art method for predicting the wake of a wind farm;

[0035] Figure 2 is a flowchart of the method of the present invention;

[0036] Figure 3 is a flowchart of the wake prediction using high-precision fluid simulation combined with the engineering wake model provided by the embodiment of the present invention;

[0037] Figure 4 is a wake schematic diagram of each unit in a certain wind farm provided by the embodiment of the present invention, where x, y, and z are the three components of the three-dimensional space coordinate system, and T1, T2, T3, T4, and T99 are the serial numbers of the wind turbines. Detailed Embodiment

[0038] In the present invention, processes such as modeling and simulation calculations are all implemented on a computer. When predicting the wake effect of a wind farm with terrain changes, the processing of atmospheric boundary conditions and the modeling of wind turbines are divided into two modules. That is, a high-precision background wind field with only terrain effects is established using large-eddy simulation, and then it is used as input conditions to combine with the engineering wake model to achieve wake prediction.

[0039] As Figures 2 to 4 shown, a method for predicting the wake of a wind farm in a terrain-changing scenario includes the following steps:

[0040] Step a: According to the actual engineering situation, determine the evaluation time period, determine the site terrain range selection, the layout of the wind turbine positions, and the unit foundation information. In step a, the selection of the site terrain range is carried out according to the following steps:

[0041] Step a1: Determine the layout of the wind turbines within the site, ensure that the selected area includes all the turbines, use GIS to determine the coordinates of the turbines within the site and extract the longitude and latitude of each turbine;

[0042] Step a2: Based on the average wind direction rose diagram of the local meteorological station at the site, determine the changes in the primary and secondary wind directions, and ensure that there is sufficient wake development area within the selected area;

[0043] Step a3: For the three-dimensional terrain geometric model, generate it by fitting the elevation point cloud, and select the spatial resolution of the data set ≤ 30m to ensure that it meets the requirements of engineering calculations;

[0044] Step a4: For complex terrain, if "artificial cliffs" are generated due to terrain truncation and the boundary conditions cannot be determined, slope treatment is performed on the terrain boundary to make the terrain at the same elevation around;

[0045] Step a5: Select the selected area, ensure that the length and width are close, simplify the calculation and make the grid cells more evenly distributed in space; combine the selected area after slope treatment, considering the calculation complexity, try to make the grid cells more evenly distributed in space. Finally, the plane of the selected area is a square area, and the value of the height of the computational domain should be greater than 6 times the maximum terrain height difference;

[0046] Step a6: Divide structured grids for the selected computational domain, encrypt the grids in the area where the wind turbines are located, control the number of grids, and use the skewness, distortion rate, and overall quality as standards to check the grid quality to ensure that it meets the engineering calculation accuracy.

[0047] Step b: Perform large-eddy simulation based on the modeling information of the selected area and the given boundary conditions to establish a high-precision CFD background wind field with terrain effects; when performing large-eddy simulation, ignore the influence of the wind turbine model and only simulate a background wind field with terrain effects. The specific steps of this step are as follows:

[0048] Step b1: Export the information of the selected area and the terrain elevation map, construct the computational domain, and based on the given computational domain, construct the sub-grid stress model required for large-eddy simulation;

[0049] Step b2: Set the inflow boundary condition, outflow boundary condition, sidewall condition of the simulated atmospheric boundary layer and the viscosity coefficient of the air, create a grid with appropriate resolution according to the terrain complexity and the expected turbulence scale, and perform large-eddy simulation after setting the solution mode;

[0050] Step b3: When performing large-eddy simulation, first perform steady-state calculation. When the residual reaches 10 -4 and then use large-eddy simulation for unsteady-state calculation based on this to simulate a stable wind field with terrain effects and atmospheric boundary layer characteristics;

[0051] Step b4: Configure the parameters of each wind turbine in the subgrid stress model. Use the wind field data obtained from large-eddy simulation as the input condition, and solve it in combination with the engineering wake model to generate the wake effect diagram of each unit in the wind farm and calculate the power generation of the unit.

[0052] In order to divide the physical quantities of the flow into resolvable scale quantities \(x\) and unresolvable scale quantities \(x'\), a box filter is selected, and its filtering function is

[0053]

[0054] where \(\Delta\) is the grid average scale. In three dimensions, \(\Delta=(\Delta 1 \Delta 2 \Delta 3 ) 1 / 3 , \(\Delta 1 , \Delta 2 , \Delta 3 are the grid scales in the \(x 1 , x 2 , x 3 directions respectively. When \(\Delta\rightarrow0\), the large-eddy simulation LES transforms into the direct numerical simulation DNS. In step b3, the subgrid stress is the momentum transport between the filtered small-scale pulsations and the resolvable large-scale pulsations. The dynamic Smagorinsky-Lilly model is adopted as the subgrid stress model. The dynamic Smagorinsky-Lilly model can overcome the shortcomings of the Smagorinsky-Lilly model, such as excessive dissipation, suppression of transition occurrence, and even causing flow laminarization. To meet the engineering requirements, the time step \(\Delta t\) of the large-eddy simulation is \(\Delta 0 / U max , where \(U max is 1.5 times to several times the free-stream velocity, and the requirements can be relaxed further in the dead water area. For the side walls and top surface of the computational simulation atmospheric boundary layer domain, symmetric boundaries are adopted, and the ground adopts a no-slip wall boundary. Combining boundary conditions such as wind speed and wind direction, inhomogeneous flow probe points are set to simulate a stable wind field with terrain effects and atmospheric boundary layer characteristics, and the wind field simulation results under the selected wind direction are obtained.

[0055] Step c: Interpolate and reconstruct the flow field by combining the background wind field calculated by CFD simulation, and predict the wake effect of the wind farm and the total power generation of the wind farm units using the engineering wake model. The specific steps of this step are as follows:

[0056] Step c1: Based on the high-precision CFD background wind field with terrain effects established in step b, generate a wind farm terrain file, cut the wind farm terrain file, and the number of segmentation steps is the number of time steps of the large-eddy simulation calculation to generate vtk files of the instantaneous wind field at each step length;

[0057] Step c2: Convert the transient wind field data into steady-state data;

[0058] Step c3: Configure the wind turbine parameters and set the initial solution conditions for the engineering wake model;

[0059] Step c4: Calculate using a solver based on the engineering wake model to obtain the wake velocity field and the total power of the unit under the condition of heterogeneous inflow.

[0060] The specific steps of step c3 are as follows: Configure the wind turbine point information, the solver type is the wind turbine grid, the number of grid points is 3, set the initial atmospheric conditions of atmospheric density, turbulence intensity, wind speed and wind direction, the wake combination model is the Sosfs model, the wake deflection model and the velocity deficit model are the Gauss models, and the wake turbulence model is the Crespo - hernandez model.

[0061] In step c4, the engineering wake model is selected from the Gaussian Rotating Hybrid Wake Model (GCH), and it is set that the velocity deficit of the wake follows a Gaussian distribution. The flow velocity u in the flow field after passing through the turbine G has the following analytical expression:

[0062]

[0063] where C is the velocity deficit at the wake center, U ∞ is the free stream velocity, δ is the wake deflection, δ y is the wake deflection in the y direction, δ z is the wake deflection in the z direction, y 0 is the spanwise position of the wind turbine, y is the calculated component of the wind turbine hub in the y direction, z is the calculated component of the wind turbine hub in the z direction, z h is the hub height of the wind turbine, σ y and σ z are respectively defined as the wake widths in the y direction and the z direction, σ y0 and σ z0 refer to the initial values at the start of the far wake, and they depend on the ambient turbulence intensity and the thrust coefficient C T , and accurately describe the spatial distribution of the velocity drop in the wake region according to this formula.

[0064] The embodiments of the present invention aim to solve the problem of wake prediction in a wind farm with terrain variations. By using large-eddy simulation technology to obtain a CFD background wind field with only terrain effects, and combining with an engineering wake model to predict the wake effect of the wind farm, it is a method that takes into account both computational accuracy and computational efficiency. The present invention can promote the application of CFD numerical simulation technology in engineering practice, and is an efficient wind farm wake prediction method in actual engineering. The method of the present invention uses a high-precision background wind field to replace the homogeneous wind field, which can greatly improve the accuracy of wind farm wake prediction and reasonably reflect the wake overlap effect of each wind turbine under the terrain. While breaking through the limitations of traditional engineering wake models, it avoids the situation of excessive computational resources occupied by simulating the wake of a wind farm with terrain variations using traditional CFD methods, and takes into account both computational accuracy and computational efficiency.

[0065] Finally, it should be emphasized that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for predicting wind farm wake in a scene with terrain changes, characterized in that: The method comprises the following steps: Step a: According to the actual project situation, determine the evaluation time period, site terrain range selection, machine position layout and basic unit information; Step b, performing large eddy simulation according to the modeling information of the framed area and the given boundary conditions to establish a CFD high-precision background wind field with terrain effect; Step c: combining the background wind field calculated by CFD simulation, reconstructing the flow field by interpolation, and using the engineering wake model to predict the wake effect of the wind farm and the total power generation of the wind farm units.

2. A method for predicting wind farm wakes in a terrain change scenario according to claim 1, characterized in that: In step a, the site terrain range selection is performed according to the following steps: Step a1: Determine the layout of wind turbines on the site and ensure that the selected area includes all units; Step a2: Determine the changes in the primary and secondary wind directions to ensure that there is enough wake development area in the selected area; Step a3: Generate a 3D terrain geometry model using elevation point cloud fitting, and select a data set with a spatial resolution of ≤30m to ensure that it meets the requirements of engineering calculations; Step a4: Slope the terrain boundary to make the terrain at the same elevation; Step a5, select the frame area and ensure that the length and width are close to each other, which simplifies the calculation and makes the grid units more evenly distributed in space; Step a6: Divide the calculated domain into structured grids, control the number of grids, and check the quality of the grids to ensure that the engineering calculation accuracy is met.

3. The method for predicting the wind farm wake in a terrain change scenario according to claim 1, characterized in that: In the step b, when performing large eddy simulation, the influence of the wind wheel model is ignored and only a background wind field with terrain effect is simulated.

4. A method for predicting wind farm wakes in a terrain change scenario according to claim 3, characterized in that: The specific steps of step b are: Step b1, export the information of the selected area and the terrain elevation map, build a calculation domain, and construct a sub-grid stress model required for large eddy simulation based on the given calculation domain; Step b2, setting the inflow boundary conditions, outflow boundary conditions and side wall conditions of the simulated atmospheric boundary layer and the viscosity coefficient of the air, creating a grid with appropriate resolution according to the terrain complexity and the expected turbulence scale, and performing large eddy simulation after setting the solution mode; Step b3: When performing large eddy simulation, first perform steady-state calculation. When the residual reaches 10 -4 Then, based on this, large eddy simulation is used to perform unsteady-state calculations to simulate a stable wind field with terrain effects and atmospheric boundary layer characteristics. Step b4, configure the parameters of each wind turbine in the sub-grid stress model, use the wind field data obtained by large eddy simulation as input conditions, combine it with the engineering wake model to solve, generate the wake effect diagram of each unit in the wind farm, and calculate the power generation of the unit.

5. A method for predicting wind farm wakes in a terrain change scenario according to claim 4, characterized in that: In the step b3, the dynamic Smagorinsky-Lilly model is used as the sub-grid stress model when performing large eddy simulation, the side walls and top surfaces of the atmospheric boundary layer are calculated and simulated, and the ground is adopted as a no-slip wall boundary. In combination with the boundary conditions of wind speed and wind direction, the inhomogeneous flow probe points are set to simulate the stable wind field with terrain effects and atmospheric boundary layer characteristics, and obtain the wind field simulation results under the selected wind direction.

6. The method for predicting the wind farm wake in a terrain change scenario according to claim 1, characterized in that: The specific steps of step c are: Step c1, based on the CFD high-precision background wind field with terrain effect established in step b, generate a wind field terrain file, cut the wind field terrain file, the number of cut steps is the number of time steps of large eddy simulation calculation, and generate a vtk file of the instantaneous wind field of each step; Step c2, converting transient wind field data into steady-state data; Step c3, configuring wind turbine parameters and setting initial solution conditions of the engineering wake model; Step c4: Based on the engineering wake model, the wake velocity field and the total power of the unit are calculated using a solver to obtain the wake velocity field under the condition of non-homogeneous inflow.

7. A method for predicting wind farm wakes in a terrain change scenario according to claim 6, characterized in that: The specific steps of step c3 are: configuring wind turbine point information, the solver type is wind turbine grid, the number of grid points is 3, the initial atmospheric conditions of atmospheric density, turbulence intensity, wind speed and direction are set, the wake combination model is Sosfs model, the wake deflection model and velocity loss model are Gauss model, and the wake turbulence model is Crespo-Hernandez model.

8. The method for predicting the wind farm wake in a terrain change scenario according to claim 6, characterized in that: In step c4, the engineering wake model is selected from the Gaussian rotating mixed wake model, and the velocity loss of the wake is set to follow the Gaussian distribution, and the flow velocity u in the flow field after flowing through the turbine is G The analytical expression is as follows: Where C is the velocity loss at the wake center, U ∞ is the free stream velocity, δ is the wake deflection, is Wake deflection in direction, is Wake deflection in direction, is the spanwise position of the wind turbine, The wind turbine hub is The calculation component in the z direction, z is the calculation component of the wind turbine hub in the z direction, h is the hub height of the wind turbine, They are defined as The wake width in the a and z directions, refers to the initial value at the beginning of the far wake, which depends on the ambient turbulence intensity and the thrust coefficient C T , according to which the formula can accurately describe the spatial distribution of the velocity drop in the wake region.

Images