Improved mesoscale numerical modeling method for large wind farm

Abstract

The application discloses a large-scale wind farm mesoscale numerical simulation method considering wind turbine interference. The method first obtains the basic data of the target wind farm wind turbine; numerical modeling is carried out on each wind turbine by using the basic data; then in the numerical modeling, the wind turbine sub-grid interference model is added, and the atmospheric inflow conditions of each wind turbine in the same grid are corrected, so that the accuracy of the large-scale wind farm mesoscale numerical simulation is improved. Finally, the modeling results are coupled to the numerical weather prediction model to carry out numerical simulation research of the large-scale wind farm. Compared with the traditional wind farm mesoscale numerical modeling method, the modeling method of the application has better accuracy and robustness, and is more suitable for research on the interference in the large-scale wind farm and the wake characteristics of the whole farm.

Description

Technical Field

[0001] This invention belongs to the field of numerical simulation, specifically relating to an improved method for mesoscale numerical modeling of large wind farms. Background Technology

[0002] Wind energy, as a clean and renewable energy source, is developing rapidly. However, blind development without prior assessment and planning can easily lead to economic and resource waste. Therefore, accurate assessment and analysis of wind resources at target sites and the operational characteristics of wind farms themselves have become extremely important. In recent years, mesoscale numerical simulation analysis methods have gradually become a commonly used tool for analyzing wind resources and wind farm wake characteristics due to their advantages such as lower economic and time costs and a wider research scope. For example, application number CN202011237572.3 developed an automated mesoscale wind resource analysis system based on the Linux system using the mesoscale numerical model WRF combined with the preprocessing tool WPS. Application number CN202011237154.4, still based on the WRF model, used the global reanalysis dataset to downscale the 2km mesoscale wind resource information to 200m, obtaining more refined regional wind resources.

[0003] However, none of the above methods consider the impact of the wind farm itself on wind resources and cannot analyze the operational characteristics of the wind farm. This is because the impact of wind turbine operation on the atmosphere is not considered; that is, no modeling analysis is performed specifically for the wind turbine. Currently, most wind farm models are based on data modeling and do not consider the physical characteristics of the wind turbine. Application No.: CN201710280318.3 constructs an equivalent wind farm model for different regions by equally grouping the output power of the wind turbine and using data algorithms, and evaluates its effectiveness. Application No.: CN202011099920.5 uses machine learning methods to model and correlate the relationship between velocity and power in different wind farm regions. An equivalent model of the wind farm is constructed using this method. However, most of the above modeling methods are detached from the actual atmospheric environment, the physical mechanisms are not analyzed, and they are difficult to cope with complex and diverse external environments, resulting in poor model practicality.

[0004] For mesoscale numerical simulations of large-scale wind farms, the commonly used simulation method both domestically and internationally is the coupled wind farm parameterized model, namely the Fitch model. This model treats the wind turbine as a brake disc acting on the incoming atmospheric flow, causing changes in atmospheric momentum and turbulent kinetic energy. However, due to the mesoscale nature of the model, multiple wind turbines exist within the same grid, and the subgrid characteristics of the turbines within the grid are not analyzed, resulting in wind turbines within the same grid having the same atmospheric response. This approximation overestimates the atmospheric effect of the wind turbines, causing simulation errors. Therefore, the model needs further development and improvement. Summary of the Invention

[0005] The purpose of this invention is to solve the problems existing in the current technology and to provide an improved method for mesoscale numerical modeling of large wind farms.

[0006] The objective of this invention is achieved through the following technical solution: An improved mesoscale numerical modeling method for large-scale wind farms comprises the following steps:

[0007] (1) Collect basic information data of each wind turbine in the target wind farm; the basic information data includes geographical location information, wind turbine geometric and physical information, and wind turbine operating characteristics.

[0008] (2) Parametric modeling of the entire wind farm is performed using the basic information data in step (1); each wind turbine is regarded as a drag disk, and the wind turbine applies a momentum sink to the incoming atmospheric flow, converting the kinetic energy of the incoming atmospheric flow into electrical energy and turbulent kinetic energy; the interaction source terms between the wind turbine and the atmosphere are calculated based on the basic information data of the wind turbine in step (1), namely the transient momentum change term, the turbulent kinetic energy change term and the power output.

[0009] (3) For a single computational grid with multiple wind turbines, it is necessary to consider the mutual interference effect between the wind turbines. The accuracy of numerical modeling of large wind farms can be improved by correcting the incoming atmospheric velocity of each wind turbine. The idea of ​​time difference is used to construct a model considering the mutual interference of wind turbines in the sub-grid to correct the incoming velocity error caused by the mutual interference of wind turbines in the same grid. Specifically, within a certain computational time step, wind turbines in the same grid are placed into the grid in sequence. During the time interval between each placement, the placed wind turbines act as brake discs on the incoming atmospheric flow, causing momentum loss in the atmospheric flow field. This momentum loss decays with a predetermined function and diffuses to the next placed wind turbine. This process is repeated, and each wind turbine in the same grid will be affected by its upstream wind turbine, thereby achieving different incoming velocities and finally obtaining the corrected momentum change term, turbulent kinetic energy change term, and power output.

[0010] (4) The modified momentum change term, turbulent kinetic energy change term and power output are coupled to the numerical prediction model WRF, and the wind farm modeling process is embedded into the planetary atmospheric boundary layer (PBL) parameterization scheme of the WRF model, so as to finally realize the mesoscale numerical evaluation and prediction of the flow and operation characteristics of large-scale wind farms considering the mutual interference of wind turbines.

[0011] Furthermore, in step (1), the geographical location information includes the wind farm site and the latitude and longitude information of each wind turbine; the wind turbine's geometric and physical information includes the hub height and the radius of the rotor's rotation surface; and the wind turbine's operating characteristics include the power curve and the thrust curve.

[0012] Furthermore, the WRF model is a weather forecasting model, version WRF version 4.3.

[0013] Furthermore, the atmospheric boundary layer (PBL) parameterization scheme is specifically the MYNN2.5 scheme.

[0014] Furthermore, the decay function is a Gaussian decay function.

[0015] Furthermore, in step (3), the incoming flow velocity experienced by the (n+1)th fan placed in a single computational grid is:

[0016]

[0017] Where u and v are the horizontal velocity components of the incoming flow velocity, |U| is the scalar of the incoming flow velocity, and Δu and Δv are the momentum loss caused by the nth wind turbine already in place, which is calculated as follows:

[0018]

[0019]

[0020] Where, N ij Let A be the number of wind turbines within grid (i,j). ijk Let Δz be the area intercepted by the rotating surface of the wind turbine on the vertical layer k of the grid (i,j). k C is the vertical distance between vertical layer k and vertical layer k+1. T Δt is the thrust coefficient, Δt is the time step of the WRF mode, and Δt / (N ij -1) represents the placement time interval, α and β are the spatial and arrangement correction factors, respectively. The spatial correction factor limits the maximum range of complete diffusion of momentum loss (the smallest scale in the subgrid), while the arrangement correction factor specifies that momentum loss decays in the form of a Gaussian function and gradually diffuses to the surroundings.

[0021] The beneficial effects of this invention are as follows: The improved mesoscale numerical modeling method for large-scale wind farms established in this invention is an innovation and improvement on traditional methods, possessing advantages such as higher accuracy, lower sensitivity to grid resolution, and strong robustness. The improved wind farm numerical model can be applied to large-scale wind farms or wind farm clusters, simulating areas of hundreds or even thousands of square kilometers. The improvement effect is even better when a single grid contains multiple wind turbines. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the steps of the model's operation.

[0023] Figure 2 This is a schematic diagram illustrating the accuracy of the wind turbine grid calculation simulation.

[0024] Figure 3 This is a comparison and verification diagram of the model's mesh sensitivity.

[0025] Figure 4 This is a comparison diagram of the flow field in the model. Detailed Implementation

[0026] The present invention will be further described and illustrated below with reference to the accompanying drawings and specific embodiments.

[0027] This invention primarily constructs an improved mesoscale numerical modeling method for large-scale wind farms to simulate and analyze atmospheric motions around large-scale wind farms and their operational characteristics. The specific steps are as follows:

[0028] 1) Collect basic information data for each wind turbine in the target wind farm. This mainly includes geographical location information (wind farm site and latitude and longitude information of each wind turbine), wind turbine geometric and physical information (hub height and rotor radius), and wind turbine operating characteristics (power curve and thrust curve). Among them, the location information is used to determine the computational domain and the relative position of each wind turbine in the computational domain, while the geometric and physical and operating characteristic information of the wind turbines is used for subsequent parametric modeling calculation steps.

[0029] 2) Utilize the basic data from step 1) to perform parametric modeling of the entire wind farm. WRF, as an open-source mesoscale numerical model, possesses a series of parametric methods for atmospheric physical motions, such as atmospheric boundary layer, atmospheric radiation, and cumulus clouds. Based on the Fortran language, WRF allows researchers to develop new parametric schemes, including previously developed wind farm parametric models. However, due to its mesoscale characteristics, the horizontal size of the WRF grid is typically set to the kilometer level, resulting in a single grid potentially containing multiple wind turbines in traditional models. Traditional models, however, neglect the interactions between wind turbines within the same grid, simplifying the total atmospheric interaction term for wind turbines within the same grid to N for a single turbine. ij Times, where N ij Let be the total number of wind turbines within grid (i,j). Each wind turbine is considered a drag disk, applying a momentum sink to the incoming airflow, converting the kinetic energy into electrical and turbulent kinetic energy. The interaction terms between the wind turbines and the atmosphere are calculated based on the basic information data of the wind turbines obtained in step 1), namely, the transient momentum change term, the turbulent kinetic energy change term, and the power output. The expressions for the total momentum and turbulent kinetic energy interaction terms are:

[0030]

[0031]

[0032]

[0033]

[0034] Where u ijk ,v ijk For the horizontal velocity component in the grid, |U| ijk For velocity scalars, TKE ijk For turbulent kinetic energy, C T C is the thrust coefficient. TKE The turbulent kinetic energy coefficient, its value is C. TKE =C T -C P C P This is the power coefficient. A ijk Let |U| be the area intercepted by the rotating surface of the wind turbine on the vertical layer k of the grid (i,j); P is the power output, |U| hub Let A be the wind speed at the hub height. rotor The area swept by the wind turbine.

[0035] 3) In the parameterization process of step 2), the processing method undoubtedly overestimates the effect of wind turbines within the grid. For a single computational grid with multiple wind turbines, the mutual interference effect between the turbines needs to be considered. To compensate for this error, this invention borrows the ideas of "sub-grid model" and "time difference" to construct a wind turbine interference model within an independent grid. By correcting the incoming flow velocity of each wind turbine, the accuracy of numerical modeling of large-scale wind farms is improved. Combined with... Figure 1 The construction process of the wind turbine interference model will be described in detail. First, within one time step, this invention will divide the N grids (i,j,k) into... ij Wind turbines are placed sequentially. Each turbine begins operation within a designated time interval, acting as a brake disc on the incoming airflow, causing a momentum deficit in the atmospheric flow. This momentum deficit decays with a Gaussian function and diffuses to the next turbine placed. This process repeats, with each turbine within the same grid being influenced by its upstream turbine, resulting in different incoming flow velocities and ultimately yielding corresponding momentum, turbulent kinetic energy, and power output changes. The resulting total grid momentum deficit and turbulent kinetic energy increment are more realistic than those obtained using traditional models. The momentum deficit Δu is calculated separately. ijk and Δv ijk , as well as atmospheric effects—momentum effects and turbulent kinetic energy effects.

[0036] According to the momentum theorem, the changes in the horizontal component of velocity are as follows:

[0037]

[0038]

[0039] Where Δt / (N) ij-1) represents the placement interval, and Δt is the time step of the WRF model. Since the influence range of a placed wind turbine is limited, and the interference effect decreases with increasing distance, two correction coefficients, α and β, are introduced here: a spatial correction factor and an arrangement correction factor, respectively. The spatial correction factor limits the maximum range (smallest scale in the subgrid) for complete diffusion of momentum loss, while the arrangement correction factor specifies that momentum loss decays as a Gaussian function and gradually diffuses outwards. Therefore, the interference effect of a placed wind turbine on a wind turbine about to be placed also depends on the distance between them. The expressions for the correction factors are as follows:

[0040] α=dxdy / δx 2

[0041]

[0042] Where dx and dy are the horizontal grid sizes of the WRF mode, and δx 2 To define the area, d represents the distance between wind turbines. This model assumes that when the distance between turbines exceeds 6δx, the interference effect of the wind turbines is negligible, i.e.: D cr =2.08δx.

[0043] Therefore, the total momentum interference caused by placing a wind turbine on the (n+1)th wind turbine is:

[0044]

[0045]

[0046] Therefore, the horizontal vector of the incoming wind speed corresponding to the (n+1)th wind turbine and and wind speed for:

[0047]

[0048]

[0049]

[0050] Through the obtained and The value is used to calculate the momentum term, turbulent kinetic energy term, and power output of the wind turbine. When all wind turbines are placed, the total momentum term of the wind turbine within the grid (i,j,k) is calculated. Turbulent kinetic energy term ΔTKE ijk and power output term P ij for:

[0051]

[0052]

[0053]

[0054]

[0055] 4) Combining steps 2) and 3) and coupling them to the numerical weather prediction model WRF (weather forecast model, version WRF version 4.3), the wind farm modeling process is embedded into the planetary atmospheric boundary layer (PBL) parameterization scheme of the WRF model (MYNN2.5 scheme) to achieve model closure; then, the meteorological field covering hundreds of kilometers including the wind farm is simulated, thereby finally realizing the mesoscale numerical evaluation and prediction of the flow and operation characteristics of large-scale wind farms considering the mutual interference of wind turbines.

[0056] The specific implementation effect of the above method is demonstrated below with reference to the embodiments.

[0057] Example

[0058] In this embodiment, a large offshore wind farm located in Hangzhou Bay, Zhejiang Province, is taken as the research object to explore the specific implementation effect and advantages of this improved mesoscale numerical modeling method for large wind farms. The parameter settings corresponding to WRF version 4.3 used are as follows:

[0059] Table 1 WRF Parameter Settings

[0060]

[0061] By comparing with measured data, the parameter δx required for the improved mesoscale numerical modeling method for large wind farms was set to 200m. The simulation period was from 00:00 on January 1, 2022 to 24:00 on January 7, 2022.

[0062] By comparing with measured data Figure 2 The standard deviation (SD) and root mean square error (RESM) of the simulated wind speed at the grid with wind turbines are presented. The percentages in the model and grid represent the percentage improvement in accuracy of the improved model (only values ​​greater than 1% are shown). The results show a high degree of agreement between the simulated and measured wind speeds, with an average similarity coefficient... With a value greater than 0.85, the improved parametric model enhances the accuracy of wind speed prediction, especially for downstream grids and grids with a large number of wind turbines. Figure 3 The wake characteristics of the wind field are shown. Due to the consideration of wind turbine interference within a single grid, part of the wake is dissipated in the subgrid, resulting in a shorter wake length in the improved model. Figure 4The model's sensitivity to grid resolution (1km, 2km and 3km) was demonstrated. The results showed that the improved model is less sensitive to grid resolution, and therefore the simulation results of the improved model are more robust and superior to the original model.

[0063] In addition, the improved model in this embodiment consumes comparable computational resources to the traditional model, and can be used in the study of larger-scale wind farms or clusters. As the grid size increases, the number of wind turbines in a single grid increases, and the improved model has a greater advantage in accuracy than the original model.

[0064] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, all technical solutions obtained through equivalent substitution or transformation fall within the protection scope of the present invention.

Claims

1. An improved method for mesoscale numerical modeling of large-scale wind farms, characterized in that, The steps of this method are as follows: (1) Collect basic information data of each wind turbine in the target wind farm; the basic information data includes geographical location information, wind turbine geometric and physical information, and wind turbine operating characteristics; (2) Parametric modeling of the entire wind farm is performed using the basic information data in step (1); each wind turbine is regarded as a drag disk, and the wind turbine applies a momentum sink to the incoming atmospheric flow, converting the kinetic energy of the incoming atmospheric flow into electrical energy and turbulent kinetic energy; the interaction source terms between the wind turbine and the atmosphere are calculated based on the basic information data of the wind turbine in step (1), namely the transient momentum change term, the turbulent kinetic energy change term and the power output. (3) For a single computational grid with multiple wind turbines, it is necessary to consider the mutual interference effect between each wind turbine. The accuracy of numerical modeling of large wind farms can be improved by correcting the incoming atmospheric velocity of each wind turbine. The idea of ​​time difference is used to construct a model that considers the mutual interference of wind turbines in the sub-grid to correct the incoming velocity error caused by the mutual interference of wind turbines in the same grid. Specifically, within a certain computational time step, wind turbines in the same grid are placed into the grid in sequence. During the time interval of each placement, the placed wind turbines act as brake discs on the incoming atmospheric flow, causing momentum loss in the atmospheric flow field. This momentum loss decays with a predetermined function and diffuses to the next placed wind turbine. In this way, each wind turbine in the same grid will be affected by the wind turbine upstream, thus achieving different incoming flow velocities, and finally obtaining the corrected momentum change term, turbulent kinetic energy change term and power output; (4) The modified momentum change term, turbulent kinetic energy change term and power output are coupled to the numerical prediction model WRF, and the wind farm modeling process is embedded into the planetary atmospheric boundary layer PBL parameterization scheme of the WRF model, so as to finally realize the mesoscale numerical evaluation and prediction of the flow and operation characteristics of large-scale wind farms considering the mutual interference of wind turbines.

2. The improved mesoscale numerical modeling method for large-scale wind farms according to claim 1, characterized in that, In step (1), the geographical location information includes the site of the wind farm and the latitude and longitude information of each wind turbine; the geometric and physical information of the wind turbine includes the hub height and the radius of the rotor surface; the operating characteristics of the wind turbine include the power curve and the thrust curve.

3. The improved mesoscale numerical modeling method for large-scale wind farms according to claim 1, characterized in that, The WRF mode mentioned is a weather forecast mode, version 4.

3.

4. The improved mesoscale numerical modeling method for large-scale wind farms according to claim 1, characterized in that, The atmospheric boundary layer PBL parameterization scheme is specifically the MYNN2.5 scheme.

5. An improved numerical modeling method for large-scale wind farms according to claim 1, characterized in that, The given function is a Gaussian decay function.

6. An improved numerical modeling method for large-scale wind farms according to claim 1, characterized in that, In step (3), the incoming flow velocity experienced by the (n+1)th fan placed in a single computational grid is: Where u and v are the horizontal velocity components of the incoming flow velocity. Let the incoming flow velocity be a scalar. , The momentum loss caused by the nth wind turbine already in place is calculated as follows: in, Let be the number of wind turbines within grid (i, j). Let be the area intercepted by the rotating surface of the wind turbine on the vertical layer k of the grid (i, j). C is the vertical distance between vertical layer k and vertical layer k+1. T For thrust coefficient, The time step for WRF mode. The placement time interval, and These are the spatial and arrangement correction factors, respectively. The spatial correction factor limits the maximum extent of complete diffusion of momentum loss, i.e., the smallest scale in the subgrid. The arrangement correction factor specifies that momentum loss decays in the form of a Gaussian function and gradually diffuses to the surrounding area.

Images