System and method for three-dimensional reconstruction of radiator flow channel and fluid performance prediction

Abstract

The present application relates to the field of heat sink design analysis, in particular to a heat sink flow channel three-dimensional reconstruction and fluid performance prediction system and method thereof, comprising: a parameter input module receives a heat sink three-dimensional model and boundary information, and calculates parameters of a flow channel cross section and an inlet cross section; a streamline drawing module receives the parameters, and calculates and determines streamline starting coordinates and boundary points; a streamline smoothing module performs smoothing processing on the streamlines. A flow channel curved surface meshing module receives the smoothed streamline coordinates, generates a three-dimensional curved surface based on space construction points, and performs adaptive non-uniform mesh partitioning and boundary constraint intelligent mesh optimization; a fluid performance prediction module receives the flow channel curved surface mesh, inputs fluid performance analysis parameters, and predicts fluid performance; a calculation result output module receives the prediction results, organizes and summarizes, and then outputs; streamline tracking and smoothing processing, accurate reconstruction of complex flow channels of the heat sink is realized, modeling accuracy is improved by more than 40%, and the accuracy and efficiency of heat sink design are improved.

Description

Technical Field

[0001] This invention relates to the field of radiator design and analysis, and more specifically to a system and method for three-dimensional reconstruction of radiator flow channels and prediction of fluid performance. Background Technology

[0002] As a crucial component of heat exchange equipment, the internal flow channel structure of a radiator directly impacts its heat dissipation efficiency. Traditional radiator design relies on rules of thumb and repeated trials, resulting in long design cycles, high costs, and difficulty in achieving optimization. While advancements in computational fluid dynamics (CFD) technology have made radiator performance prediction possible, existing technologies still face the following challenges:

[0003] First, traditional radiator modeling methods typically rely on manual modeling using CAD software, a cumbersome process with limited accuracy, making it difficult to accurately describe complex flow channel structures. Second, existing mesh generation techniques mostly employ uniform meshes or simple gradient strategies, failing to optimize for the unique and complex flow characteristics of radiator channels, resulting in low computational efficiency or insufficient accuracy. Third, boundary treatment methods are inadequate, particularly in critical areas such as the connection between the flow channel and fins, leading to poor mesh quality and impacting computational stability and result reliability. Furthermore, fluid performance prediction and flow channel geometry modeling are often disconnected, lacking an effective integrated system.

[0004] Therefore, there is an urgent need to develop a system that can automatically reconstruct the three-dimensional structure of radiator flow channels and accurately predict their fluid performance in order to improve radiator design efficiency and performance. Summary of the Invention

[0005] The purpose of this invention is to provide a system and method for three-dimensional reconstruction and fluid performance prediction of radiator flow channels, aiming to solve the problems of low modeling accuracy, poor mesh quality, imperfect boundary treatment, and lack of integrated system in the prior art.

[0006] This invention proposes a three-dimensional reconstruction and fluid performance prediction system for radiator flow channels, comprising:

[0007] The parameter input module is used to receive the 3D model and boundary information of the heat sink and calculate the parameters of the flow channel section and the inlet section.

[0008] The streamline drawing module, connected to the parameter input module, is used to receive parameters of the flow channel cross section and the inlet cross section, and to calculate and determine the starting coordinates and boundary points of the streamlines;

[0009] A streamline smoothing module, connected to the streamline drawing module, is used to receive the streamline starting coordinates and boundary points, and to smooth the streamline.

[0010] The flow channel surface meshing module is connected to the streamline smoothing module. It is used to receive the smoothed streamline coordinates, generate a three-dimensional surface based on spatial points, perform adaptive non-uniform mesh subdivision, and implement boundary constraint intelligent mesh optimization.

[0011] A fluid performance prediction module, connected to the flow channel surface meshing module, is used to receive the flow channel surface mesh, input fluid performance analysis parameters, and predict fluid performance; and

[0012] The calculation result output module is connected to the fluid performance prediction module and is used to receive the fluid performance prediction results, organize and summarize them, and then output them.

[0013] Preferably, the parameter input module includes:

[0014] The model input unit is used to receive the 3D model and boundary information of the heat sink.

[0015] A parameter calculation unit, connected to the model input unit, is used to calculate the parameters of the flow channel cross-section and inlet cross-section based on the 3D model and boundary information of the heat sink; and

[0016] The parameter output unit, connected to the parameter calculation unit, is used to output the parameters of the flow channel section and the inlet section to the streamline drawing module.

[0017] Preferably, the streamline drawing module includes:

[0018] The first input unit is used to receive the parameters of the flow channel cross section and the inlet cross section;

[0019] The second input unit is used to receive the boundary information;

[0020] A streamline calculation unit, connected to the first input unit and the second input unit, is used to calculate the streamline starting coordinates and streamline point coordinates based on the parameters of the flow channel cross-section and the inlet cross-section, as well as the boundary information; and

[0021] The parameter output unit, connected to the streamline calculation unit, is used to send the streamline starting coordinates and streamline point coordinates to the streamline smoothing module.

[0022] Preferably, the streamline smoothing module includes:

[0023] A smoothing input unit is used to receive the streamline starting coordinates and boundary points;

[0024] A smoothing processing unit, connected to the smoothing input unit, is used to smooth the streamlines and generate smoothed streamline coordinates; and

[0025] A smoothing output unit, connected to the smoothing processing unit, is used to send the smoothed streamline coordinates to the flow channel surface meshing module.

[0026] Preferably, the flow channel surface meshing module includes:

[0027] A surface generation unit is used to receive the smoothed streamline coordinates and generate a three-dimensional surface of the flow channel based on the spatial streamline points.

[0028] A mesh generation unit, connected to the surface generation unit, is used for adaptive non-uniform mesh generation based on the three-dimensional surface of the flow channel; and

[0029] A boundary mesh optimization unit, connected to the mesh subdivision unit, is used to perform intelligent boundary constraint optimization on the mesh to generate an optimized flow channel surface mesh.

[0030] Preferably, the surface generation unit includes:

[0031] The streamline preprocessing subunit is used to parameterize the smoothed streamline coordinates and establish topological relationships between streamlines.

[0032] A section construction subunit, connected to the streamline preprocessing subunit, is used to construct a sequence of flow channel sections based on the parameterized streamlines; and

[0033] The surface interpolation subunit, connected to the cross-section construction subunit, is used to generate a continuous three-dimensional surface based on the flow channel cross-section sequence.

[0034] Preferably, the mesh subdivision unit includes:

[0035] The characteristic analysis subunit is used to analyze the geometric features of the three-dimensional curved surface of the flow channel and predict the fluid characteristics;

[0036] A density function construction sub-unit, connected to the characteristic analysis sub-unit, is used to construct a mesh density function based on the geometric features and fluid properties; and

[0037] A mesh generation sub-unit, connected to the density function construction sub-unit, is used to generate an adaptive non-uniform mesh based on the mesh density function.

[0038] Preferably, the boundary mesh optimization unit includes:

[0039] Boundary identification sub-unit, used to identify and classify mesh boundary types;

[0040] Boundary layer mesh sub-units, connected to the boundary identification sub-units, are used to construct boundary layer meshes near the wall boundaries;

[0041] A transition region processing subunit, connected to the boundary identification subunit, is used to process the transition region mesh between boundaries; and

[0042] The quality optimization sub-unit, connected to the boundary layer mesh sub-unit and the transition zone processing sub-unit, is used to perform multi-objective quality optimization on the boundary mesh.

[0043] Preferably, the fluid performance prediction module includes:

[0044] A mesh input unit is used to receive the flow channel surface mesh;

[0045] The parameter setting unit is used to set fluid performance analysis parameters;

[0046] A prediction calculation unit, connected to the mesh input unit and the parameter setting unit, is used to perform fluid performance prediction calculations based on the flow channel surface mesh and the fluid performance analysis parameters; and

[0047] The result generation unit is connected to the prediction calculation unit and is used to generate fluid performance prediction results and send them to the calculation result output module.

[0048] Methods for three-dimensional reconstruction of radiator flow channels and prediction of fluid performance include:

[0049] Receive the 3D model and boundary information of the radiator, and calculate the parameters of the flow channel section and the inlet section;

[0050] Based on the parameters of the flow channel cross section and the inlet cross section, the starting coordinates and boundary points of the streamlines are calculated and determined;

[0051] The streamlines are smoothed to generate smoothed streamline coordinates;

[0052] Receive the smoothed streamline coordinates, generate a three-dimensional surface based on spatial points, perform adaptive non-uniform mesh partitioning, and implement boundary constraint intelligent mesh optimization to generate a flow channel surface mesh.

[0053] Based on the flow channel surface mesh, fluid performance analysis parameters are input to predict fluid performance; and

[0054] Receive fluid performance prediction results, organize and summarize them, and then output them.

[0055] The beneficial effects of this invention include:

[0056] 1. Improve modeling accuracy: Through streamline tracing and smoothing, the complex flow channels of the radiator can be accurately reconstructed, improving modeling accuracy by more than 40%.

[0057] 2. Optimize mesh quality: Employ 3D surface generation technology based on spatial point construction and adaptive non-uniform mesh generation technology to accurately match mesh resources with computational requirements, improving computational efficiency by more than 60%.

[0058] 3. Improved boundary treatment: Introducing boundary constraint intelligent mesh optimization technology, with special treatment for key areas such as fin roots, to improve mesh quality and computational stability, reducing defective cells by 95%.

[0059] 4. Achieve integrated processing: Establish a complete processing flow from inputting the 3D model of the radiator to outputting the fluid performance prediction results, shortening the design cycle by more than 70%.

[0060] 5. Enhanced prediction capability: Generates high-quality flow channel meshes and reasonable boundary conditions, improving the accuracy of fluid performance prediction and reducing the deviation from experimental data by more than 40%. Attached Figure Description

[0061] Figure 1 This is an overall architecture diagram of the radiator flow channel three-dimensional reconstruction and fluid performance prediction system of the present invention;

[0062] Figure 2 This is a schematic diagram of the parameter input module of the present invention;

[0063] Figure 3 This is a schematic diagram of the streamline drawing module of the present invention;

[0064] Figure 4 This is a schematic diagram of the streamline smoothing module of the present invention;

[0065] Figure 5 This is a schematic diagram of the flow channel surface meshing module of the present invention;

[0066] Figure 6 This is a schematic diagram of the surface generation unit of the present invention;

[0067] Figure 7 This is a schematic diagram of the structure of the mesh subdivision unit of the present invention;

[0068] Figure 8 This is a schematic diagram of the boundary mesh optimization unit of the present invention;

[0069] Figure 9 This is a schematic diagram of the fluid performance prediction module of the present invention;

[0070] Figure 10 This is a flowchart of the method for three-dimensional reconstruction of heat sink flow channels and prediction of fluid performance according to the present invention. Detailed Implementation

[0071] Please refer to the attached document. Figure 1-10The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0072] Reference Figure 1 The present invention provides a three-dimensional reconstruction and fluid performance prediction system for radiator flow channels, including a parameter input module 1, a streamline drawing module 2, a streamline smoothing module 3, a flow channel surface meshing module 4, a fluid performance prediction module 5, and a calculation result output module 6.

[0073] Parameter input module 1 is used to receive the 3D model and boundary information of the radiator, and calculate the parameters of the flow channel cross-section and inlet cross-section. In one embodiment of the present invention, the 3D model of the radiator includes radiator flow channel structure data and material data. The boundary information includes the flow channel outlet, flow channel inlet, and flow channel boundary. Preferably, the flow channel cross-section parameters include the major axis length, minor axis length, and angular direction parameters, and the inlet cross-section parameters include shape (e.g., circular, rectangular, elliptical) and size data. Taking an automotive radiator as an example, the inlet cross-section is usually circular or elliptical, with a major axis length of approximately 30–60 mm, a minor axis length of approximately 15–30 mm, and the angular direction depending on the installation position.

[0074] The streamline drawing module 2 is connected to the parameter input module 1 and is used to receive the parameters of the flow channel cross-section and the inlet cross-section, and calculate and determine the starting coordinates and boundary points of the streamlines. Specifically, the streamline drawing module 2 determines the starting point of the streamline by analyzing the shape of the inlet cross-section and boundary information, and calculates the spatial trajectory of the streamline based on the fluid flow law and boundary constraints. In a typical automotive engine radiator, streamline drawing will evenly distribute 30 to 50 starting points on the inlet cross-section to ensure that the flow channel shape can be accurately captured.

[0075] The streamline smoothing module 3 is connected to the streamline drawing module 2 and is used to receive the starting coordinates and boundary points of the streamlines, and to smooth the streamlines. In this invention, the smoothing process uses a spline curve interpolation method to improve the smoothness of the drawn streamlines. Preferably, the spline curve is a cubic spline curve, which can ensure the continuity of the first and second derivatives of the streamlines and avoid sharp angles or abrupt changes in the streamlines. For complex heat sink channels, such as S-shaped curved pipes, smoothing is crucial to ensuring the quality of subsequent surface generation.

[0076] The flow channel surface meshing module 4 is connected to the streamline smoothing module 3. It receives the smoothed streamline coordinates, generates a three-dimensional surface based on spatial points, performs adaptive non-uniform mesh partitioning, and implements boundary constraint intelligent mesh optimization. This module is the core innovation of this invention and will be described in detail later.

[0077] The fluid performance prediction module 5 is connected to the flow channel surface meshing module 4, and is used to receive the flow channel surface mesh, input fluid performance analysis parameters, and predict fluid performance. In embodiments of the present invention, the fluid performance analysis parameters include fluid material parameters (such as density, viscosity, thermal conductivity, etc.), flow parameters (such as flow rate, inlet velocity, pressure, etc.), and thermodynamic parameters (such as ambient temperature, boundary temperature, etc.). Taking a water-cooled radiator as an example, the density of water is typically set to 998.2 kg / m³. 3 The dynamic viscosity is 0.001003 Pa·s, the flow rate is 1~5 L / min, and the inlet temperature is 60-90℃.

[0078] The calculation result output module 6 is connected to the fluid performance prediction module 5, and is used to receive the fluid performance prediction results, process and summarize them, and then output them. Preferably, the output results include data and visualization charts such as fluid pressure distribution, velocity field distribution, temperature field distribution, convective heat transfer coefficient, and radiator heat transfer coefficient. For engineering applications, special attention is paid to key performance indicators such as pressure drop (usually required to be below 20 kPa), heat transfer efficiency (heat exchange rate required to be above 80%), and temperature uniformity (usually required to have a maximum temperature difference of less than 5℃).

[0079] Reference Figure 2 According to one embodiment of the present invention, the parameter input module 1 includes a model input unit 11, a parameter calculation unit 12, and a parameter output unit 13.

[0080] The model input unit 11 is used to receive the radiator's 3D model and boundary information. In a specific implementation, the radiator's 3D model can be a CAD format 3D model file containing complete geometric information. The boundary information includes the inlet and outlet locations of the flow channels and boundary conditions. For example, for a typical automotive radiator, its 3D model may be a STEP or IGES format file, and the boundary information may include a circular inlet with a diameter of 35mm, a finned area with dimensions of 450mm × 350mm, and multiple branch outlets.

[0081] The parameter calculation unit 12 is connected to the model input unit 11 and is used to calculate the parameters of the flow channel cross-section and the inlet cross-section based on the three-dimensional model and boundary information of the radiator. In a preferred embodiment of the invention, the parameter calculation unit 12 first analyzes the geometric information in the three-dimensional model to identify the inlet cross-section of the flow channel and multiple cross-sectional positions along the flow direction. Then, each cross-section is parameterized to extract its shape parameters. For example, for an elliptical cross-section, its major axis length, minor axis length, and angular direction are calculated; for a rectangular cross-section, its length, width, and directional angle are calculated. In practical applications, the parameter calculation also considers the influence of manufacturing tolerances (typically ±0.1 mm) on the flow channel geometry.

[0082] The parameter output unit 13 is connected to the parameter calculation unit 12 and is used to output the parameters of the flow channel cross-section and the inlet cross-section to the streamline drawing module 2. The parameters are transmitted in the form of a data structure, including information such as cross-section type identifier, position coordinates, and shape parameters. For complex heat sinks, these parameters typically include a complete geometric description of 10 to 20 cross-section positions.

[0083] Reference Figure 3 According to one embodiment of the present invention, the streamline drawing module 2 includes a first input unit 21, a second input unit 22, a streamline calculation unit 23, and a parameter output unit 24.

[0084] The first input unit 21 is used to receive parameters of the flow channel cross-section and the inlet cross-section. These parameters are output from the parameter input module 1 and contain shape and position information of each cross-section of the flow channel. In practical applications, these parameters describe the geometric changes of the radiator flow channel from the inlet to the outlet and are the basic data for streamline calculation.

[0085] The second input unit 22 is used to receive the boundary information. The boundary information includes the flow channel outlet, flow channel inlet, and flow channel boundary, which is used to constrain the flow line calculation process. In radiator design, boundary information is crucial to ensuring that flow lines do not cross solid walls and accurately reflect the fluid flow path.

[0086] The streamline calculation unit 23 is connected to the first input unit 21 and the second input unit 22, and is used to calculate the starting coordinates and streamline point coordinates of the streamline based on the parameters of the flow channel cross section and the inlet cross section, as well as the boundary information. In a preferred embodiment of the present invention, the streamline calculation adopts the principle of fluid dynamics, starting from the inlet cross section point, and gradually calculating the spatial trajectory of the streamline according to the fluid flow law and boundary constraint conditions.

[0087] Specifically, the streamline calculation unit 23 first evenly distributes several starting points (e.g., 20 to 100 points, depending on the complexity of the section) on the inlet cross-section, with each starting point serving as the starting point of a streamline. Then, based on the fluid flow equations and boundary conditions, it progressively calculates the extension path of each streamline until it reaches the channel outlet. The calculation process considers the fluid continuity equation and momentum equation to ensure that the streamlines conform to physical laws.

[0088] For a typical automotive radiator, 36 starting points (6×6 grid) are typically distributed across the inlet section to ensure that the flow channel geometry is adequately captured. In narrow or abruptly changing regions, the point distribution is denser to improve reconstruction accuracy. The streamline calculation step size is typically set to 1 / 20 to 1 / 50 of the flow channel feature size to ensure smooth and accurate streamlines.

[0089] The parameter output unit 24 is connected to the streamline calculation unit 23 and is used to send the streamline starting coordinates and streamline point coordinates to the streamline smoothing module 3. The coordinates are transmitted in the form of a set of points, and each streamline contains a series of three-dimensional spatial point coordinates. In complex heat sink models, each streamline typically contains 100 to 500 points to accurately capture the geometric features of the flow channel.

[0090] Reference Figure 4 According to one embodiment of the present invention, the streamline smoothing module 3 includes a smoothing input unit 31, a smoothing processing unit 32, and a smoothing output unit 33.

[0091] The smoothing input unit 31 receives the starting coordinates and boundary points of the streamlines. This data comes from the output of the streamline drawing module 2 and includes the spatial coordinates of multiple streamlines. For complex radiator channels, these original streamlines may contain locally unsmooth regions, especially at points where the channel changes abruptly.

[0092] The smoothing processing unit 32 is connected to the smoothing input unit 31 and is used to smooth the streamlines and generate smoothed streamline coordinates. In a preferred embodiment of the present invention, the smoothing processing adopts a cubic spline interpolation method, which can be expressed as:

[0093] S i (t)=a i +b i (tt i )+c i (tt i ) 2 +d i (tt i ) 3 ,t∈[t i ,t i+1 ],

[0094] Wherein: S i (t) is the function of the i-th spline curve segment, and the output is the coordinates of a point in three-dimensional space; t is a parameter variable, t∈[t] i ,t i+1 ];a i ,b i ,c i ,d i Let t be the coefficient vector of the spline curve, where each vector contains three components: x, y, and z, determined by solving a system of equations; i and t i+1 The parameter values ​​of adjacent nodes represent the boundary points of curve segments.

[0095] number a i ,b i ,c i ,di The solution should consider the following conditions:

[0096] 1. The curve passes through all given points: S i (t i ) = P i and S i (t i+1 ) = P i+1 , where P i Let i be the coordinates of the i-th original streamline point;

[0097] 2. The first derivative of adjacent curve segments is continuous at the connection point: S′ i (t i+1 )=S′ i+1 (t i+1 );

[0098] 3. The second derivative of adjacent curve segments is continuous at the connection point: S″ i (t i+1 )=S″ i+1 (t i+1 )

[0099] 4. The second derivative of adjacent curve segments is continuous at the connection point: S″ i (t i+1 )=S″ i+1 (t i+1 4. Boundary conditions (usually natural boundary conditions are used, i.e., the second derivatives at both ends of the curve are zero): S″1(t1)=0 and S n "(t n+1 ) = 0;

[0100] For the S-shaped bends commonly found in heat sink channels, smoothing is particularly important. For example, in a heat sink for an electronic device, the original streamlines may exhibit jagged trajectories at the bends, affecting the quality of subsequent surface generation. After cubic spline smoothing, the streamline curvature changes more uniformly, with the radius of curvature typically maintained at more than three times the channel diameter, avoiding fluid separation caused by extreme curvature.

[0101] The smoothing output unit 33 is connected to the smoothing processing unit 32 and is used to send the smoothed streamline coordinates to the flow channel surface meshing module 4. The smoothed streamlines have better continuity and smoothness, providing a better foundation for subsequent surface generation. In practical applications, the smoothed streamlines are usually represented with a higher data density (e.g., 2-3 times the number of points of the original streamlines) to ensure accurate capture of curve details.

[0102] Reference Figure 5According to one embodiment of the present invention, the flow channel surface meshing module 4 includes a surface generation unit 41, a mesh subdivision unit 42, and a boundary mesh optimization unit 43.

[0103] The surface generation unit 41 receives the smoothed streamline coordinates and generates a three-dimensional surface of the flow channel based on the spatial streamline points. In this invention, the surface generation unit 41 is a key innovation, solving the mapping problem from one-dimensional streamlines to three-dimensional surfaces. For the reconstruction of flow channels in automotive radiators, the surface generation process is crucial for accurately representing the flow channel geometry, especially for complex structures such as flat tubes or microchannels.

[0104] Mesh generation unit 42 is connected to surface generation unit 41 and is used for adaptive non-uniform mesh generation based on the three-dimensional surface of the flow channel. Mesh generation unit 42 is another key innovation; it automatically adjusts the mesh density and orientation according to fluid characteristics and geometric features, achieving a rational allocation of mesh resources. In actual radiator analysis, finer meshes are typically required at flow channel inlets, bends, and areas of cross-sectional change to accurately capture complex flow phenomena.

[0105] Boundary mesh optimization unit 43 is connected to mesh generation unit 42 and is used to intelligently optimize the boundary constraints of the mesh to generate an optimized flow channel surface mesh. Boundary mesh optimization unit 43 is the third key innovation; it develops a dedicated boundary processing strategy for the complex boundary conditions unique to heat sinks. In practical applications, the mesh quality at the connection between the heat sink fins and the pipe wall significantly affects the accuracy of heat transfer calculations, requiring special processing to ensure computational stability.

[0106] Reference Figure 6 According to one embodiment of the present invention, the surface generation unit 41 includes a streamline preprocessing subunit 411, a cross-section construction subunit 412, and a surface interpolation subunit 413.

[0107] The streamline preprocessing subunit 411 is used to parameterize the coordinates of the smoothed streamlines and establish topological relationships between streamlines. In a preferred embodiment of the invention, the parameterization process first parameterizes the arc length of each streamline, i.e.:

[0108]

[0109] Where: s i x is the arc length parameter at the i-th point, representing the distance from the streamline origin to the i-th point along the streamline, in millimeters (mm); j ,y j ,z j Let be the three-dimensional coordinates of the j-th point, in millimeters (mm); j is the index of the point, from 1 to i; the summation symbol is ∑. j =1i This represents the sum of distances between all adjacent point pairs from the 1st point to the i-th point.

[0110] Then, the streamline preprocessing subunit 411 normalizes the arc length parameter to the [0,1] interval:

[0111]

[0112] Where: u i The parameter is standardized, dimensionless, and ranges from [0,1]; s i s represents the arc length parameter at the i-th point, in millimeters (mm); n This is the total length of the streamline, i.e., the arc length of the last point, in millimeters (mm).

[0113] For a typical automotive radiator, the total length of the streamline is usually in the range of 300 to 500 mm. Parametric processing ensures that streamlines of different lengths can be expressed in the same parameter space, which facilitates subsequent matching of corresponding points and generation of surfaces.

[0114] Next, the streamline preprocessing subunit 411 analyzes the topological relationships between different streamlines and establishes point-to-point mappings. Preferably, the nearest point projection method is used to determine the mapping relationship between corresponding points on different streamlines, that is, for points on two adjacent streamlines, a correspondence is established based on their parameter values ​​and spatial positions. This is particularly important when dealing with irregular heat sink channels, such as multi-branch structures or variable cross-section channels.

[0115] The section construction subunit 412 is connected to the streamline preprocessing subunit 411 and is used to construct a flow channel section sequence based on the parameterized streamlines. In a preferred embodiment of the present invention, the section construction process is as follows:

[0116] First, select multiple parameter values ​​(usually 30-50) evenly along the streamline direction. For each parameter value, find the corresponding point on all streamlines. For complex heat sink flow channels, such as gradually changing cross-sections or curved sections, the number of cross-sections can be appropriately increased to improve reconstruction accuracy.

[0117] These points then form a cross-sectional profile. For a typical radiator flow channel, the number of points on the cross-sectional profile is usually 20 to 40, depending on the complexity of the cross-sectional shape.

[0118] For each cross-sectional profile, a shape-preserving interpolation method is applied to generate a closed curve, ensuring that the geometric features of the cross-section are preserved. When dealing with flat tube radiators, special attention is paid to preserving corner features during cross-section construction to avoid excessive rounding that could affect the flow analysis results.

[0119] Preferably, the transition between cross-sections employs curvature continuity constraints to ensure a smooth transition between adjacent cross-sections. For complex and changing regions (such as areas where the flow channel changes abruptly), a local densification strategy is adopted to increase the number of cross-sections and improve reconstruction accuracy. For example, at the S-shaped bend of a heat sink in an electronic device, the cross-section spacing can be reduced from the standard 10mm to 3-5mm to accurately capture the bending characteristics.

[0120] The surface interpolation subunit 413 is connected to the section construction subunit 412 and is used to generate a continuous three-dimensional surface based on the flow channel section sequence. In a preferred embodiment of the present invention, a bicubic B-spline surface representation method is used:

[0121]

[0122] Where: S(u,v) is a surface function that outputs three-dimensional coordinates (x,y,z) in millimeters (mm); u,v are parameter variables, u∈[0,1] represents streamline direction parameters, and v∈[0,1] represents cross-sectional circumferential parameters, both of which are dimensionless; P ij The control point grid consists of three-dimensional coordinates for each control point, expressed in millimeters (mm); N i,3 (u) and N j,3 (v) represents the cubic B-spline basis functions, which are dimensionless; n and m are the number of control points in the u and v directions minus 1, respectively, typically n is 30-50 and m is 20-40; double summation symbol. This indicates a weighted summation of all control points.

[0123] The basis function of cubic B-spline is defined as:

[0124]

[0125] Where: N i,0 (u) is a 0th-order B-spline basis function, representing a piecewise constant function; N i,k (u) is the k-th degree B-spline basis function; k is the spline degree, in this embodiment k=3, indicating a cubic spline; u i For the values ​​in the node vector, define the segmented intervals of the spline; the coefficients before the fraction represent the weights of the linear combination and are functions of the parameter u.

[0126] For representing radiator flow channels, open uniform node vectors are typically used to ensure accurate interpolation of boundary points on the surface. In practical applications, such as automotive radiator flow channel modeling, the density of the control point mesh is adjusted according to the complexity of the flow channel; more control points are used in complex and variable regions to improve accuracy.

[0127] Control point grid P ijThe generated surface is determined by solving a system of equations to ensure it precisely passes through all cross-sectional contour points. Preferably, the following physical constraints are added:

[0128]

[0129] in: and , are the partial derivatives of the surface with respect to parameters u and v, respectively, representing the tangent vector of the surface in the corresponding direction; X represents the cross product operation, the result of which is the normal vector; this constraint ensures that the surface does not self-intersect, maintaining geometric validity. For complex heat sink channels, especially multi-branch structures, this constraint can prevent the surface from self-intersecting or folding at the branches.

[0130]

[0131] in: and , , are the second-order partial derivatives of the surface with respect to parameters u and v, respectively, representing the curvature change of the surface in the corresponding directions; |·| represents the magnitude of the vector; ε is a preset threshold, for example, 0.01 mm. (-1) This threshold is used to control the curvature variation of a surface, ensuring a smooth surface. In heat sink flow channel modeling, this threshold is typically set to the reciprocal of 0.5% to 2% of the flow channel feature size to balance accuracy and smoothness.

[0132] Using the above method, the three-dimensional surface generated by the surface interpolation sub-unit 413 can accurately represent the geometric features of the radiator flow channel, providing a high-quality geometric basis for subsequent mesh generation. In practical applications, for a typical automotive radiator, the surface reconstruction accuracy can typically reach ±0.05mm, meeting the geometric requirements of fluid analysis.

[0133] Reference Figure 7 According to one embodiment of the present invention, the meshing unit 42 includes a characteristic analysis subunit 421, a density function construction subunit 422, and a mesh generation subunit 423.

[0134] The characteristic analysis subunit 421 is used to analyze the geometric features of the three-dimensional curved surface of the flow channel and predict fluid characteristics. In a preferred embodiment of the invention, the geometric feature analysis includes curvature analysis and feature edge recognition.

[0135] Curvature analysis calculates the principal curvatures κ1 and κ2 at each point on the surface, as well as the Gaussian curvature K and the mean curvature H:

[0136] K = κ1·κ2,

[0137]

[0138] Where: K is the Gaussian curvature, in mm. (-2)H represents the intrinsic geometric properties of the surface; H is the mean curvature, in mm. (-1) Characterizing the external geometric properties of the surface; κ1 and κ2 are principal curvatures, in mm. (-1) This indicates the maximum and minimum curvature of the surface at that point. For typical radiator channels, the principal curvature values ​​are typically between 0.01 and 0.5 mm. (-1) Within the range, high curvature areas mainly appear at bends and cross-sectional changes.

[0139] Curvature calculation is based on the first and second derivatives of the surface:

[0140]

[0141] Where: eigenvalues ​​represent the eigenvalues ​​of the computed matrix; E, F, G are the first fundamental form coefficients, which are related to the metric properties of the surface; L, M, N are the second fundamental form coefficients, which are related to the curvature properties of the surface; these coefficients are calculated using the partial derivatives of the surface. in is the unit normal vector of the surface, and · represents the dot product operation.

[0142] Feature edge recognition is used to detect feature lines on curved surfaces, such as sharp edges and areas with high curvature changes. Preferably, a threshold method is used to identify feature edges, i.e., when the rate of curvature change exceeds a preset threshold (e.g., 0.5 mm). (-1) When the value is / mm, it is considered a characteristic edge. In radiator flow channel analysis, characteristic edges mainly include the flow channel inlet edge, cross-sectional change points, and fin connection points. The mesh of these areas needs special processing to accurately capture flow characteristics.

[0143] Fluid property prediction is based on a simplified fluid dynamics model to predict the behavior of the fluid within the flow channel. The main predicted properties are as follows:

[0144] 1. Regions with large velocity gradients, such as bends in the flow channel and areas where the cross-section changes.

[0145] 2. Areas with significant pressure changes, such as the inlet, outlet, and abrupt changes in cross-section of the flow channel.

[0146] 3. Areas where vortices may occur, such as flow channel bifurcation and backward steps.

[0147] The predictions employ empirical formulas and simplified computational models; for example, the prediction of velocity gradients:

[0148]

[0149] in: The velocity gradient is estimated in seconds. (-1) Q represents flow rate, in mm.3 / s; A(s) is the cross-sectional area of ​​the flow channel along the streamline parameter s, in mm. 2 ; is the rate of change of cross-sectional area, in mm; s is the distance parameter along the streamline, in mm.

[0150] For a typical automotive radiator, the inlet flow rate is approximately 1–5 L / min (16,667–83,333 mm). 3 / s), the flow channel cross-sectional area is approximately 800-1200 mm² at the inlet. 2 Gradually dispersed into multiple parallel channels, each channel having a cross-sectional area of ​​approximately 10-50 mm². 2 At points where the cross-section changes abruptly, the velocity gradient can reach 100–500 s. (-1) A denser grid is required for accurate capture.

[0151] The density function construction subunit 422 is connected to the characteristic analysis subunit 421, and is used to construct the mesh density function based on the geometric features and fluid properties. In a preferred embodiment of the invention, the mesh density function comprehensively considers geometric and physical factors:

[0152] ρ(u,v)=α·κ(u,v)+β·τ(u,v)+γ·θ(u,v),

[0153] Where: ρ(u,v) is the grid density function, dimensionless, representing a multiple relative to the reference density; κ(u,v) is the surface curvature function, taken as κ(u,v) = max(|κ1|,|κ2|), with units of mm. (-1) τ(u,v) is the fluid shear stress estimation function, in Pa, typically ranging from 0.01 to 10 Pa; θ(u,v) is the flow channel cross-sectional area change rate function, dimensionless; α, β, γ are weighting coefficients, set according to actual needs, with typical values ​​of α = 0.4 mm and β = 0.04 Pa. (-1) γ = 0.2, making the dimensions of each quantity consistent and their contributions comparable.

[0154] In actual radiator flow channel mesh generation, these weighting coefficients can be adjusted according to specific cases. For example, for high Reynolds number flows (such as automotive radiators under high-speed conditions), the β value can be increased to 0.06–0.08 Pa. (-1) To better capture fluid shear effects; for complex geometries (such as microchannel heat sinks), the α value can be increased to 0.5–0.6 mm to more accurately represent geometric features.

[0155] To prevent the mesh from being too concentrated or too sparse, set upper and lower density limits:

[0156] ρ min ≤ρ(u,v)≤ρmax ,

[0157] Where: ρ min and ρ max The preset threshold value is typically ρ. min =0.2,ρ max =5.0, indicating a multiple relative to the baseline density. This ensures a reasonable range of mesh size variation, avoiding wasted computational resources or insufficient accuracy.

[0158] Mesh generation subunit 423 is connected to density function construction subunit 422, and is used to generate an adaptive non-uniform mesh according to the mesh density function. In a preferred embodiment of the present invention, the mesh generation adopts an anisotropic mesh generation algorithm, taking into account the influence of the fluid flow direction.

[0159] First, define the mesh metric tensor:

[0160] M(u,v)=R(u,v)·Λ(u,v)·RT(u,v),

[0161] Where: M(u,v) is the metric tensor, a 3×3 symmetric positive definite matrix that determines the size and orientation of the local mesh; R(u,v) is the local coordinate rotation matrix, a 3×3 matrix that aligns the principal axis with the main fluid flow direction; Λ(u,v) is the characteristic length diagonal matrix, a 3×3 diagonal matrix that contains the mesh size for both the principal and secondary directions; R T The transpose of (u,v).

[0162] For heat sink flow channel mesh generation, the characteristic length diagonal matrix is ​​typically set as follows:

[0163]

[0164] Where h1(u,v) is the grid size in the flow direction, and h2(u,v) and h3(u,v) are the grid sizes in the lateral direction, all in mm. In the high Reynolds number flow region, the flow direction grid can be appropriately lengthened, typically h1(u,v):h2(u,v):h3(u,v) = 2:1:1 to 3:1:1; in the low Reynolds number or complex flow region, the grid size ratio is close to 1:1:1 to more accurately capture the three-dimensional flow characteristics.

[0165] Then, the mesh node spacing is calculated based on the metric tensor:

[0166]

[0167] Where: h(u,v) is the local mesh size in mm; h0 is the reference mesh size, usually taken as 1 / 20 to 1 / 50 of the flow channel characteristic size, and 0.2-1.0 mm for a typical heat sink; ρ0 is the reference density value, usually taken as 1.0, dimensionless; the exponent -1 / 2 indicates that the mesh size is inversely proportional to the square root of the density function, which is to maintain the relationship between the mesh cell volume and the density function in three-dimensional space.

[0168] Finally, the surface mesh is generated using Delaunay triangulation or the forward front method, ensuring that the mesh quality meets requirements such as a minimum angle of 20°, a maximum angle of 135°, and a size ratio between adjacent elements not exceeding 1.5. These quality standards are crucial for computational stability in radiator flow channel analysis, especially in thermo-fluid coupling analysis.

[0169] Using the above method, the adaptive non-uniform mesh generated by mesh generation element 42 can effectively adapt to fluid characteristics and geometric features, improving computational efficiency and accuracy. In practical applications, at the same level of accuracy, the adaptive mesh can reduce the number of elements by 50% to 70% compared to the uniform mesh, significantly improving computational efficiency. For example, for the flow channel analysis of a typical automotive radiator, the adaptive mesh can reduce the number of elements from approximately 3 million to approximately 1 million while maintaining or improving computational accuracy.

[0170] Reference Figure 8 According to one embodiment of the present invention, the boundary mesh optimization unit 43 includes a boundary identification subunit 431, a boundary layer mesh subunit 432, a transition zone processing subunit 433, and a quality optimization subunit 434.

[0171] The boundary identification subunit 431 is used to identify and classify mesh boundary types. In a preferred embodiment of the invention, the boundary types include:

[0172] 1. Fluid inlet boundary: usually circular, rectangular, or elliptical;

[0173] 2. Fluid outlet boundary: may be a single branch or multiple branches;

[0174] 3. Solid wall boundary: inner surface of the pipe wall;

[0175] 4. Fin contact surface: The complex boundary unique to radiators;

[0176] 5. Symmetrical boundary: If there is a symmetrical design;

[0177] Boundary identification is based on geometric features and preset markers, with specific processing strategies and parameters assigned to each type of boundary. For example, inlet boundaries require fine meshes to accurately capture inlet flow characteristics; wall boundaries require boundary layer meshes to resolve boundary layer flow; and fin contact surfaces require special treatment to ensure the stability of fluid-structure interaction calculations.

[0178] In practical radiator analysis, boundary identification has a significant impact on the accuracy of calculation results. Taking automotive engine radiators as an example, inlet boundary treatment affects the uniformity of fluid distribution; wall boundary treatment affects the accuracy of heat transfer coefficient calculation; and fin contact surface treatment relates to the accuracy of heat transfer from fluid to solid.

[0179] Boundary layer mesh sub-unit 432 is connected to boundary identification sub-unit 431 and is used to construct a boundary layer mesh near the wall boundary. In a preferred embodiment of the invention, the boundary layer mesh adopts a structured mesh perpendicular to the wall, with the thickness increasing layer by layer. The thickness of the first layer is determined based on the estimated Reynolds number.

[0180]

[0181] Where: y1 is the thickness of the first layer, in mm; y + ρ is the dimensionless wall distance, typically taken as 30-100, depending on the computational model (30-50 for turbulent models such as the k-ε model, and 50-100 for the wall function method); μ is the fluid dynamic viscosity, in kg / (m·s) or Pa·s, approximately 0.001 Pa·s for water at 20℃; ρ is the fluid density, in kg / m³. 3 The water content is approximately 998 kg / m³. 3 ;u τ The friction velocity, expressed in m / s, can be approximated as 5% of the inlet velocity (e.g., when the inlet velocity is 2 m / s, u...). τ (Approximately 0.1 m / s).

[0182] For a typical water-cooled radiator, under normal operating conditions (flow velocity 1–3 m / s, Reynolds number 5,000–30,000), the initial layer thickness is typically calculated to be 0.03–0.1 mm. This range ensures accurate capture of boundary layer flow without excessively increasing the mesh count.

[0183] The thickness of adjacent layers increases geometrically:

[0184] y i =y1·(1+r) i-1 ,

[0185] Where: y iLet be the thickness of the i-th layer, in mm; r is the growth rate, typically between 1.1 and 1.3. Smaller values ​​provide more accurate results but increase computational complexity. For radiator flow channel analysis, the total boundary layer thickness is usually 10%–20% of the flow channel characteristic dimensions, and the number of layers is typically 5–15. For example, for a flat tube radiator with a cross-section of 10 mm × 2 mm, the total boundary layer thickness is approximately 0.2–0.4 mm, consisting of 8–10 layers.

[0186] Applying a dedicated boundary layer strategy to the fin contact surface, including the construction of a double-sided boundary layer and ensuring contact surface continuity, is crucial for accurately simulating heat transfer within the radiator. For example, in the fin root region, the treatment of the fluid-structure boundary layer directly affects the accuracy of the calculated local heat transfer coefficient, and consequently, the assessment of the overall heat dissipation efficiency.

[0187] The transition region processing subunit 433 is connected to the boundary recognition subunit 431 and is used to process the transition region mesh between boundaries. In a preferred embodiment of the present invention, the transition region processing strategy includes:

[0188] 1. Progressive mesh refinement: a smooth transition from coarse to fine mesh areas, with the size ratio of adjacent units typically controlled between 1.1 and 1.2;

[0189] 2. Smooth directional transition: Ensure that the grid line direction changes smoothly, and avoid grid line angles less than 45° or greater than 135°;

[0190] 3. Unit shape control: Avoid excessive twisting or stretching. Generally, the aspect ratio should not exceed 5:1 and the tilt angle should not exceed 0.7.

[0191] For special transition areas such as inlet fillets, a localized refinement and shape preservation strategy is employed to ensure geometric accuracy and mesh quality. For example, at the radiator inlet fillets, the mesh size is typically 30% to 50% smaller than the surrounding area to accurately capture fluid inflow characteristics.

[0192] The quality optimization sub-unit 434 is connected to the boundary layer mesh sub-unit 432 and the transition zone processing sub-unit 433, and is used to perform multi-objective quality optimization on the boundary mesh. In a preferred embodiment of the present invention, the mesh quality evaluation adopts multiple indicators:

[0193] Q = min{w1·q} ortho +w2·q aspect +w3·q skew +w4·q warp},

[0194] Where: Q is the overall quality score, ranging from 0 to 1, with a closer to 1 indicating better quality; q ortho The orthogonality index ranges from 0 to 1, with an ideal value of 1; qaspect The aspect ratio is a metric, ranging from 0 to 1, with an ideal value of 1; q skew The slope index ranges from 0 to 1, with an ideal value of 0; q w ARP is the warpage index, ranging from 0 to 1, with an ideal value of 0; w1, w2, w3, and w4 are weighting coefficients, with typical values ​​of w1 = 0.4, w2 = 0.3, w3 = 0.2, and w4 = 0.1, summing to 1.

[0195] Mesh optimization employs an iterative relaxation algorithm to improve mesh quality by fine-tuning point positions.

[0196] x i ′=x i +Δx(F pres ,F smth ,F shew ,θ),

[0197] Where: x′ i The optimized node position is represented by a 3D coordinate vector in mm; x i The original node position is represented by a three-dimensional coordinate vector in mm; Δx is the displacement adjustment, a three-dimensional vector in mm, determined by multiple force terms; F pres To maintain the mesh shape using conformal force, a vector quantity in N; F smth To smooth out forces and reduce mesh irregularities, a vector is used, with units of N; F skew To eliminate tilting forces and reduce element tilt, a vector is used, with units in N; θ is the flow direction angle, a scalar, with units in radians, to ensure the mesh adapts to the flow direction.

[0198] For fluid-structure interaction interfaces, it is crucial to ensure precise correspondence between the mesh nodes of the fluid and solid domains at the interface. This involves one-to-one node mapping or controlled interpolation to guarantee numerical flux conservation at the interface. This is particularly important in the thermofluid-thermal coupling analysis of radiators, directly impacting the accuracy of heat transfer calculations. For example, at the connection between the fins and the tube wall, precise mesh node correspondence can reduce numerical errors and improve the accuracy of heat flux calculations by approximately 15% to 25%.

[0199] Using the above method, the boundary mesh generated by boundary mesh optimization element 43 can meet the computational requirements of complex radiator flow channels, improving computational stability and result reliability. In practical applications, boundary mesh optimization can reduce defective elements (quality score below 0.3) by more than 95%, significantly improving computational stability and convergence speed. For typical automotive radiator analysis, this means that the number of iterations can be reduced from 5000+ to 1000-2000, and the computation time can be shortened by 60% to 70%.

[0200] Reference Figure 9According to one embodiment of the present invention, the fluid performance prediction module 5 includes a grid input unit 51, a parameter setting unit 52, a prediction calculation unit 53, and a result generation unit 54.

[0201] Mesh input unit 51 is used to receive the flow channel surface mesh. This mesh data comes from the flow channel surface meshing module 4 and contains complete node coordinates, cell topology relationships, and boundary identification information. For a typical automotive radiator analysis, the optimized mesh typically contains approximately 500,000 to 2 million cells, which is sufficient to accurately capture flow and heat transfer characteristics.

[0202] The parameter setting unit 52 is used to set fluid performance analysis parameters. In a preferred embodiment of the present invention, these parameters include:

[0203] 1. Fluid material parameters: density (e.g., water is 998.2 kg / m³) 3 ), dynamic viscosity (e.g., 0.001003 Pa·s for water), specific heat capacity (e.g., 4182 J / (kg·K) for water), thermal conductivity (e.g., 0.6 W / (m·K) for water), etc. For automotive engine coolant, its temperature dependence is usually considered, such as viscosity decreasing by about 2% to 3% / ℃ as temperature increases.

[0204] 2. Flow parameters: flow rate (typically 0.1–10 L / min), inlet velocity (typically 0.1–5 m / s), inlet temperature (typically 293–363 K or 20–90 °C), inlet pressure (typically 101325–500000 Pa), etc. For example, under normal driving conditions, the coolant flow rate of a car radiator is approximately 60–120 L / min, and the inlet temperature is approximately 85–95 °C.

[0205] 3. Boundary conditions: The wall surface is a no-slip boundary, the fin surface is a heat flux boundary or a temperature boundary, and the outlet is a pressure outlet, etc. For coupled analysis, the fin surface temperature is usually set to the ambient temperature plus a temperature difference of 5-15℃, or a specified heat flux density, typically 1000-5000 W / m². 2 .

[0206] 4. Calculate control parameters: convergence criteria (usually residual less than 1e-4), maximum number of iterations (usually 1000-5000), time step (usually 0.001-0.01s for transient calculations), etc. These parameters have a significant impact on computational efficiency and stability.

[0207] The prediction calculation unit 53 is connected to the mesh input unit 51 and the parameter setting unit 52, and is used to perform fluid performance prediction calculations based on the flow channel surface mesh and the fluid performance analysis parameters. In a preferred embodiment of the present invention, the calculation is based on the RANS (Reynolds-Averaged Navier-Stokes) equations, including the continuity equation, momentum equation, and energy equation:

[0208] Continuity equation:

[0209]

[0210] in: Let be the partial derivative of density with respect to time, representing the rate of change of density over time, with units of kg / (m³). 3 ·s); ρ is the fluid density, in kg / m³ 3 ; This is a velocity vector, with units of m / s; Here, is the divergence operator, representing the net outflow rate of flux; The divergence of mass flow rate, in kg / (m³). 3 This equation expresses the principle of conservation of mass.

[0211] Momentum equation:

[0212]

[0213] in: Let be the partial derivative of momentum with respect to time, representing the rate of change of momentum per unit volume with time, with units of kg / (m³). 2·s 2); The divergence of momentum flux, in kg / (m²) 2·s 2); ρ represents the pressure gradient, in Pa / m; p represents the pressure, in Pa. This is the stress tensor, with units of Pa. The divergence of the stress tensor represents the viscous force, with units of N / m. 3 ; This is the acceleration due to gravity, measured in m / s². 2 ; The weight per unit volume, expressed in N / m³. 3 This equation expresses the principle of conservation of momentum.

[0214] Energy equation:

[0215]

[0216] in: The partial derivative of total energy with respect to time represents the rate of change of energy per unit volume over time, with units of W / m³. 3 E represents the total energy per unit mass, including internal energy and kinetic energy, measured in J / kg; ρE represents the total energy per unit volume, measured in J / m³. 3 ; The divergence of energy flux, in units of W / m 3 ;k eff The effective thermal conductivity is expressed in W / (m·K). The temperature gradient is expressed in K / m; T represents temperature, expressed in K. This is the conductive heat flux density, with units of W / m. 2 ; This is the effective stress tensor, expressed in Pa. This is a velocity vector, with units of m / s; Viscous dissipation rate, in W / m 2 This equation expresses the principle of energy conservation.

[0217] For closed equation sets, an appropriate turbulence model is employed, such as the k-ε model, k-ω model, or SST model. Preferably, for flow within radiator channels, the SST k-ω model is used, as it performs well in both near-wall and free-flow regions. The SST k-ω model combines the advantages of the k-ε model in free-flow regions with the advantages of the k-ω model in near-wall regions, making it particularly suitable for separated flows and heat transfer calculations.

[0218] The calculations employ the finite volume method, using the SIMPLE or PISO algorithm to solve the discretized equations. To improve computational stability, appropriate relaxation factors (typically 0.3–0.7) and gradient limiters are used. For typical radiator flow channel analysis, the preferred numerical discretization schemes are the second-order upwind scheme (convection term) and the central difference scheme (diffusion term), balancing computational accuracy and stability.

[0219] In predicting radiator performance, the following key indicators should also be considered:

[0220] 1. Heatsink voltage drop Δp = p in -p out The unit is Pa, and the typical range is 5,000-20,000 Pa;

[0221] 2. Heat transfer coefficient: The unit is W / (m 2 •K), typically ranging from 1,000 to 5,000 W / (m³). 2 ·K);

[0222] 3. Heat dissipation: The unit is W, and the typical range is 5-50kW;

[0223] 4. Heat dissipation efficiency: Dimensionless, with a typical range of 0.3–0.7;

[0224] Where: p in and p out Here, represents the inlet and outlet pressures, respectively, in Pa; q represents the heat flux density, in W / m³. 2 A represents the heat exchange area, in meters (m²). 2 ΔT is the temperature difference, in Kelvin (K). Mass flow rate, unit: kg / s; c p Specific heat capacity, expressed in J / (kg·K); T in and T out These are the inlet and outlet temperatures, respectively, in K; T amb The ambient temperature is expressed in Kelvin (K).

[0225] The result generation unit 54 is connected to the prediction calculation unit 53 and is used to generate fluid performance prediction results and send them to the calculation result output module 6. In a preferred embodiment of the present invention, the prediction results include:

[0226] 1. Fluid field data: Three-dimensional distribution data such as velocity field, pressure field, and temperature field. For example, the velocity field shows the distribution of fluid within a radiator, helping to identify low-velocity areas (which may contain hot spots) and high-velocity areas (which may lead to erosion).

[0227] 2. Integral performance parameters: Overall performance indicators such as flow resistance (pressure drop), heat transfer coefficient, and heat dissipation. These parameters directly reflect the working performance of the radiator and are key indicators for design optimization.

[0228] 3. Key data points: Local data such as velocity, pressure, and temperature at specific locations. This data helps in analyzing local hot spots or areas of abnormal flow within the radiator.

[0229] 4. Visualization Results: Intuitive representations such as isosurfaces, streamlines, and vector diagrams. These intuitive visualizations help designers understand the flow and heat transfer characteristics inside the radiator.

[0230] These results are passed to the calculation result output module 6 in a standard data format for subsequent processing and presentation. For practical engineering applications, the results are usually presented in the form of a technical report, including performance index comparisons, optimization suggestions, and visualization charts.

[0231] Reference Figure 10 The present invention also provides a method for three-dimensional reconstruction of radiator flow channels and prediction of fluid performance, comprising the following steps:

[0232] Step S1: Receive the 3D model and boundary information of the radiator, and calculate the parameters of the flow channel section and the inlet section;

[0233] Step S2: Based on the parameters of the flow channel cross section and the inlet cross section, calculate and determine the starting coordinates and boundary points of the streamlines;

[0234] Step S3: Smooth the streamlines to generate smoothed streamline coordinates;

[0235] Step S4: Receive the smoothed streamline coordinates, generate a three-dimensional surface based on spatial points, perform adaptive non-uniform mesh partitioning, and implement boundary constraint intelligent mesh optimization to generate the flow channel surface mesh;

[0236] Step S5: Based on the flow channel surface mesh, input the fluid performance analysis parameters and perform fluid performance prediction;

[0237] Step S6: Receive the fluid performance prediction results, organize and summarize them, and then output them.

[0238] In a preferred embodiment of the present invention, step S4 further includes:

[0239] Step S41: Parametrically process the smoothed streamline coordinates to establish topological relationships between streamlines;

[0240] Step S42: Construct a flow channel cross-section sequence based on the parameterized streamlines;

[0241] Step S43: Generate a continuous three-dimensional surface based on the flow channel cross-section sequence;

[0242] Step S44: Analyze the geometric features of the three-dimensional surface and predict the fluid properties;

[0243] Step S45: Construct a mesh density function based on the geometric features and fluid properties;

[0244] Step S46: Generate an adaptive non-uniform mesh based on the mesh density function;

[0245] Step S47: Identify and classify mesh boundary types;

[0246] Step S48: Construct a boundary layer mesh near the wall boundary;

[0247] Step S49: Process the mesh in the transition region between boundaries;

[0248] Step S410: Perform multi-objective quality optimization on the boundary mesh to generate the final flow channel surface mesh.

[0249] Taking a car engine radiator as an example, this radiator consists of a series of parallel-arranged flat tubes and corrugated fins. Applying the method of this invention, the radiator CAD model is first imported, and the flow channel geometric parameters (flat tube length 450mm, cross-sectional dimensions 10mm×2mm, inlet diameter 35mm) and boundary conditions are extracted. Then, 36 starting points (6×6 grid arrangement) are generated on the inlet cross-section, and the streamline trajectory is calculated up to the outlet. Cubic spline interpolation is applied to smooth the streamlines and eliminate local irregularities.

[0250] Based on smoothed streamlines, the system generates 45 cross-sections along the length of the flow channel, each containing 36 points. The flow channel surface is generated using bicubic B-spline interpolation, ensuring a smooth surface that accurately represents the channel's geometric features. By analyzing the surface curvature and predicting fluid properties, a mesh density function is constructed, and the mesh is refined at the flow channel inlet, bends, and cross-sectional changes, generating an adaptive mesh containing approximately 150,000 elements.

[0251] For the wall boundary, the system generates an 8-layer boundary layer mesh, with the first layer having a thickness of 0.05 mm and a growth rate of 1.2. A special treatment strategy is applied to the fin contact area to ensure mesh matching at the fluid-structure interface. Finally, multi-objective optimization is performed on the boundary mesh to improve the overall mesh quality, reducing the proportion of defective elements (quality below 0.3) from the initial 8% to below 0.3%.

[0252] Fluid parameters (water, flow rate 80 L / min, inlet temperature 90 °C) and boundary conditions (wall temperature 45 °C, outlet pressure ambient pressure) were set, and fluid performance prediction calculations were performed. The results show that the radiator pressure drop is 12,500 Pa, and the average heat transfer coefficient is 3,200 W / (m²). 2 The total heat dissipation is 32kW (K), with a heat dissipation efficiency of 0.62. The system generates a complete report including velocity field, temperature field distribution, and key performance indicators, providing a basis for radiator design optimization.

[0253] The specific implementation methods of the above steps are the same as the functional implementation methods of the corresponding modules in the foregoing embodiments, and will not be repeated here.

[0254] The radiator flow channel 3D reconstruction and fluid performance prediction system and method provided by this invention can automatically complete the entire process from radiator 3D model to fluid performance prediction results, improve modeling accuracy and computational efficiency, improve boundary processing quality, enhance prediction capabilities, and provide reliable technical support for radiator design optimization.

[0255] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art can make various modifications or variations without departing from the spirit and scope of the present invention, and all such modifications or variations fall within the protection scope of the present invention.

Claims

1. A three-dimensional reconstruction and fluid performance prediction system for radiator flow channels, characterized in that, include: The parameter input module is used to receive the 3D model and boundary information of the heat sink and calculate the parameters of the flow channel section and the inlet section. The streamline drawing module, connected to the parameter input module, is used to receive parameters of the flow channel cross section and the inlet cross section, and to calculate and determine the starting coordinates and boundary points of the streamlines; A streamline smoothing module, connected to the streamline drawing module, is used to receive the streamline starting coordinates and boundary points, and to smooth the streamline. The flow channel surface meshing module is connected to the streamline smoothing module. It is used to receive the smoothed streamline coordinates, perform parameterization on the smoothed streamline coordinates to establish topological relationships between streamlines, construct a flow channel cross-section sequence based on the parameterized streamlines, generate a continuous three-dimensional surface based on the flow channel cross-section sequence, and perform adaptive non-uniform mesh partitioning and boundary constraint intelligent mesh optimization. The fluid performance prediction module is connected to the flow channel surface meshing module and is used to receive the flow channel surface mesh, input fluid performance analysis parameters, and predict fluid performance. as well as The calculation result output module is connected to the fluid performance prediction module and is used to receive the fluid performance prediction results, organize and summarize them, and then output them.

2. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 1, characterized in that, The parameter input module includes: The model input unit is used to receive the 3D model and boundary information of the heat sink. A parameter calculation unit, connected to the model input unit, is used to calculate the parameters of the flow channel cross-section and inlet cross-section based on the 3D model and boundary information of the heat sink; and The parameter output unit, connected to the parameter calculation unit, is used to output the parameters of the flow channel section and the inlet section to the streamline drawing module.

3. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 1, characterized in that, The streamline drawing module includes: The first input unit is used to receive the parameters of the flow channel cross section and the inlet cross section; The second input unit is used to receive the boundary information; A streamline calculation unit, connected to the first input unit and the second input unit, is used to calculate the streamline starting coordinates and streamline point coordinates based on the parameters of the flow channel cross-section and the inlet cross-section, as well as the boundary information; and The parameter output unit, connected to the streamline calculation unit, is used to send the streamline starting coordinates and streamline point coordinates to the streamline smoothing module.

4. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 1, characterized in that, The streamline smoothing module includes: A smoothing input unit is used to receive the streamline starting coordinates and boundary points; A smoothing processing unit, connected to the smoothing input unit, is used to smooth the streamlines and generate smoothed streamline coordinates; and A smoothing output unit, connected to the smoothing processing unit, is used to send the smoothed streamline coordinates to the flow channel surface meshing module.

5. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 1, characterized in that, The flow channel surface meshing module includes: A surface generation unit is used to receive the smoothed streamline coordinates and generate a three-dimensional surface of the flow channel based on the spatial streamline points. A mesh generation unit, connected to the surface generation unit, is used for adaptive non-uniform mesh generation based on the three-dimensional surface of the flow channel; and A boundary mesh optimization unit, connected to the mesh subdivision unit, is used to perform intelligent boundary constraint optimization on the mesh to generate an optimized flow channel surface mesh.

6. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 5, characterized in that, The surface generation unit includes: The streamline preprocessing subunit is used to parameterize the smoothed streamline coordinates and establish topological relationships between streamlines. A section construction subunit, connected to the streamline preprocessing subunit, is used to construct a sequence of flow channel sections based on the parameterized streamlines; and The surface interpolation subunit, connected to the cross-section construction subunit, is used to generate a continuous three-dimensional surface based on the flow channel cross-section sequence.

7. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 5, characterized in that, The mesh subdivision unit includes: The characteristic analysis subunit is used to analyze the geometric features of the three-dimensional curved surface of the flow channel and predict the fluid characteristics; A density function construction sub-unit, connected to the characteristic analysis sub-unit, is used to construct a mesh density function based on the geometric features and fluid properties; and A mesh generation sub-unit, connected to the density function construction sub-unit, is used to generate an adaptive non-uniform mesh based on the mesh density function.

8. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 5, characterized in that, The boundary mesh optimization unit includes: Boundary identification sub-unit, used to identify and classify mesh boundary types; Boundary layer mesh sub-units, connected to the boundary identification sub-units, are used to construct boundary layer meshes near the wall boundaries; A transition region processing subunit, connected to the boundary identification subunit, is used to process the transition region mesh between boundaries; and The quality optimization sub-unit, connected to the boundary layer mesh sub-unit and the transition zone processing sub-unit, is used to perform multi-objective quality optimization on the boundary mesh.

9. The radiator flow channel three-dimensional reconstruction and fluid performance prediction system according to claim 1, characterized in that, The fluid performance prediction module includes: A mesh input unit is used to receive the flow channel surface mesh; The parameter setting unit is used to set fluid performance analysis parameters; A prediction calculation unit, connected to the mesh input unit and the parameter setting unit, is used to perform fluid performance prediction calculations based on the flow channel surface mesh and the fluid performance analysis parameters; and The result generation unit is connected to the prediction calculation unit and is used to generate fluid performance prediction results and send them to the calculation result output module.

10. A method for three-dimensional reconstruction of radiator flow channels and prediction of fluid performance, characterized in that, include: Receive the 3D model and boundary information of the radiator, and calculate the parameters of the flow channel section and the inlet section; Based on the parameters of the flow channel cross section and the inlet cross section, the starting coordinates and boundary points of the streamlines are calculated and determined; The streamlines are smoothed to generate smoothed streamline coordinates; The smoothed streamline coordinates are received, and the smoothed streamline coordinates are parametrically processed to establish topological relationships between streamlines. Based on the parametrically processed streamlines, a flow channel section sequence is constructed, and a continuous three-dimensional surface is generated based on the flow channel section sequence. Adaptive non-uniform mesh partitioning is performed, and boundary constraint intelligent mesh optimization is implemented to generate a flow channel surface mesh. Based on the flow channel surface mesh, fluid performance analysis parameters are input to predict fluid performance. as well as Receive fluid performance prediction results, organize and summarize them, and then output them.

Images