A method for airflow simulation modeling in cleanrooms of electronics factories

Abstract

This invention relates to a method for simulating airflow modeling in cleanrooms of electronic factories, belonging to the field of simulation modeling technology. It solves the problem that existing modeling methods cannot simultaneously guarantee accuracy and speed / efficiency. The method includes: obtaining 3D information of simulation primitives based on various design drawings of the cleanroom; performing modeling in CFD preprocessing software based on the 3D information of the simulation primitives to generate a geometric model; treating all FFUs and all raised floors in the geometric model as a single unit corresponding to a face; meshing the geometric model, using each vertex of each FFU and each raised floor in the simulation primitives as virtual nodes; reconstructing the face mesh of the FFUs and raised floors; optimizing all face meshes; and then reconstructing and optimizing the volume mesh to obtain the input model for airflow simulation calculations. This method simultaneously improves modeling efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of simulation modeling technology, and in particular to a method for simulating airflow modeling in a cleanroom of an electronics factory. Background Technology

[0002] Computational Fluid Dynamics (CFD) technology, as an effective airflow simulation and analysis tool, can provide technical support for the optimization of airflow organization in cleanrooms. However, the accuracy of CFD analysis depends primarily on the accuracy of the modeling. Electronic factories are characterized by large areas, wide spans, and numerous components in their cleanroom air conditioning systems with complex layouts. Taking a 30,000-square-meter space as an example, depending on the different cleanroom requirements, the cleanroom system would require several thousand to tens of thousands of FFUs (Fan Filter Units), and nearly 100,000 raised floors. Furthermore, the layout of the raised floors requires extensive iterative calculations, making the modeling workload and difficulty substantial. Therefore, the current bottleneck for airflow simulation in electronic factories is how to quickly and accurately construct models of the FFUs and raised floor layers.

[0003] Currently, there are two techniques for CFD simulation of airflow organization in electronics factories. One method uses precise modeling, where FFUs and raised floors are modeled individually and arranged according to the actual shape of the equipment installation. The other method uses simplified modeling, representing a large area of ​​FFUs or raised floors with a single equivalent value.

[0004] Using precise modeling methods, constructing geometric models for thousands of FFUs and tens of thousands of raised floor panels is an enormous undertaking. Furthermore, because each FFU and each raised floor panel may differ—for example, the outlet air velocity of some FFUs may vary, or the opening ratio of the raised floor panels may differ—setting boundary conditions in the later stages is not only labor-intensive but also prone to errors. This results in long modeling cycles, high computational costs, and an inability to adapt to the comparison and optimization of multiple schemes. Conversely, simplified modeling methods cannot reflect real airflow conditions, leading to a lack of modeling accuracy and consequently, the inability to provide optimization and adjustment solutions, thus negating the purpose of airflow simulation. Summary of the Invention

[0005] Based on the above analysis, the present invention aims to provide a method for simulating and modeling airflow in cleanrooms of electronic factories, in order to solve the problem that existing modeling methods cannot simultaneously guarantee accuracy and speed and efficiency.

[0006] This invention provides a method for simulating and modeling airflow in a cleanroom of an electronics factory, comprising the following steps:

[0007] Based on various design drawings of cleanrooms in electronic factories, obtain the three-dimensional information of simulation elements;

[0008] Based on the 3D information of the simulation primitives, modeling is performed in CFD preprocessing software to generate a geometric model; among them, all FFUs and all raised floors in the simulation primitives are treated as a whole corresponding to a surface;

[0009] The geometric model is meshed to obtain the surface meshes of the FFU and the raised floor. Each vertex of each FFU and each raised floor in the simulation primitive is used as a virtual node. After reconstructing the surface meshes of the FFU and the raised floor, all surface meshes are optimized. Then, the volume mesh is reconstructed and optimized to obtain the input model for airflow simulation calculation.

[0010] Based on the further improvement of the above method, each vertex of each FFU and each raised floor in the simulation primitive is used as a virtual node to reconstruct the surface mesh of the FFU and the raised floor, including:

[0011] Based on the three-dimensional information of the simulation primitives, the two-dimensional coordinates of each vertex of each FFU and each raised floor are obtained, and then the virtual area of ​​each FFU and each raised floor is constructed.

[0012] Adjust the mesh nodes of the mesh elements in the surface mesh of FFU and raised floor according to the positional relationship between the mesh elements in the surface mesh and the virtual region;

[0013] The system identifies whether the number of mesh nodes inside the virtual region of each FFU and the virtual region of each raised floor is less than a set threshold. If it is less than the set threshold, the system adds a corresponding number of mesh nodes based on the difference between the number of mesh nodes and the threshold to obtain the reconstructed surface mesh of the FFU and the raised floor.

[0014] Based on a further improvement to the above method, the mesh nodes of the mesh elements in the surface mesh of the FFU and the raised floor are adjusted according to the positional relationship between the mesh elements in the surface mesh and the virtual region, including:

[0015] Each virtual region is extracted sequentially as the current virtual region, and the positional relationship between each grid cell in the surface mesh and the current virtual region is identified sequentially:

[0016] If some of the grid nodes of a certain grid cell are outside the current virtual region but within the region tolerance, the grid nodes outside can be adjusted by moving the nodes.

[0017] If some of the grid nodes of a certain grid cell are outside the current virtual region and outside the region tolerance range, then the grid nodes on the outside are adjusted by grid splitting.

[0018] A further improvement to the above method involves adjusting external mesh nodes through node movement, including:

[0019] Find the first intersection point between any grid edge containing an external grid node and the current virtual region, and adjust the position of the external grid node to the first intersection point.

[0020] A further improvement to the above method involves adjusting the external mesh nodes through mesh splitting, including:

[0021] Obtain the two second intersection points between the grid cell and the current virtual region. If the distance between the two second intersection points is less than the minimum size of the grid cell, the position of the outer grid node will be adjusted to either of the second intersection points. Otherwise, the two second intersection points and the inner grid node will be combined to form a new grid.

[0022] Based on the above method, before optimizing all surface meshes, the method also includes: establishing quality diagnosis conditions for surface meshes based on the diagnostic indicators of surface meshes and performing quality diagnosis on each surface mesh cell.

[0023] Before optimizing the volume mesh, the method also includes: establishing quality diagnosis conditions for the volume mesh based on the diagnostic indicators of the volume mesh, and performing quality diagnosis on each volume mesh cell.

[0024] Based on further improvements to the above methods, the diagnostic indicators for surface meshes include: aspect ratio of surface mesh elements in non-boundary layers, aspect ratio of surface mesh elements in boundary layers, interior angles of surface mesh elements, parallelism of quadrilateral meshes, skewness of surface mesh elements, and orthogonality of surface mesh elements; the diagnostic indicators for volume meshes include: distortion, volume change rate, Jacobian matrix, orthogonality of volume mesh elements, and aspect ratio of volume mesh elements.

[0025] Further improvements to the above method are made to optimize all surface meshes and volume meshes using an improved Laplacian smoothing method. The improved Laplacian smoothing method is obtained by adding weights to the adjacent vertices of the vertex to be optimized in the Laplacian smoothing method.

[0026] Based on the above method, a further improvement, the improved Laplace smoothing method, calculates the new position of the vertex to be optimized using the following formula:

[0027] ,

[0028] in, and They represent the first The positions of the vertices before and after optimization This represents the relaxation factor, used to control the movement step size; Indicates the first The number of adjacent vertices of a vertex. Indicates the first The vertex of the first vertex The positions of the adjacent vertices Indicates the first The vertex of the first vertex The weights of the adjacent vertices, Indicates the decay index, Indicates the first From vertex to the first Euclidean distance between adjacent vertices Indicates the first Quality scores of adjacent vertices Indicates control factor. This represents the set of grid cells containing adjacent vertices. This indicates the number of grid cells containing adjacent vertices. Represents the grid cells containing adjacent vertices. The quality score is obtained by weighting the values ​​of the corresponding diagnostic indicators.

[0029] Further improvements to the above method involve modeling the simulation primitives in CFD preprocessing software using 3D information. This is achieved by first creating a command stream file in Python to automatically draw the primitives, and then importing it into the CFD preprocessing software.

[0030] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0031] 1. For the two types of constructions that consume the most time in modeling, namely FFU and raised floor, only one geometric surface needs to be generated based on the outline of the FFU and raised floor layout area during modeling. There is no need to model thousands of individual FFUs or raised floors separately. Instead of building thousands of models, only one model needs to be built, which greatly reduces the modeling time and improves the modeling efficiency.

[0032] 2. By introducing virtual nodes to construct virtual regions, the surface and volume meshes of FFUs and raised floors are reconstructed. Multiple quality diagnostic conditions are established to diagnose the quality of the surface and volume meshes, and various optimization methods are used to optimize the surface and volume meshes. This ensures that the information of each FFU and each raised floor is completely mapped to the model, thereby improving the accuracy of modeling.

[0033] 3. By developing a program, CAD element information can be quickly obtained, and modeling can be rapidly performed in the CFD preprocessing model through drawing command flow files. This reduces repetitive work and effectively avoids errors caused by human factors. When the design changes, only the relevant parameters need to be adjusted to quickly generate a new model, thus improving the accuracy and efficiency of modeling.

[0034] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained from what is particularly pointed out in the description and drawings. Attached Figure Description

[0035] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0036] Figure 1 This is a flowchart of a method for simulating and modeling airflow in a cleanroom of an electronics factory, as described in an embodiment of the present invention.

[0037] Figure 2 This is a schematic diagram illustrating the adjustment of external grid nodes by node movement in an embodiment of the present invention;

[0038] Figure 3 This is a schematic diagram illustrating the adjustment of external grid nodes through grid splitting in an embodiment of the present invention. Detailed Implementation

[0039] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0040] A specific embodiment of the present invention discloses a method for simulating and modeling airflow in a cleanroom of an electronics factory, such as... Figure 1 As shown, it includes the following steps:

[0041] S1. Based on the various design drawings of the cleanroom in the electronics factory, obtain the three-dimensional information of the simulation elements.

[0042] It should be noted that, after determining the purpose and requirements of the simulation, the design drawings of the cleanroom of the electronics factory are collected, usually CAD files, including but not limited to: architectural design drawings, cleanroom air conditioning design drawings, process equipment design drawings, as well as architectural elevation or section drawings, and front, side or top views of process equipment.

[0043] Furthermore, simulation elements are determined based on the cleanroom air conditioning design drawings and process equipment design drawings of the electronic factory cleanroom, including but not limited to: various process equipment (such as semiconductor photolithography, etching, thin film, ion implantation, grinding, diffusion, cleaning and other process equipment), transmission equipment (such as automated logistics systems, STK three-dimensional storage warehouses), return air ducts, fan filter units (FFUs) and raised floors.

[0044] Furthermore, from the collected design drawings of the cleanroom in the electronics factory, the three-dimensional information of the simulation primitives is obtained, namely the three-dimensional coordinates of each vertex of the primitive, including:

[0045] The layers of each design drawing are analyzed. Based on the determined simulation elements, the remaining elements are cleared by the reverse selection principle. The remaining elements in each layer are then obtained in turn, and the two-dimensional information of each simulation element is output according to its category.

[0046] Read the building elevation or section drawing, as well as the front view, side view or top view of the process equipment, obtain the height of the corresponding simulation elements, and then convert the two-dimensional information of the simulation elements into three-dimensional information.

[0047] Preferably, a program is written using Visual LISP, Python, VC, or their respective languages ​​to automatically obtain element information from each design drawing.

[0048] S2. Based on the 3D information of the simulation primitives, model the geometric model in the CFD preprocessing software; where all FFUs and all raised floors in the simulation primitives are treated as a whole corresponding to a surface.

[0049] It should be noted that during modeling, the area occupied by all FFUs and all raised floors in the simulation primitives (i.e., the layout area) is calculated separately, and a surface is generated according to the outline of each area. This eliminates the need to model thousands of individual FFUs and each raised floor separately, greatly reducing modeling time.

[0050] To further improve modeling speed and efficiency, this embodiment uses CFD preprocessing software for modeling. First, a command stream file for automatically drawing primitives is compiled using Python, and then imported into CFD (computational fluid dynamics) preprocessing software to complete rapid modeling, such as Icem and Spaceclaim.

[0051] Specifically, the command stream file for drawing primitives includes:

[0052] ① Read the three-dimensional information of the simulation primitives except FFU and raised floor in sequence. In the order of points, lines, surfaces and volumes, generate lines according to adjacent points, obtain lines that are connected to each other and form closed regions to generate surfaces, and generate volumes by stretching or building new surfaces. The volumes include: the clean room of the electronic factory and its internal structural columns, return air ducts and equipment.

[0053] It should be noted that generating a 3D solid through extrusion involves extruding the device along its height (Z direction) based on the device's projection surface in the XY plane, which is identical to its bottom surface. Generating a 3D solid by constructing new surfaces involves combining the device's three views (front view, side view, and top view) to create new surfaces.

[0054] ② Read the three-dimensional information of the simulation primitives corresponding to the FFU and the raised floor, obtain the height and area occupied by the FFU and the raised floor, and generate a surface composed of the outline of the area occupied by each at their respective heights.

[0055] ③ Generate command stream files in the corresponding format according to the type of CFD preprocessing software.

[0056] For example, if the CFD preprocessing software is Pointwise, a command stream file with the suffix .glf is generated; if the CFD preprocessing software is Gmsh, a command stream file with the suffix .geo is generated; and if the CFD preprocessing software is ICEM, a command stream file with the suffix .rpl is generated.

[0057] The command stream file is imported into the CFD preprocessing software, parsed by the built-in script interpreter, and a geometric model of the cleanroom in the electronics factory is generated.

[0058] Furthermore, considering that some devices have different levels of detail in their graphics and different uses, and may include various small connectors, tiny holes, and machined chamfers, the generated geometric models are repaired in two dimensions and three dimensions in the CFD preprocessing software to lay a solid foundation for high-quality simulation.

[0059] Furthermore, the fluid domain is extracted and defined on the repaired geometric model. The extraction of the fluid domain refers to identifying the solid region and dividing the inlet and outlet of the fluid and the flow area.

[0060] S3. Mesh the geometric model to obtain the surface meshes of the FFU and the raised floor; take each vertex of each FFU and each raised floor in the simulation primitive as a virtual node, reconstruct the surface meshes of the FFU and the raised floor, optimize all surface meshes, and then reconstruct and optimize the volume mesh to obtain the airflow simulation calculation input model.

[0061] Different mesh types and sizes are set for different regions based on their impact on the simulation results. The geometric model is then initially meshed. Preferably, the mesh type is triangular or quadrilateral.

[0062] In the geometric model, all FFUs correspond to one face, and all raised floors correspond to one face. After the geometric model is meshed, the face meshes of the FFUs and the face meshes of the raised floors are obtained.

[0063] It should be noted that each vertex of each FFU and each raised floor in the simulation primitives is used as a virtual node to reconstruct the surface mesh of the FFU and the raised floor, including:

[0064] ① Based on the three-dimensional information of the simulation primitives, obtain the two-dimensional coordinates of each vertex of each FFU and each raised floor, and then construct the virtual area of ​​each FFU and each raised floor.

[0065] It should be noted that obtaining two-dimensional coordinates means not considering the height coordinates in the three-dimensional information. The virtual region of each FFU is a rectangular region formed by constructing lines from the adjacent vertices of each FFU in sequence; similarly, the virtual region of each raised floor is a rectangular region formed by constructing lines from the adjacent vertices of each raised floor in sequence.

[0066] ② Based on the positional relationship between the mesh elements in the surface mesh of the FFU and the raised floor and the corresponding virtual regions of each FFU and each raised floor, adjust the mesh nodes of the mesh elements in the surface mesh of the FFU and the raised floor.

[0067] It should be noted that the positional relationship between the grid cells in the surface mesh of an FFU and the virtual region of each FFU is the same as the positional relationship between the grid cells in the surface mesh of a raised floor and the virtual region of each raised floor. Moreover, the grid nodes are updated in the same way. Therefore, FFU and raised floor will no longer be distinguished below and will be collectively referred to as surface mesh, grid cell, and virtual region.

[0068] The positional relationships between grid cells and the virtual region in a surface mesh include: all grid nodes of a grid cell are within the virtual region; some grid nodes of a grid cell are outside the virtual region but within the region's tolerance range; and some grid nodes of a grid cell are outside the virtual region but outside the region's tolerance range. The region's tolerance range is the area formed by expanding outwards from each edge of the virtual region according to a preset tolerance distance.

[0069] Specifically, each virtual region is extracted sequentially as the current virtual region, and the positional relationship between each grid cell in the surface grid and the current virtual region is identified sequentially:

[0070] If some grid nodes of a certain grid cell are outside the current virtual region but within the region tolerance, then obtain the first intersection point between any grid edge where the external grid node is located and the current virtual region, and adjust the position of the external grid node to the first intersection point.

[0071] For example, such as Figure 2As shown, the current virtual region is a solid-line rectangular region composed of vertices 1, 2, 3, and 4. The area between the solid-line rectangular region and the dashed-line rectangular region is the region tolerance range. c, d, and e form a certain grid cell. Point c in this grid cell is outside the current virtual region but within the region tolerance range. Then, the first intersection points of the grid edges cd and ce where c is located with the current virtual region are m and n, respectively. The position of point c is adjusted to m or n.

[0072] If some of the grid nodes of a certain grid cell are outside the current virtual region and outside the region tolerance range, then obtain the two second intersection points between the grid cell and the current virtual region. If the distance between the two second intersection points is less than the minimum size of the grid cell, then the position of the outer grid node will be adjusted to any of the second intersection points. Otherwise, the two second intersection points and the inner grid nodes will form a new grid, and the two second intersection points and the outer grid nodes will form a new grid.

[0073] For example, such as Figure 3 As shown, point c in the grid cell is outside the current virtual region and outside the region tolerance range. The two second intersection points of this grid cell with the current virtual region are m and n. If the distance between m and n is less than the minimum size of the grid cell, then c is adjusted to point m or n. Otherwise, d, e, n and m form a new grid, and m, n and c form a new grid.

[0074] Otherwise, the positions of the grid nodes will not be adjusted, including situations where all grid nodes of a certain grid cell are within the current virtual area, or where all grid nodes of a certain grid cell are outside the current virtual area.

[0075] ③ Identify whether the number of grid nodes inside the virtual region of each FFU and the virtual region of each raised floor is less than the set threshold. If it is less than the set threshold, add the corresponding number of grid nodes according to the difference from the threshold to obtain the reconstructed surface mesh of the FFU and the raised floor.

[0076] It should be noted that if the number of grid nodes within any virtual region is less than the set threshold, the distances between the grid nodes within the virtual region are sorted from largest to smallest. Then, new nodes are inserted between the grid nodes sequentially using linear interpolation, the connection relationships between the grid nodes are updated, and the grid cells are regenerated to obtain the reconstructed surface mesh.

[0077] After the surface meshes of the FFU and the raised floor are reconstructed, all surface meshes in the geometric model are optimized, and then the volume mesh is reconstructed and optimized to obtain the input model for airflow simulation calculation.

[0078] It should be noted that before optimizing all surface meshes, the following steps are also included: establishing quality diagnostic conditions for surface meshes based on diagnostic indicators, and performing quality diagnosis on each surface mesh unit in the geometric model.

[0079] The diagnostic indicators for surface meshes include: aspect ratio of surface mesh cells in non-boundary layers, aspect ratio of surface mesh cells in boundary layers, interior angles of surface mesh cells, parallelism of quadrilateral meshes, skewness of surface mesh cells, and orthogonality of surface mesh cells.

[0080] The quality diagnostic conditions for surface meshes include: the aspect ratio of surface mesh cells in non-boundary layers is less than a first threshold, the aspect ratio of surface mesh cells in boundary layers is within a second threshold range, the interior angles of surface mesh cells are within a third threshold range, the parallelism of quadrilateral meshes is less than a fourth threshold, the skewness of surface mesh cells is less than a fifth threshold, and the orthogonality of surface mesh cells is greater than a sixth threshold.

[0081] It should be noted that the aspect ratio of the face mesh element is calculated by constructing a rectangle inside the mesh element and using the long and short sides of the rectangle. Specifically, for a triangular mesh, any mesh node is selected, and the first line is formed by connecting this mesh node with the midpoint of the corresponding mesh edge. The second line is formed by connecting the midpoints of the other two mesh edges. Then, a rectangle is constructed with these two lines as the center, and the aspect ratio of the mesh is calculated according to formula (1). For a quadrilateral mesh, the midpoints of the four sides of the quadrilateral mesh are connected to form two intersecting lines. A rectangle is constructed with these two lines as the center, passing through the four midpoints. The aspect ratio of the mesh is obtained by dividing the length of the long side of the rectangle by the length of the short side.

[0082] The formula for calculating the aspect ratio of a triangular mesh is shown below:

[0083] (1),

[0084] in, Indicates the aspect ratio of the grid. and These represent the lengths of the long side and the short side of the rectangle, respectively.

[0085] Preferably, the first threshold is set to 5, and the second threshold range is set to [10, 100].

[0086] It should be noted that the larger the interior angle of the face mesh element, the worse the mesh quality. For triangular meshes, the third threshold range is set to [30°, 120°]; for quadrilateral meshes, the third threshold range is set to [45°, 135°].

[0087] Parallelism of a quadrilateral mesh is determined by constructing unit vectors from the mesh edges, calculating the dot product of the relative edge vectors, and then using the inverse cosine of the calculated values ​​to obtain the angle. Higher parallelism results in lower mesh quality; the fourth threshold is 150°.

[0088] The skewness of the surface mesh element is calculated using the normalized equal-angle deviation method. In this embodiment, the fifth threshold is set to 0.7.

[0089] The orthogonality of the surface mesh element is obtained by calculating the cosine of the angle between the mesh surface normal vector and the line connecting the centers of adjacent elements. In this embodiment, the sixth threshold is set to 0.2.

[0090] A surface mesh element is considered to be of acceptable quality when it meets all the quality diagnostic conditions of the surface mesh. A surface mesh element that does not meet any of the quality diagnostic conditions of the surface mesh is considered to be of unacceptable quality.

[0091] Furthermore, each substandard surface mesh element is optimized until all quality diagnostic conditions for the surface mesh are met. Surface mesh optimization methods include: local refinement, adjusting mesh size, Laplacian smoothing, and adjusting mesh type (e.g., converting triangular meshes to quadrilateral meshes).

[0092] When all surface mesh elements meet all the quality diagnostic conditions of the surface mesh, the surface mesh in the geometric model is of qualified quality. After setting parameters such as the number of boundary layers, mesh element type, and mesh growth rate, the geometric model is reconstructed into a volume mesh based on the optimized surface mesh.

[0093] Furthermore, based on the diagnostic indicators of the volume mesh, quality diagnostic conditions for the volume mesh are established, and quality diagnosis is performed on each volume mesh cell.

[0094] It should be noted that the diagnostic indicators for volume meshes include: distortion, volume change rate, Jacobian matrix, orthogonality of volume mesh elements, and aspect ratio of volume mesh elements. Distortion refers to the degree to which the element shape deviates from an ideal shape (such as a regular hexahedron or tetrahedron); volume change rate refers to the degree of abrupt change in volume between adjacent volume mesh elements; the Jacobian matrix is ​​calculated based on the vertex coordinates of the volume mesh elements to reflect the degree of distortion; the orthogonality of volume mesh elements is measured by calculating the angle between the normal vectors of adjacent mesh faces to determine whether the mesh is orthogonal; and the aspect ratio of volume mesh elements is obtained by calculating the ratio of the longest side to the shortest side of the volume mesh element.

[0095] Furthermore, based on the diagnostic indicators of the volume mesh, the quality diagnostic conditions for the volume mesh are established as follows: the distortion is less than the sixth threshold, the volume change rate is less than the seventh threshold, the Jacobian matrix is ​​positive and has no obvious fluctuations, the orthogonality of the volume mesh cells is greater than the eighth threshold, and the aspect ratio of the volume mesh cells is less than the ninth threshold.

[0096] In this embodiment, the sixth threshold is set to 0.7, the seventh threshold is 5, the eighth threshold is 0.2, the ninth threshold is 5 if the volume mesh element is a structured mesh, and the ninth threshold is 20 if the volume mesh element is an unstructured mesh.

[0097] A volume mesh element is considered to be of acceptable quality when it meets all the quality diagnostic conditions of the volume mesh. A volume mesh element that does not meet any of the quality diagnostic conditions of the volume mesh is considered to be of unacceptable quality.

[0098] Furthermore, volume mesh optimization is performed on each substandard volume mesh element, including:

[0099] If the diagnostic condition of distortion being less than the sixth threshold is not met, it indicates that the volume mesh has large distortion. The following methods are used to optimize the volume mesh: homogenize the mesh density in the distorted region, adjust the position of the volume mesh nodes using the Laplacian smoothing method, or adjust the volume mesh type (e.g., convert the tetrahedral mesh to a polyhedral mesh).

[0100] If the diagnostic condition that the volume change rate is less than the seventh threshold is not met, the following methods are used for volume mesh optimization: set a mesh size gradient (e.g., gradually transition from coarse to fine), or insert a buffer layer (e.g., add 2 to 3 transition meshes in the region of abrupt size change).

[0101] If the diagnostic condition that the Jacobian matrix is ​​positive and has no significant fluctuation is not met, the following methods are used for volume mesh optimization: local remeshing (deleting distorted elements, manually refining the mesh in areas with large geometric curvature or narrow gaps), adjusting the position of volume mesh nodes using the Laplacian smoothing method or node offset method, or repairing small gaps or overlapping surfaces in the model.

[0102] If the diagnostic condition that the orthogonality of the volume mesh elements is greater than the eighth threshold is not met, the following methods are used to optimize the volume mesh: generate prism or hexahedral layers with good orthogonality near the wall, adjust the mesh nodes to fit the curvature of the surface, or insert pyramid or prism elements as buffer layers in the unstructured mesh.

[0103] If the diagnostic condition that the aspect ratio of the volume mesh element is less than the ninth threshold is not met, the following methods are used to optimize the volume mesh: add mesh nodes in the long side direction, adjust the volume mesh type, or, for the wall boundary layer mesh, if y+ does not meet the recommended requirements, adjust the height of the first layer mesh; if the interlayer expansion ratio is too large, increase the number of boundary layers.

[0104] When all volume grid elements meet the diagnostic conditions for volume grids, the volume grid quality is qualified. Each volume grid element is then renumbered to obtain the input model for airflow simulation calculation.

[0105] It should be noted that the Laplacian smoothing method is used to reduce twisted elements, improve mesh orthogonality, and homogenize node distribution. Therefore, this method is suitable for optimizing surface and volume mesh elements with substandard quality.

[0106] This embodiment improves the Laplace smoothing method. The improved Laplace smoothing method is obtained by adding weights to the adjacent vertices of the vertex to be optimized in the Laplace smoothing method. That is, the new position of the vertex to be optimized is calculated by the following formula:

[0107] (2),

[0108] in, and They represent the first The positions of the vertices before and after optimization This represents the relaxation factor, used to control the movement step size; Indicates the first The number of adjacent vertices of a vertex. Indicates the first The vertex of the first vertex The positions of the adjacent vertices Indicates the first The vertex of the first vertex The weight of each adjacent vertex is obtained based on the position of the adjacent vertex and the quality score of the grid cell in which the adjacent vertex is located.

[0109] Specifically, the weights of adjacent vertices are calculated using the following formula, so that when the vertex to be optimized moves, it will consider the influence of high-quality adjacent vertices more and reduce the influence of low-quality adjacent vertices:

[0110] (3),

[0111] in, Indicates the decay index, Indicates the first From vertex to the first Euclidean distance between adjacent vertices Indicates the first The quality score of each adjacent vertex is obtained by aggregating the quality scores of the grid cells containing the adjacent vertices, as shown in the following formula:

[0112] (4),

[0113] in, This represents a control factor used to control the weight of the worst-quality grid cell. This represents the set of grid cells containing adjacent vertices. This indicates the number of grid cells containing adjacent vertices. Represents the grid cells containing adjacent vertices. Quality rating.

[0114] It should be noted that the quality score of a grid cell is obtained by weighting the values ​​of the various diagnostic indicators corresponding to the grid cell.

[0115] For surface mesh elements, the quality score of the mesh element is calculated using the following formula:

[0116] (5),

[0117] in, Represents surface cell mesh aspect ratio, Indicates based on surface mesh elements The score for interior angle mapping is obtained by mapping angles that do not meet the interior angle quality diagnosis conditions to values ​​within the range of [0,1]. Represents surface cell mesh The degree of skewness, Represents surface cell mesh orthogonality, , , and These represent the weighting coefficients of the diagnostic indicator values ​​for the corresponding items. .

[0118] If it is a volume mesh element, the quality score of the mesh element is calculated using the following formula:

[0119] (6),

[0120] in, Represents volume element mesh aspect ratio, Represents volume element mesh orthogonality, Represents volume element mesh The degree of distortion, Represents volume element mesh The rate of change of volume, Represents the absolute value of the Jacobian determinant, when When the value is less than 0, the volume mesh element quality is unqualified, and it should be set directly. =0; , , , and These represent the weighting coefficients of the diagnostic indicator values ​​for the corresponding items. .

[0121] Finally, the optimized airflow simulation model is exported and imported into CFD software for airflow simulation.

[0122] Compared with existing technologies, the airflow simulation modeling method for cleanrooms in electronic factories provided in this embodiment addresses the two most time-consuming components: FFUs and raised floors. During modeling, only a single geometric surface needs to be generated based on the outline of the FFU and raised floor area, eliminating the need to model thousands of individual FFUs or raised floors separately. This significantly reduces modeling time and improves efficiency by reducing the number of models from thousands to a single model. Virtual nodes are introduced to construct virtual regions, reconstructing the surface and volume meshes of FFUs and raised floors. Multiple quality diagnostic conditions are established to diagnose the quality of the surface and volume meshes, and various optimization methods are employed to optimize them. This ensures a complete mapping between the information of each FFU and each raised floor and the model, improving modeling accuracy. A program is developed to quickly acquire CAD primitive information, and command flow files are used to quickly model in the CFD preprocessing model, reducing repetitive work and effectively avoiding errors caused by human factors. When design changes occur, only relevant parameters need to be adjusted to quickly generate a new model, further improving modeling accuracy and efficiency.

[0123] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0124] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for simulating and modeling airflow in a cleanroom of an electronics factory, characterized in that, Includes the following steps: Based on various design drawings of cleanrooms in electronic factories, obtain the three-dimensional information of simulation elements; Based on the 3D information of the simulation primitives, a geometric model is generated in the CFD preprocessing software; wherein, all FFUs and all raised floors in the simulation primitives are treated as a whole corresponding to a surface; The geometric model is meshed to obtain the surface meshes of the FFU and the raised floor. Each vertex of each FFU and each raised floor in the simulation primitive is used as a virtual node. After reconstructing the surface meshes of the FFU and the raised floor, all surface meshes are optimized. Then, the volume mesh is reconstructed and optimized to obtain the airflow simulation calculation input model. The step of reconstructing the surface mesh of the FFU and the raised floor by treating each vertex of each FFU and each raised floor in the simulation primitive as a virtual node includes: Based on the three-dimensional information of the simulation primitives, the two-dimensional coordinates of each vertex of each FFU and each raised floor are obtained, and then the virtual area of ​​each FFU and each raised floor is constructed. Based on the positional relationship between the grid cells in the surface mesh and the virtual region, adjust the grid nodes of the grid cells in the surface mesh of the FFU and the raised floor. Specifically, take each virtual region as the current virtual region in turn, and identify the positional relationship between each grid cell in the surface mesh and the current virtual region in turn: If some grid nodes of a certain grid cell are outside the current virtual region but within the region tolerance range, adjust the grid nodes outside by moving the nodes, including: obtaining the first intersection point between any grid edge where the grid node is located and the current virtual region, and adjusting the position of the grid node outside to the first intersection point; If some grid nodes of a certain grid cell are outside the current virtual region and outside the region tolerance range, adjust the grid nodes outside by mesh splitting, including: obtaining two second intersection points between the grid cell and the current virtual region, if the distance between the two second intersection points is less than the minimum size of the grid cell, adjust the position of the grid node outside to any one of the second intersection points; otherwise, form a new mesh by combining the two second intersection points and the internal grid nodes, and form a new mesh by combining the two second intersection points and the external grid nodes. The system identifies whether the number of mesh nodes inside the virtual region of each FFU and the virtual region of each raised floor is less than a set threshold. If it is less than the set threshold, the system adds a corresponding number of mesh nodes based on the difference between the number of mesh nodes and the threshold to obtain the reconstructed surface mesh of the FFU and the raised floor.

2. The airflow simulation modeling method for cleanrooms in electronic factories according to claim 1, characterized in that, Before optimizing all surface meshes, the method further includes: establishing quality diagnostic conditions for surface meshes based on diagnostic indicators of the surface meshes and performing quality diagnostics on each surface mesh cell; Before optimizing the volume mesh, the method further includes: establishing quality diagnosis conditions for the volume mesh based on the diagnostic indicators of the volume mesh, and performing quality diagnosis on each volume mesh cell.

3. The airflow simulation modeling method for cleanrooms in electronic factories according to claim 2, characterized in that, The diagnostic indicators for the surface mesh include: aspect ratio of surface mesh cells in non-boundary layers, aspect ratio of surface mesh cells in boundary layers, interior angles of surface mesh cells, parallelism of quadrilateral meshes, skewness of surface mesh cells, and orthogonality of surface mesh cells; the diagnostic indicators for the volume mesh include: distortion, volume change rate, Jacobian matrix, orthogonality of volume mesh cells, and aspect ratio of volume mesh cells.

4. The airflow simulation modeling method for cleanrooms in electronic factories according to claim 1 or 3, characterized in that, The optimization of all face meshes and the optimization of volume meshes are performed using a modified Laplacian smoothing method; the modified Laplacian smoothing method is obtained by adding weights to the adjacent vertices of the vertex to be optimized in the Laplacian smoothing method.

5. The airflow simulation modeling method for cleanrooms in electronic factories according to claim 4, characterized in that, The improved Laplacian smoothing method calculates the new position of the vertex to be optimized using the following formula: ， in, and They represent the first The positions of the vertices before and after optimization This represents the relaxation factor, used to control the movement step size; Indicates the first The number of adjacent vertices of a vertex. Indicates the first The vertex of the first vertex The positions of adjacent vertices Indicates the first The vertex of the first vertex The weights of the adjacent vertices, Indicates the decay index, Indicates the first From vertex to the first Euclidean distance between adjacent vertices Indicates the first Quality scores of adjacent vertices Indicates control factor. This represents the set of grid cells containing adjacent vertices. This indicates the number of grid cells containing adjacent vertices. Represents the grid cell containing adjacent vertices. The quality score is obtained by weighting the values ​​of the corresponding diagnostic indicators.

6. The airflow simulation modeling method for cleanrooms in electronic factories according to claim 1, characterized in that, The process of modeling in CFD preprocessing software based on the 3D information of the simulated primitives involves first using Python to compile a command stream file for automatically drawing the primitives, and then importing it into the CFD preprocessing software.

Images