A method and system for refined modeling of the micro-geometry of metal foam

Abstract

This invention belongs to the field of microscopic modeling and numerical simulation technology for porous metallic materials, and specifically provides a method and system for refined modeling of the microscopic geometric structure of metal foam. It includes the following steps: S1: Establishing a cuboid computational domain in Cartesian coordinates; S2: Extracting the center coordinates and radii of all spheres in the stacked sphere model as weighted seed points; S3: Generating a Laguerre-Voronoi polyhedral cell network; S4: Constructing a three-dimensional solid skeleton of the metal foam with a set ligament thickness; S5: Obtaining a refined simulation model of the representative volume elements of the metal foam that eliminates boundary effects. This invention achieves extremely high fidelity microstructure reconstruction, significantly improving simulation accuracy at high Reynolds numbers. It can be directly and seamlessly integrated into lattice Boltzmann methods or finite volume methods solvers, significantly improving the robustness and computational efficiency of the simulation.

Description

Technical Field

[0001] This invention belongs to the field of microscopic modeling and numerical simulation technology of porous metal materials, specifically involving a method and system for refined modeling of the microscopic geometric structure of metal foam. Background Technology

[0002] Metal foams, due to their extremely high porosity, large specific surface area, excellent thermal and electrical conductivity, and strong fluid mixing ability, have broad application prospects in compact heat exchangers, fuel cell flow field distribution, microchannel heat dissipation, and catalyst supports. To conduct in-depth research and optimization design of the complex fluid flow and heat transfer characteristics within metal foams, it is essential to accurately reconstruct their complex and tortuous microscopic three-dimensional geometric structure using numerical simulation methods.

[0003] Current conventional research often uses idealized periodic geometric models (such as the Kelvin truncated octahedron model and the Weaire-Phelan model) to replace real metal foam structures. However, real open-cell metal foams typically exhibit irregular characteristics such as varying pore sizes and randomly distributed skeletons. Studies have shown that under low Reynolds number fluid conditions, idealized models can predict pressure drop and forced convection heat transfer relatively well; however, as the flow velocity increases (i.e., under high Reynolds number conditions), the simulation results deviate significantly from experimental data because idealized models lack the tortuosities and surface complexity of real structures.

[0004] To accurately represent the irregularity of materials, the conventional Voronoi diagram (VT) algorithm has been introduced into the modeling of porous media. However, the spatial partitioning boundary of the conventional VT algorithm relies solely on the Euclidean distance between seed points. This is equivalent to assuming that the radii of all pore spheres in the generated metal foam are completely uniform, failing to reflect the normally distributed pore size characteristic prevalent in the fabrication process of real metal foams. Therefore, there is an urgent need for a refined modeling method that can faithfully reproduce the random pore size distribution of metal foams, eliminate boundary effects, and seamlessly integrate with fluid dynamics numerical solvers, thereby overcoming the technical bottleneck of distortion in the simulation of porous media fluid dynamics and heat transfer at high Reynolds numbers. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0006] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution: a method for refined modeling of the microscopic geometry of metal foam, comprising the following steps:

[0007] S1: Establish a cuboid computational domain in Cartesian coordinate system, and generate stacked sphere models that satisfy the target normal aperture distribution and do not overlap within the computational domain using the discrete element algorithm;

[0008] S2: Extract the center coordinates and radii of all spheres in the stacked sphere model as weighted seed points, and remove the sphere geometric entities;

[0009] S3: Based on the extracted weighted seed points, perform Laguerre-Voronoi spatial polyhedron subdivision within the computational domain using the weighted distance function to generate a Laguerre-Voronoi polyhedron cell network.

[0010] S4: Extract the intersecting edges and nodes of the polyhedral cell network, stretch the solid along the intersecting edges, and perform chamfering and smoothing at the intersecting nodes to construct a three-dimensional solid skeleton of metal foam with a set ligament thickness.

[0011] S5: The boundary surface of the cuboid computational domain is used to periodically cut and trim the three-dimensional solid skeleton along the outward normal direction, and the trimmed continuous solid structure is discretized by voxelization to obtain a refined simulation model of the representative volume element of the metal foam that eliminates the boundary effect.

[0012] As a preferred embodiment of the refined modeling method for the microscopic geometric structure of metal foam described in this invention, in step S1, generating a stacked sphere model that satisfies the target normal pore size distribution and is non-overlapping, specifically involves setting the geometric dimensions of the cuboid computational domain as follows: And generate discrete spheres within the domain according to the normal probability density function, wherein the normal probability density function satisfy:

[0013]

[0014] Where r is the radius of the generated sphere; The set target aperture average value; The standard deviation of the target aperture.

[0015] As a preferred embodiment of the refined modeling method for the microscopic geometric structure of metal foam described in this invention, the specific method for performing Laguerre-Voronoi space polyhedron subdivision in step S3 is as follows:

[0016] S3.1: Extract the first step from step two. The coordinates of the center of each sphere Set it as the seed point for partitioning, and square the radius of its corresponding sphere. Set as the weight of this seed point ;

[0017] S3.2: Define any point in the computational domain space With seed point Weighted distance between for:

[0018]

[0019] S3.3: Generate Laguerre-Voronoi cells based on weighted distance The cell The mathematical definition of is the distance from the seed point in space. The weighted distance is no greater than the distance to any other seed point. The set of points whose weighted distance satisfies:

[0020]

[0021] All cells They are seamlessly spliced ​​together in the computational domain without overlapping, forming a Laguerre-Voronoi polyhedral cell network.

[0022] As a preferred embodiment of the refined modeling method for the microscopic geometric structure of metal foam according to the present invention, the specific process of constructing the three-dimensional solid skeleton of metal foam in step S4 is as follows:

[0023] S4.1: Identify and extract the boundary topology information of all Laguerre-Voronoi cells, and obtain the set of intersecting edges and the set of intersecting nodes formed by the intersection of edges;

[0024] S4.2: Using the set cross-sectional shape and geometric dimensions as the outline, and the intersecting edges as the center of the scanning path, perform solid stretching through Boolean addition to generate a connecting rod structure representing the foam ligament; the cross-sectional shape includes a circle or a regular polygon;

[0025] S4.3: At the set of intersecting nodes, a Boolean summation operation is performed on the multiple intersecting connecting rod structures, and a three-dimensional chamfer and surface smoothing fitting is performed on the sharp corner contour at the node connection using a preset fillet radius to simulate the surface tension effect of real metal foam during the preparation process, forming a solid skeleton with smooth transition characteristics.

[0026] As a preferred embodiment of the refined modeling method for the microscopic geometric structure of metal foam described in this invention, the specific process of trimming and voxel discretization in step S5 is as follows:

[0027] S5.1: Using Boolean subtraction, with the six outer surfaces of the cuboid computational domain as cutting planes, the three-dimensional solid skeleton that extends beyond the computational domain is removed and deleted. Within the range, periodic representative volume elements are obtained;

[0028] S5.2: Based on the set discrete mesh resolution, the continuous RVE solid model space is mapped into a three-dimensional discrete voxel matrix. If the center point of the voxel mesh is located inside the solid skeleton, the voxel mesh is assigned solid material properties; otherwise, it is assigned fluid porosity properties. Finally, a mesh file for high Reynolds number fluid dynamics and heat transfer simulation is exported.

[0029] A system for refining the microscopic geometry of metal foam includes:

[0030] The computational domain definition and discrete element stacking module is used to establish a Cartesian coordinate system computational domain and generate a stacked sphere model that satisfies non-overlapping distance constraints and normal aperture distribution.

[0031] The data extraction and transformation module is used to remove the geometric entities of the sphere and extract the center and radius of the sphere to form a weighted seed point;

[0032] The weighted distance polyhedron subdivision module is used to calculate the weighted distance and generate the Laguerre-Voronoi polyhedron network;

[0033] The Boolean stretching and chamfering smoothing module is used to stretch the connecting rods along the edges of a polyhedron and perform surface chamfering and smoothing at the intersection nodes to build a three-dimensional solid skeleton.

[0034] The periodic trimming and voxel mapping module is used to cut the boundaries of the solid skeleton and map it into a three-dimensional discrete voxel matrix, outputting a refined simulation model.

[0035] A computer device includes a processor; and a memory communicatively connected to the processor, the memory storing a computer program, the processor executing the computer program, the computer program, when executed by the processor, implementing the steps of a method for fine modeling the micro-geometry of metal foam.

[0036] A computer-readable storage medium storing a computer program that, when executed by a processor, implements steps for a method of fine-grained modeling of the microstructure of metal foam.

[0037] Compared with the prior art, the beneficial effects of this invention are: (1) High-fidelity microstructure restoration: This invention breaks the limitation of the conventional Voronoi diagram (VT) assuming a single pore size, and cleverly uses the discrete element non-overlapping sphere stacking algorithm to introduce the target normal pore size distribution, and substitutes the square of the sphere radius as a weight into the Laguerre-Voronoi (LVT) weighted distance equation. This mechanism enables the generated model to accurately reflect the gradient difference and randomness of the pore size inside the real metal foam structure, and the fidelity far exceeds that of the traditional Kelvin or Weaire-Phelan ideal model.

[0038] (2) Significantly improves simulation accuracy at high Reynolds numbers: By performing Boolean solid stretching along the edges of the polyhedron and combining it with chamfering and smoothing of the nodes, the smooth transition nodes and complex flow channel tortuosity formed by surface tension during the manufacturing process of real foam materials are highly reproduced. This method greatly reduces the errors generated by the ideal model in the simulation of fluid mechanics and forced convection heat transfer under high flow rate (high Reynolds number) conditions, and provides an accurate geometric basis for the design of high-efficiency heat exchangers.

[0039] (3) Eliminating boundary effects while ensuring efficient solution: The original periodic boundary cutting operation completely eliminates the "wall effects" caused by irregular pores at the edge of the finite computational domain; at the same time, the solid model is discretized by voxelization, avoiding the mesh distortion and negative volume problems that are very easy to occur when complex porous media are meshed in conventional tetrahedral meshing. It can be directly and seamlessly connected to the Lattice Boltzmann Method (LBM) or Finite Volume Method (FVM) solvers, which significantly improves the robustness and computational efficiency of the simulation.

[0040] (4) Excellent parameterized control and expansion capabilities: This method can achieve directional control of the overall porosity and pore density of the model by independently adjusting the normal distribution parameters (mean and variance) and the size of the skeleton stretching section, and has strong modularity and engineering applicability. Attached Figure Description

[0041] To more clearly illustrate the technical solutions of the embodiments of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and detailed embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein:

[0042] Figure 1 This is a logical block diagram illustrating the generation process of the method and system provided in the embodiments of the present invention;

[0043] Figure 2A flowchart illustrating a method for refining the microscopic geometry of metal foam based on Laguerre-Voronoi diagrams, provided in this embodiment of the invention;

[0044] Figure 3 The images are SEM (scanning electron microscope) images of the actual microstructure of metal foam, used as a reference for comparing the fidelity of this invention.

[0045] Figure 4 The figure shows a three-dimensional structural diagram of the generation process of a representative volume element of metal foam in an embodiment of the present invention (Figures (a), (b), and (c) show that the porosity of the model is 0.88 and the pore density is 60 PPI, 90 PPI, and 120 PPI, respectively).

[0046] Figure 5 The following is a comparison of the probability density of the stacked sphere model generated in the embodiments of the present invention with the normal pore size distribution set in the target (Figure (a) shows the pore size analysis results of the model with a porosity of 0.88 and a pore density of 60 PPI; Figure (b) shows the pore size analysis results of the model with a porosity of 0.88 and a pore density of 90 PPI; Figure (c) shows the pore size analysis results of the model with a porosity of 0.88 and a pore density of 120 PPI). Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0048] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0049] like Figure 1 The diagram shown is a logical flowchart illustrating the generation process of a refined modeling system for the microscopic geometric structure of metal foam based on Laguerre-Voronoi diagrams, provided by an embodiment of the present invention. As can be seen from the flowchart, the system comprises three main layers from top to bottom: input parameter configuration, core algorithm processing, and model output. The core algorithm processing layer specifically includes: a computational domain definition and discrete element stacking module, a data extraction and transformation module, a weighted distance polyhedron partitioning module, a Boolean stretching and chamfering smoothing module, and a periodic trimming and voxel mapping module. These modules are connected sequentially, realizing fully automated logical control from macroscopic parameter input (such as porosity, pore size normal distribution, etc.) to microscopic geometric topology network construction, and then to 3D solid stretching and voxelized mesh output.

[0050] like Figure 2The diagram shown is an overall flowchart of a method for refined modeling of the microscopic geometry of metal foam based on Laguerre-Voronoi diagrams, provided by an embodiment of the present invention. This process is executed sequentially by each module of the aforementioned system. Combined with... Figure 4 (Schematic diagram of the generated metal foam structure) Figure 5 (Probability density plot of the generative model) and Figure 3 (SEM image of a real metal foam structure) The specific implementation steps of the method of the present invention are described in detail:

[0051] Step 1: Define the computational domain and generate the stacked sphere model;

[0052] Reference Figure 2 (a) First, a cuboid computational domain is defined in the Cartesian coordinate system. To comprehensively evaluate the fluid flow and subsequent heat transfer characteristics, the physical boundary size of the representative volume element (RVE) is selected in this embodiment. Subsequently, referring to Figure 2 (b) A large number of filled spheres are generated and arranged within the computational domain using the discrete element method. Unlike traditional methods that use uniform spheres, the sphere radii generated in this embodiment follow a preset target normal aperture distribution law. To ensure the rationality of the geometric topology, a strict spatial non-overlapping distance constraint condition is applied during the arrangement process:

[0053] With a target porosity of 0.88 and pore densities of 60 PPI, 90 PPI, and 120 PPI, the input sphere radii for the three models can be calculated using the following formula:

[0054]

[0055] in Let be the pore radius of the metal foam. From the above formula, the average pore radii of the three models are calculated to be 423.3. 282.3 211.67 .

[0056] Fully integrate Figure 5 Further explanation: Figure 5 This shows the probability density plot of the model generated in this step. The horizontal axis represents the diameter of the generated spheres (pores), and the vertical axis represents the probability density. From... Figure 5 As can be clearly seen in (a), (b), and (c), the histogram (representing the statistical distribution of the discrete sphere size actually generated by the discrete element method) and the smooth normal probability density curve (representing the ideal target distribution function) are highly consistent. This fully demonstrates that the stacking algorithm of this invention can accurately control the aperture gradient and randomness of the sphere, overcoming the distortion defects of the conventional uniform aperture model.

[0057] Step 2: Weighted seed point extraction;

[0058] Reference Figure 2 (c) Since the sphere generated in step 1 is only used as a space placeholder for the reserved aperture, the algorithm automatically extracts it in this step. Figure 2 (b) Coordinates of the center of all stacked spheres and the corresponding sphere radius Subsequently, all spherical solid geometry was completely removed from the model, retaining only the extracted 3D spatial point map. These points, carrying radius information, served as "weighted seed points" to guide subsequent non-uniform spatial partitioning.

[0059] Step 3: Laguerre-Voronoi space polyhedron subdivision;

[0060] Square the radius of the sphere extracted in step 2. Assign the corresponding center As weight Based on this, the weighted distances from all points in the global spatial domain to various sub-points are calculated. . Reference Figure 2 (d) Based on this weighted distance criterion, the computational domain is divided into multiple polyhedral cells, generating a Laguerre-Voronoi (LVT) spatial polyhedral cell network. Figure 2 (d) Clearly demonstrates that these polyhedral cell units in Seamlessly joined within the space without overlapping; more importantly, it is precisely because the mathematical determination incorporates radius weighting that... This completely breaks the limitation of conventional VT network cells evolving to the same volume, resulting in cells of varying sizes that perfectly inherit the microscopic topological morphology. Figure 5 The normal porosity distribution law was verified in the experiment.

[0061] Step 4: Extrusion and smoothing of the 3D solid skeleton;

[0062] extract Figure 2 (d) The intersecting edges of all LVT cells are used as the center scan line of the metal foam ligament skeleton. The ligament thickness is set according to the target porosity of 0.88. Boolean addition solid stretching is performed along the intersecting edges with a specific cross section. The stretching is stopped when the overall porosity reaches the target porosity to generate the initial linkage structure.

[0063] Fully integrate Figure 3 Further explanation: Figure 3 This is a scanning electron microscope (SEM) image of the actual microstructure of metal foam. (Observation) Figure 3It can be clearly observed that in the manufacturing process of real metal foam (such as melt foaming or deposition), due to the surface tension of the molten metal, the joints where the struts meet exhibit very rounded and smooth curved transitions, without any absolutely sharp geometric angles.

[0064] To reproduce with extremely high fidelity Figure 3 This physical microscopic feature is addressed in this step, which, based on the tensioning of the connecting rod, specifically involves three-dimensional chamfering and surface smoothing fitting operations at the nodes where multiple connecting rods intersect. (Refer to...) Figure 2 (e) The morphology of the nodes of the metal foam three-dimensional solid skeleton generated after smoothing is similar to... Figure 3 The SEM images are highly consistent, which greatly eliminates the calculation errors of pseudo-turbulence and eddy current caused by non-physical sharp corners in high Reynolds number fluid simulation.

[0065] Step 5: Periodic pruning and voxelized discrete output;

[0066] Reference Figure 2 (f), Figure 2 (e) The generated solid skeleton contains truncated, incomplete pores near the cuboid boundary. To eliminate "wall effects," the six outer surfaces of the cuboid's computational domain are used as cutting surfaces to remove pores extending beyond the boundary. In the range portion, periodic representative volume element (RVE) solid structures are obtained.

[0067] Subsequently, to facilitate numerical calculations, the clipped geometric entities underwent voxelization spatial mapping. From Figure 2 (f) shows that the continuous curved surface skeleton is transformed into a discrete mesh matrix composed of tiny cubes. Figure 4 The three models shown are the metal foam models generated in this embodiment, with a target porosity of 0.88 and pore densities of 60 PPI, 90 PPI, and 120 PPI, respectively. In this embodiment, the voxel resolution is set to... This means that the physical size of each voxel element is precisely 0.013 mm. This voxelization operation effectively avoids the mesh distortion and negative volume problems that easily occur when complex porous surfaces are meshed using conventional tetrahedral meshes. The final output model file can be directly imported into the Lattice Boltzmann Method (LBM) or Computational Fluid Dynamics (CFD) solvers for efficient and accurate simulation verification of multiphase flow and convective heat transfer.

[0068] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for refined modeling of the microscopic geometry of metal foam, characterized in that, Includes the following steps: S1: Establish a cuboid computational domain in Cartesian coordinate system, and generate stacked sphere models that satisfy the target normal aperture distribution and do not overlap within the computational domain using the discrete element algorithm; S2: Extract the center coordinates and radii of all spheres in the stacked sphere model as weighted seed points, and remove the sphere geometric entities; S3: Based on the extracted weighted seed points, perform Laguerre-Voronoi spatial polyhedron subdivision within the computational domain using the weighted distance function to generate a Laguerre-Voronoi polyhedron cell network. S4: Extract the intersecting edges and nodes of the polyhedral cell network, stretch the solid along the intersecting edges, and perform chamfering and smoothing at the intersecting nodes to construct a three-dimensional solid skeleton of metal foam with a set ligament thickness. S5: The boundary surface of the cuboid computational domain is used to periodically cut and trim the three-dimensional solid skeleton along the outward normal direction, and the trimmed continuous solid structure is discretized by voxelization to obtain a refined simulation model of the representative volume element of the metal foam that eliminates the boundary effect.

2. The method for refined modeling of the microscopic geometry of metal foam according to claim 1, characterized in that, In step S1, a stacked sphere model that satisfies the target normal aperture distribution and is non-overlapping is generated. Specifically, the geometric dimensions of the cuboid computational domain are set as follows: And generate discrete spheres within the domain according to the normal probability density function, wherein the normal probability density function satisfy: Where r is the radius of the generated sphere; The set target aperture average value; The standard deviation of the target aperture.

3. The method for refined modeling of the microscopic geometry of metal foam according to claim 1, characterized in that, In S3, the specific method for performing Laguerre-Voronoi space polyhedron subdivision is as follows: S3.1: Extract the first step from step two. The coordinates of the center of each sphere Set it as the seed point for partitioning, and square the radius of its corresponding sphere. Set as the weight of this seed point ; S3.2: Define any point in the computational domain space With seed point Weighted distance between for: S3.3: Generate Laguerre-Voronoi cells based on weighted distance The cell The mathematical definition of is the distance from the seed point in space. The weighted distance is no greater than the distance to any other seed point. The set of points whose weighted distance satisfies: All cells They are seamlessly spliced ​​together in the computational domain without overlapping, forming a Laguerre-Voronoi polyhedral cell network.

4. The method for refined modeling of the microscopic geometry of metal foam according to claim 1, characterized in that, In step S4, the specific process of constructing the three-dimensional solid skeleton of the metal foam is as follows: S4.1: Identify and extract the boundary topology information of all Laguerre-Voronoi cells, and obtain the set of intersecting edges and the set of intersecting nodes formed by the intersection of edges; S4.2: Using the set cross-sectional shape and geometric dimensions as the outline, and the intersecting edges as the center of the scanning path, perform solid stretching through Boolean addition to generate a connecting rod structure representing the foam ligament; the cross-sectional shape includes a circle or a regular polygon; S4.3: At the set of intersecting nodes, a Boolean summation operation is performed on the multiple intersecting connecting rod structures, and a three-dimensional chamfer and surface smoothing fitting is performed on the sharp corner contour at the node connection using a preset fillet radius to simulate the surface tension effect of real metal foam during the preparation process, forming a solid skeleton with smooth transition characteristics.

5. The method for refined modeling of the microscopic geometry of metal foam according to claim 1, characterized in that, In S5, the specific process of clipping and voxel discretization is as follows: S5.1: Using Boolean subtraction, with the six outer surfaces of the cuboid computational domain as cutting planes, the three-dimensional solid skeleton that extends beyond the computational domain is removed and deleted. Within the range, periodic representative volume elements are obtained; S5.2: Based on the set discrete mesh resolution, the continuous RVE solid model space is mapped into a three-dimensional discrete voxel matrix. If the center point of the voxel mesh is located inside the solid skeleton, the voxel mesh is assigned solid material properties; otherwise, it is assigned fluid porosity properties. Finally, a mesh file for high Reynolds number fluid dynamics and heat transfer simulation is exported.

6. A system for fine-grained modeling of the micro-geometric structure of metal foam, used to implement the method for fine-grained modeling of the micro-geometric structure of metal foam as described in any one of claims 1-5, characterized in that, include: The computational domain definition and discrete element stacking module is used to establish a Cartesian coordinate system computational domain and generate a stacked sphere model that satisfies non-overlapping distance constraints and normal aperture distribution. The data extraction and transformation module is used to remove the geometric entities of the sphere and extract the center and radius of the sphere to form a weighted seed point; The weighted distance polyhedron subdivision module is used to calculate the weighted distance and generate the Laguerre-Voronoi polyhedron network; The Boolean stretching and chamfering smoothing module is used to stretch the connecting rods along the edges of a polyhedron and perform surface chamfering and smoothing at the intersection nodes to build a three-dimensional solid skeleton. The periodic trimming and voxel mapping module is used to cut the boundaries of the solid skeleton and map it into a three-dimensional discrete voxel matrix, outputting a refined simulation model.

7. A computer device, characterized in that, The device includes a processor; and a memory communicatively connected to the processor, the memory storing a computer program, the processor executing the computer program, which, when executed by the processor, implements the steps of the method for fine modeling the micro-geometry of metal foam as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for refining the micro-geometry of metal foam as described in any one of claims 1 to 5.

Images