A 3D printing-based micro-channel heat sink design method, system, electronic device and medium

Abstract

The application provides a 3D printing-based micro-channel heat sink design method, comprising: constructing a conjugate heat transfer topology optimization model, determining a design domain, boundary conditions and thermal loads, the boundary conditions at least including an inlet and an outlet, and taking material pseudo-density representing fluid-solid distribution as a design variable, and setting a target function with a fluid flow dissipation and heat sink temperature difference minimization as an optimization target; in the design domain, a variable density topology optimization method is used to iteratively solve the model, and a three-dimensional printing manufacturing constraint is applied in the optimization process to obtain an initial pseudo-density distribution field; the initial pseudo-density distribution field is post-processed with connectivity guidance and flow channel topology shape approximation to obtain a flow channel topology structure with global connectivity and meeting the printable requirements; and a three-dimensional entity model is reconstructed based on the flow channel topology structure to generate a model file for three-dimensional printing. The application effectively solves the printable defects such as flow channel disconnection, closed cavity and missing powder discharge channel in the prior art.

Description

Technical Field

[0001] This application relates to the field of semiconductor heat dissipation device manufacturing technology, specifically to a microchannel heat sink design method, system, electronic device, and medium based on 3D printing. Background Technology

[0002] With the continuous increase in the integration and heat flux density of high-power electronic devices, lasers, and radar, microchannel heat sinks with complex internal flow network have become key components for achieving efficient heat dissipation. Metal additive manufacturing technologies such as laser powder bed melting provide the process feasibility for forming flow channels with complex topologies, making topology-optimized heat sink design methods a research hotspot.

[0003] However, the pseudo-density field directly obtained by existing topology optimization generally contains isolated fluid domains and disconnected non-connected channels. When extracting isosurfaces from these to construct flow channel entities, it is difficult to avoid printability defects such as flow channel interruptions, closed cavities, and missing powder discharge channels. Furthermore, subsequent geometric reconstruction relying on manual judgment will destroy the original fluid-structure interaction optimization characteristics during the adjustment process, causing the heat transfer and flow performance of the final structure to deviate from the design expectations. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, this application provides a microchannel heat sink design method, system, electronic device, and medium based on 3D printing, as detailed below:

[0005] In a first aspect, this application provides a microchannel heat sink design method based on 3D printing, including: A conjugate heat transfer topology optimization model is constructed, which includes a defined design domain, boundary conditions, and thermal loads. Within the framework of the model, the pseudo-density of the material representing the fluid-solid distribution is used as the design variable, and the objective function is set with the minimization of fluid flow dissipation and heat sink temperature difference as the optimization objectives. Within the design domain, the model is iteratively solved using a variable density topology optimization method, and 3D printing manufacturing constraints are applied during the optimization process to obtain an initial pseudo-density distribution field. The initial pseudo-density distribution field is subjected to connectivity-guided post-processing and flow channel topology conformal approximation to obtain a flow channel topology structure with global connectivity and printability requirements. A three-dimensional solid model is reconstructed based on the flow channel topology, generating a model file for 3D printing.

[0006] In one embodiment, the post-processing of the initial pseudo-density distribution field with connectivity guidance and the conformal approximation of the channel topology to obtain a channel topology structure with global connectivity and meeting printability requirements includes: The initial fluid region is extracted from the initial pseudo-density distribution field by threshold segmentation, and the symbolic distance field of the initial fluid region is constructed. Extract the flow channel skeleton from the initial fluid region, and perform connectivity analysis on the flow channel skeleton to identify regions that cannot form a complete connection with the inlet and the outlet as void islands; Using connectivity as the driving force for geometric evolution, the symbolic distance field is subjected to horizontal set evolution, and the void islands are gradually incorporated into the connected flow channels through region growth to form a connected flow channel morphology. The connected flow channel skeleton is subjected to conformal extrusion thickening and smoothing treatment to obtain a flow channel topology surface with a smooth surface that meets printability requirements; wherein, the printability requirements include at least minimum feature size constraints and overhang angle constraints.

[0007] In one embodiment, a diffusion-reaction equation is used to perform a level set evolution on the symbolic distance field; wherein the diffusion term of the diffusion-reaction equation includes an anisotropic diffusion tensor, such that diffusion is enhanced along the direction of the flow channel skeleton and weakened perpendicular to the direction of the flow channel skeleton; the reaction term of the diffusion-reaction equation is defined on the computational grid corresponding to the connection path between the void islands and the connected flow channels, so as to drive the level set function to evolve in the direction of eliminating the void islands; the evolution process is iterative until all the void islands are connected.

[0008] In one embodiment, the connected flow channel skeleton undergoes conformal extrusion thickening and smoothing treatment to obtain a flow channel topology surface with a smooth surface that meets printable requirements; wherein, the printable requirements include at least minimum feature size constraints and overhang angle constraints, including: Based on the connected flow channel skeleton, the target flow channel width distribution is determined according to the distribution of the initial pseudo-density distribution field on the normal cross section at each position of the flow channel skeleton; A variational problem is constructed to minimize the deviation between the actual width determined by the flow channel cross-sectional profile and the target width distribution, while satisfying the minimum radius of curvature and overhang angle constraints, and a smoothing term is introduced to maintain surface continuity. Solving the variational problem yields the cross-sectional profiles at each of the flow channel skeleton locations, and fitting the cross-sectional profiles into a continuous, smooth, non-uniform rational B-spline surface, which serves as the flow channel topological surface.

[0009] In one embodiment, applying 3D printing manufacturing constraints during the optimization process includes: By performing density filtering and projection operations on the pseudo-density field defined by the pseudo-density of the material, the optimized channel width is guaranteed to meet the preset minimum feature size requirement, thereby achieving the minimum feature size constraint. The local overhang angle is determined by calculating the gradient direction of the pseudo-density field. When the local overhang angle is detected to be less than a preset threshold, a penalty term is added to the objective function to suppress the overhang structure that needs to be supported, thereby achieving overhang angle constraint.

[0010] In one embodiment, the method further includes printability verification, the printability verification including: After obtaining the initial pseudo-density distribution field, the connectivity and overhang angle characteristics of the corresponding fluid region are initially evaluated. If the evaluation is unsatisfactory, the manufacturing constraints are adjusted or the topology optimization is returned for re-iteration. And / or, after generating the model file for 3D printing, the overhang angle and minimum feature size of the model slices are detected. If the detection fails, the parameters of the connectivity-guided post-processing and the flow channel topology conformal approximation are adjusted and the model is regenerated until the printable conditions are met.

[0011] In one embodiment, the design domain is a three-dimensional geometric region determined according to the application scenario, the inlet and the outlet are located on opposite sides or on the same side, the inlet is set to a constant flow rate or a constant pressure drop, the outlet is set to a pressure outlet, and the heat load is applied to the designated heat dissipation surface of the design domain in the form of a volume heat source or a surface heat source. And / or, the objective function is set by combining the fluid flow dissipation term and the heat sink temperature difference term through a weighted summation method, wherein the value of the material pseudo density ranges from zero to one, where zero represents the solid region and one represents the fluid region.

[0012] Secondly, this application provides a 3D-printed microchannel heat sink design system, comprising: The preprocessing and modeling module is used to construct a conjugate heat transfer topology optimization model, which includes a defined design domain, boundary conditions and thermal loads. Under the model framework, the pseudo density of the material representing the fluid-solid distribution is used as the design variable, and the objective function is set with the minimization of fluid flow dissipation and heat sink temperature difference as the optimization objectives. The topology optimization solution module is used to iteratively solve the model within the design domain using a variable density topology optimization method, and to apply 3D printing manufacturing constraints during the optimization process to obtain an initial pseudo-density distribution field. The connectivity conformal approximation module is used to perform connectivity-guided post-processing and flow channel topology conformal approximation on the initial pseudo-density distribution field to obtain a flow channel topology structure with global connectivity and meeting printability requirements. The model reconstruction output module is used to reconstruct a three-dimensional solid model based on the flow channel topology and generate a model file for three-dimensional printing.

[0013] Thirdly, this application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described above.

[0014] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described above.

[0015] This application has at least the following beneficial effects: This application provides a 3D printing-based microchannel heat sink design method, system, electronic device, and medium, which effectively solves the printability defects commonly found in existing topology optimization designs, such as unconnected flow channels, closed cavities, and missing powder discharge channels. At the same time, it fundamentally avoids the destruction of thermal-fluid coupling optimization characteristics by artificial geometric reconstruction, so that the final microchannel heat sink is highly close to the design goals in terms of manufacturing feasibility and heat transfer and flow performance.

[0016] This application employs post-processing guided by connectivity of the initial pseudo-density distribution field and conformal approximation of the channel topology. This method automatically identifies and eliminates isolated fluid domains and disconnected non-connected channels, ensuring that the extracted channel topology possesses global connectivity and naturally integrates into structures such as powder discharge channels necessary for additive manufacturing. This process significantly reduces powder residue and printing failures caused by channel interruptions or closed cavities, and significantly improves the additive manufacturing quality and yield of microchannel heat sinks.

[0017] While achieving connectivity restoration, this application employs a topology-preserving approximation strategy to retain the fluid-structure interaction interface shape and flow heat transfer characteristics obtained from topology optimization to the greatest extent possible. This overcomes the problem in existing technologies where reliance on manual judgment and geometric reconstruction leads to severe damage to the optimized structure. The reconstructed heat sink entity can precisely maintain the dual optimization objectives of high heat transfer capacity and low fluid flow dissipation, ensuring the reliable transfer of design performance to manufacturing results.

[0018] Furthermore, this application directly introduces 3D printing manufacturing constraints during the variable density topology optimization iteration process, ensuring that the pseudo-density distribution field meets process requirements such as overhang angle and minimum feature size from the initial generation stage. This pre-constraint and post-processing repair form a dual-insurance mechanism, significantly reducing the pressure of post-processing corrections and improving the manufacturability of the design. The entire technical solution integrates an automated process from topology optimization and connectivity post-processing to 3D solid reconstruction and print file generation, eliminating the need for experience-based geometric reconstruction. This not only eliminates the designer's subjective bias but also rapidly generates high-performance microfluidic networks directly adaptable to metal additive manufacturing processes such as laser powder bed melting, significantly shortening the R&D cycle and improving design efficiency and consistency. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 The flowchart of a microchannel heat sink design method based on 3D printing provided in Embodiment 1 Figure 1 ; Figure 2 The flowchart of a microchannel heat sink design method based on 3D printing provided in Embodiment 1 Figure 2 ; Figure 3 The flowchart of a microchannel heat sink design method based on 3D printing provided in Embodiment 1 Figure 3 ; Figure 4 This is a schematic diagram of a microchannel heat sink design system based on 3D printing provided in Embodiment 2.

[0021] Figure label: 1-Preprocessing and modeling module; 2-Topology optimization solution module; 3-Connectivity conformal approximation module; 4-Model reconstruction output module. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0023] In the description of this application, it should be noted that the terms "vertical", "up", "down", "horizontal", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0024] In the description of this application, it should also be noted that, unless otherwise expressly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0025] Example 1 The pseudo-density field directly obtained by existing topology optimization generally contains isolated fluid domains and disconnected non-connected channels. When extracting isosurfaces from these to construct flow channel entities, it is difficult to avoid printability defects such as flow channel interruptions, closed cavities, and missing powder discharge channels. Furthermore, subsequent geometric reconstruction relying on manual judgment will destroy the original fluid-structure interaction optimization characteristics during the adjustment process, causing the heat transfer and flow performance of the final structure to deviate from the design expectations.

[0026] Based on this, this embodiment provides a 3D printing-based microchannel heat sink design method, such as... Figure 1 As shown, the method includes the following steps: Step 1: Construct a conjugate heat transfer topology optimization model, determine the design domain, boundary conditions and heat load. The boundary conditions include at least the inlet and outlet. The pseudo density of the material representing the fluid-solid distribution is used as the design variable, and the objective function is set with the minimization of fluid flow dissipation and heat sink temperature difference as the optimization objectives, thereby obtaining the conjugate heat transfer topology optimization model. Step 2: Within the design domain, the model is iteratively solved using the variable density topology optimization method, and 3D printing manufacturing constraints are applied during the optimization process to obtain the initial pseudo density distribution field. Step 3: Perform connectivity-guided post-processing and conformal approximation of the initial pseudo-density distribution field to obtain a flow channel topology structure with global connectivity and printability requirements. Step 4: Reconstruct the 3D solid model based on the flow channel topology to generate a model file for 3D printing.

[0027] This embodiment effectively solves the printability defects commonly found in existing topology optimization designs, such as unconnected flow channels, closed cavities, and missing powder discharge channels. At the same time, it fundamentally avoids the destruction of thermal-fluid coupling optimization characteristics by artificial geometric reconstruction, so that the final microchannel heat sink is highly close to the design goals in terms of manufacturing feasibility and heat transfer and flow performance.

[0028] This embodiment employs connectivity-guided post-processing and conformal approximation of the flow channel topology to the initial pseudo-density distribution field. This method automatically identifies and eliminates isolated fluid domains and disconnected non-connected channels, ensuring that the extracted flow channel topology possesses global connectivity and naturally integrates into structures such as powder discharge channels necessary for additive manufacturing. This process significantly reduces powder residue and printing failures caused by flow channel interruptions or closed cavities, and significantly improves the additive manufacturing forming quality and yield of microchannel heat sinks.

[0029] This embodiment, while restoring connectivity, employs a topology-preserving approximation strategy to retain the fluid-structure interaction interface shape and flow heat transfer characteristics obtained from topology optimization to the greatest extent possible. This overcomes the problem in existing technologies where reliance on manual judgment and geometric reconstruction leads to severe damage to the optimized structure. The reconstructed heat sink entity thus accurately maintains the dual optimization objectives of high heat transfer capacity and low fluid flow dissipation, ensuring a reliable transfer of design performance to manufacturing results.

[0030] Furthermore, this embodiment directly introduces 3D printing manufacturing constraints during the variable density topology optimization iteration process, ensuring that the pseudo-density distribution field meets process requirements such as overhang angle and minimum feature size from the initial generation stage. This pre-constraint and post-processing repair form a double-insurance mechanism, significantly reducing the pressure of post-processing corrections and improving the manufacturability of the design. The entire technical solution integrates an automated process from topology optimization and connectivity post-processing to 3D solid reconstruction and print file generation, eliminating the need for experience-based geometric reconstruction. This not only eliminates the designer's subjective bias but also rapidly generates high-performance microfluidic networks directly adaptable to metal additive manufacturing processes such as laser powder bed melting, significantly shortening the R&D cycle and improving design efficiency and consistency.

[0031] In one embodiment, the design domain in step S1 is a three-dimensional geometric region determined according to the application scenario. For example, the design domain is a cuboid with dimensions of 50mm × 50mm × 10mm (length × width × height).

[0032] In one embodiment, the inlet and outlet in step S1 are located on opposite sides (or on the same side, depending on the actual layout). The inlet is set to a constant flow rate (Reynolds number 100-1000) or a constant pressure drop, covering the mainstream heat transfer conditions from laminar to transitional flow; the outlet is set to a pressure outlet (relative pressure 0 Pa), ensuring the physical rationality and numerical stability of the flow boundary conditions.

[0033] In one embodiment, the heat load in step S1 is applied to the bottom surface of the design domain (or the designated chip attachment surface) in the form of a volume heat source or a surface heat source, with a heat flux density of 50–500 W / cm², which can meet the heat dissipation requirements of different application scenarios, from consumer electronics to high-power lasers. The refrigerant wall surface (i.e., the solid area in contact with the flow channel) is set as a non-slip wall surface and coupled heat transfer boundary, and the initial temperature field is set to the ambient temperature (e.g., 25°C).

[0034] Among them, the material pseudo density , where 0 represents solid (structural material) and 1 represents fluid (flow channel).

[0035] Step S1, "using the pseudo-density of the material representing the fluid-structure distribution as the design variable and minimizing fluid flow dissipation and heat sink temperature difference as the optimization objectives," includes setting the objective function by using the pseudo-density of the material representing the fluid-structure distribution as the design variable and combining the fluid flow dissipation term and the heat sink temperature difference term through a weighted summation method. The objective function has the following form: .

[0036] in: For fluid flow dissipation terms, Specifically, The design domain is the entire three-dimensional spatial region to be optimized, including the solid structure ( =0) and fluid flow channel ( =1) Two parts; u is the velocity vector field; For fluid dynamic viscosity; This is a reverse osmosis penalty term that depends on pseudo-density; This is the difference between the average temperature of the bottom surface of the heat sink and the temperature of the inlet fluid. The dissipation value is the initial design (homogeneous solid). The temperature difference under the initial design is used for normalization; Weighting coefficient =0.4, =0.6, adjusted according to the emphasis on heat dissipation and voltage drop requirements. =0.3~0.5, Adjust within the range of 0.5 to 0.7.

[0037] This embodiment establishes the relationship between pseudo-density and physical properties using the solid isotropic material penalty (SIMP) interpolation method: Thermal conductivity interpolation: Let the penalty factor be p=3; Reverse osmosis rate penalty: = Let the penalty factor q = 2. Wherein, Value 10 4 ~10 6 (Adjusted according to the characteristic dimensions of the design domain), this expression makes it possible in the fluid region ( When =1), ≈0, in the solid region ( When =0), ≈ It can smoothly and effectively force the flow velocity in the solid region to approach zero without explicitly treating the fluid-solid interface, thus greatly reducing the risk of numerical oscillations and iterative divergence.

[0038] In one embodiment, a hexahedral structured mesh is used for mesh generation, and the mesh size must be able to resolve the minimum flow channel feature (diameter ≥ 0.3 mm). In this embodiment, the global mesh size is 0.05–0.2 mm, and the total number of meshes is controlled within 1 million–5 million, ensuring a good match between the mesh size and the minimum feature size of 3D printing (≥ 0.3 mm). This avoids flow channel boundary distortion caused by an overly coarse mesh and prevents computational redundancy caused by an overly dense mesh, allowing the subsequent topology optimization results to be directly used for additive manufacturing.

[0039] In one embodiment, step S2, "iteratively solving the model within the design domain using a variable density topology optimization method," includes: using a solid isotropic material penalty (SIMP) framework combined with the moving asymptote method (MMA) or the optimality criterion (OC) to determine the design variables. Update. After each iteration, the physics field must be resolved. The penalty factors p and q adopt a continuously increasing strategy, starting from an initial value of 1 and increasing every 20 iterations, gradually increasing to a target value of 3 or 4. This effectively avoids local optimum traps caused by excessive penalty in the early stages of topology optimization, while ensuring clear channel-solid boundaries and a significant reduction in the proportion of grayscale units in later iterations. The convergence criterion is set as: the rate of change of design variables (…). The objective function must have a relative change of less than 0.001 and a relative change of less than 0.1%. Both conditions must be met consecutively for 10 iterations to be considered convergent. This eliminates false convergence caused by local oscillations, ensuring the stability and repeatability of the obtained pseudo-density distribution field, and providing high-quality input for subsequent connectivity-guided post-processing.

[0040] In one embodiment, step S2, “applying 3D printing manufacturing constraints during the optimization process”, includes minimum feature size constraints and overhang angle constraints.

[0041] The minimum feature size constraint is achieved by performing density filtering and projection operations on the pseudo-density field defined by the material's pseudo-density to ensure that the optimized channel width meets the preset minimum feature size requirement. Specifically, an open / closed operator-type filter from morphological image processing is used. This filter is based on the Helmholtz partial differential equation and has a filter radius of 1.5 times the grid size. This allows it to be embedded into the gradient optimization framework in a continuous and differentiable manner, avoiding the loss of sensitivity consistency caused by discontinuous morphological filtering. After each iteration, the current pseudo-density field is sequentially subjected to density filtering and projection processing. For the laser powder bed fusion process, the minimum channel width threshold is set to 0.3 mm, directly matching the process limits of mainstream laser powder bed fusion (LPBF) equipment. This ensures that the optimization result meets the geometric reachability for printing without manual correction, significantly reducing the risk of subsequent clogging or printing failure due to excessively narrow channels.

[0042] The overhang angle constraint determines the local overhang angle by calculating the gradient direction of the pseudo-density field. When a local overhang angle is detected to be less than a preset threshold, a penalty term is added to the objective function to suppress the overhanging structure that needs support. Specifically, the overhang angle is defined as the angle between the normal to the lower surface of the flow channel and the construction direction (usually taken as the +z axis). The overhang angle of each local region is calculated based on the gradient direction of the pseudo-density field. When detected If the angle is less than a preset threshold (e.g., 45°), it indicates that the area requires additional support during printing; therefore, a penalty term is introduced into the objective function.

[0043] in, For local overhang angle, ( ) is the Heaviside step function. This is a penalty weighting coefficient, ranging from 0.1 to 0.5. Designers can flexibly balance heat dissipation performance and self-support based on the support capabilities of specific printing equipment, avoiding the significant performance degradation caused by imposing hard constraints (such as forcibly deleting all cells smaller than a threshold) to eliminate overhangs in traditional methods. This penalty term makes the optimization algorithm tend to avoid generating flow channel top surface structures with excessively large horizontal spans or overly flat lower surfaces during the iteration process, thereby achieving a self-supporting flow channel geometry without relying on supports.

[0044] After the iteration, a set of values ​​with continuous values ​​is obtained at each grid node. The grayscale distribution field, i.e., the initial pseudo-density distribution field. To reduce numerical noise and improve the robustness of subsequent connectivity-guided post-processing (step S3), the grayscale field is subjected to a Gaussian smoothing filter with a standard deviation of 1.2 times the grid size. This effectively smooths numerical noise caused by iterative convergence oscillations, reduces isolated high or low values ​​in the pseudo-density field, and thus improves the robustness of subsequent symbolic distance field construction and skeleton extraction. This filtering operation suppresses the checkerboard effect while preserving the original topological features, providing a cleaner initial geometry for the connectivity-guided region growing algorithm, avoiding false gap / island identification caused by noise, and shortening the post-processing iteration time. The smoothed grayscale field will be used as input for step S3 for symbolic distance field initialization and connectivity analysis.

[0045] In one embodiment, such as Figure 2 As shown, step S3 involves post-processing the initial pseudo-density distribution field with connectivity guidance and conformal approximation of the channel topology to obtain a channel topology structure with global connectivity and printability requirements, including: S31. Extract the initial fluid region from the initial pseudo-density distribution field by threshold segmentation, and construct the symbolic distance field of the initial fluid region.

[0046] In one embodiment, step S21 further includes: reading in the initial pseudo-density distribution field obtained by topology optimization and treating it as a grayscale field. Set threshold (Its empirical value can be taken as 0.5, and can be adjusted between 0.3 and 0.7 according to the actual optimization results.) Define the initial fluid region. Subsequently, a symbolic distance field was constructed. The distance field satisfies the Eikonal equation. =1, and its boundary condition is: if =1, its boundary condition is: ,but ;like ,but ,in Point to the fluid region boundary The shortest Euclidean distance is calculated. The Fast Marching Method (FMM) is used to solve for this distance field on the background grid, and the results are saved as a distance matrix with the same dimensions as the grid.

[0047] S32. Extract the flow channel skeleton from the initial fluid region and perform connectivity analysis on the flow channel skeleton to identify regions that cannot form a complete connection with the inlet and outlet as void islands.

[0048] In one embodiment, step S22 further includes: firstly, based on the symbolic distance field The zero-level set is used to extract the flow channel centerline skeleton using a level set advancement method with geometrically adaptive stepping. Specifically, in Within the initial fluid region defined by <0, a voxelized skeleton point set is generated using methods based on Voronoi diagrams or topology refinement algorithms (such as the "erosion-preservation" morphological operation). Then, the skeleton points are clustered by connectivity, and a breadth-first search (BFS) algorithm is used to identify all skeleton branches connected to the inlet and outlet nodes (at user-specified coordinates). Skeleton branches that are not connected to the inlet or outlet are enclosed by a [missing information - likely a typo]. Regions with values ​​<0 are marked as "void islands". Finally, the volume, centroid coordinates, and shortest distance to the nearest connected channel boundary of each island are recorded as input information for subsequent step S33.

[0049] S33. To achieve the gradual annexation of isolated islands, this invention uses connectivity as the core driving force for geometric evolution, performs level set evolution on the symbolic distance field, and gradually incorporates islands into connected channels through region growing, ultimately forming a globally connected channel morphology. The key to this method lies in using a diffusion-reaction equation to drive the dynamic evolution of the level set function. The specific form of this equation is:

[0050] in, The sign distance field (level set function) has a zero level surface. Indicates the flow channel boundary; diffusion term Used to control the smooth expansion of the flow channel boundary, reaction term It then actively guides the boundary to advance towards the gap and island area.

[0051] In numerical implementation, the symbolic distance field Discretized on the background grid Time step Numerical stability conditions must be met, and typically, a value of 100% is taken as 100%. ,in For grid spacing, In actual calculations The commonly used value is between 0.01 and 0.05.

[0052] Among them, the diffusion term in the diffusion-reaction equation Including an anisotropic diffusion tensor makes the diffusion behavior spatially direction-dependent. Specifically, diffusion is enhanced along the channel skeleton direction to promote the preferential extension of the fluid region along the mainstream path; diffusion is weakened perpendicular to the skeleton direction to suppress excessive increase in channel width. For example, in the skeleton backbone region (i.e., the distance from the channel skeleton is less than twice the mesh spacing)... (position), diffusion coefficient in the parallel direction =1.0, vertical diffusion coefficient =0.1; In non-core regions, the diffusion coefficient in the parallel direction is... and vertical diffusion coefficient Both are 0.5, making the diffusion tend to be isotropic. In actual calculations, through... Construct the diffusion tensor using the gradient direction: ,in This allows the aforementioned anisotropic characteristics to be incorporated into the diffusion-reaction equation.

[0053] In this diffusion-reaction equation, the reaction terms are defined on the computational grid corresponding to the connection paths between void islands and connected channels, driving the level set function to evolve in the direction of eliminating void islands. The evolution process is iterative until all void islands are connected. For example, the reaction terms are expressed as... ,in .in, This is the reaction rate constant, ranging from 0.5 to 2.0 (a smaller value is used when the convergence rate is fast). The target fluid volume fraction (usually set to 0.3–0.5). For the current fluid region ( The volume fraction of <0); For the smoothed Dirac function, take ,in . The connectivity guiding function is constructed as follows: For each gap island, Dijkstra's algorithm is used to calculate the shortest path from the nearest connected boundary node to all nodes within the island on the skeleton graph, and the grid points traversed by this path are marked as follows. =1; if multiple isolated islands share the same path, then The summation is truncated to 1. Thus, the reaction term drives the level set function to evolve in the direction of eliminating gaps and islands. The evolution process is iterative until all islands are connected, resulting in a globally connected flow channel morphology.

[0054] The numerical solution of the diffusion-reaction equations employs the finite difference method to discretize the symbolic distance field on the background grid. The diffusion term uses a central difference scheme, and the reaction term uses an explicit upwind scheme. Time discretization utilizes an alternating direction implicit (ADI) or semi-implicit scheme, i.e. The diffusion term is implicitly handled, while the reaction term is explicitly handled, to ensure numerical stability. The resulting linear equation system is solved using the conjugate gradient method in conjunction with a preprocessor.

[0055] To meet the minimum feature size requirements of 3D printing technology, a penalty potential function is introduced during the evolution process. ,in For the reason The calculated local flow channel width (can be obtained by solving the Laplace equation or by distance transformation) Minimum allowable width (e.g., 0.3mm). =0.1, =0.05mm. This penalty term is added as an additional reaction term to the diffusion-reaction equation, where the local width... Less than the minimum allowable width At that time, penalty items Producing positive values ​​prompts Increase the width by creating locally closed flow channels, thereby avoiding the generation of overly thin, unprintable structures.

[0056] When processing multiple isolated islands, they are grown sequentially according to their volume from smallest to largest. The symbolic distance field is updated after each island is merged. And the skeleton diagram, then process the next one. If the distance between two isolated islands is less than 5 times the grid spacing (i.e., 5... If any islands are found to be isolated, they are merged to avoid competition. The convergence criterion for the evolutionary process is: all isolated islands are absorbed (through connectivity checks of the skeleton graph) and the symbolic distance field is within a certain range for five consecutive iterations. The change in L2 norm (Euclidean norm) is less than 1 × 10 -4 Typically, the entire evolution process converges within 20 to 50 iterations, resulting in a connected flow channel morphology that meets printability requirements.

[0057] S34. Perform conformal extrusion thickening and smoothing on the connected channel skeleton to obtain a channel topology surface with a smooth surface that meets printability requirements. The printability requirements include at least minimum feature size constraints (e.g., minimum channel width not less than 0.3 mm) and overhang angle constraints (e.g., the angle between the channel wall and the construction direction not less than 45°). The core objective of this step is to reconstruct the connected channel skeleton into a three-dimensional channel surface with physical thickness, while ensuring that this surface, under printability requirements such as minimum feature size constraints and overhang angle constraints, approximates as closely as possible to the optimal hydrodynamic geometry implied by the original topology optimization result.

[0058] In one embodiment, such as Figure 3 As shown, step S34 further includes the following sub-steps: Step S341: Based on the connected flow channel skeleton obtained in step S33 and the initial pseudo-density distribution field stored in step S32, extract the pseudo-density distribution on the normal cross-section at each position of the flow channel skeleton to determine the target flow channel width distribution. Specifically, fix the grown skeleton line (including the spatial coordinates of each Voronoi seed point and their interconnections), and for each skeleton branch, extract the original grayscale field on the cross-section perpendicular to the skeleton direction. Sampling is performed to obtain the "target width function" corresponding to the arc length coordinate s along this branch. This function faithfully records the ideal width of the fluid channel assigned at each cross-section by the original optimization result. Because it directly utilizes the grayscale field information generated during the optimization process rather than simple threshold binarization, this step can preserve the geometric features of the original design at local cross-sections (such as expanding, contracting, or non-circular cross-sections) to the greatest extent possible. This provides a high-fidelity geometric basis for subsequent conformal approximation, avoiding the decrease in heat dissipation performance caused by the loss of cross-sectional information in traditional methods.

[0059] Step S342: Construct a variational problem to determine the flow channel profile curve at each cross-section. This flow channel profile curve is in polar coordinates. This indicates that (where s is the arc length coordinate along the skeleton, (where the angle is polar). The goal of the variational problem is to make the reconstructed flow channel profile approximate the target width function as closely as possible while satisfying manufacturability constraints. Specifically, under the constraints of minimum radius of curvature and overhang angle, the following objective functional is minimized:

[0060] in, =0.01 smoothness coefficient along the skeleton direction, =0.001 is the smoothness coefficient in the circumferential direction of the cross section; Depend on This is achieved by combining an assumed initial cross-sectional shape (e.g., circular or elliptical). The three terms of this functional correspond to: the deviation between the actual and target width of the cross-section (form-preserving term), the geometric smoothing term along the flow path (suppressing abrupt width changes in the skeleton direction), and the circumferential smoothness regularization term (avoiding jagged fluctuations in the cross-sectional profile). By incorporating the form-preserving deviation, flow smoothing, and circumferential smoothness into a unified variational framework and explicitly adding printability hard constraints, this step can balance the original geometric fidelity and manufacturing feasibility globally, solving the dilemma in traditional post-processing of "either sacrificing performance for printability or making it unprintable."

[0061] At the same time, the variational problem must simultaneously satisfy the constraints imposed by the manufacturing process: the minimum radius of curvature of the cross-sectional profile is not less than 0.2 mm (to prevent stress concentration and difficulties in printing powder spreading), and the local corner of the flow channel wall is not less than 150° (i.e., to avoid excessively sharp corners and prevent a sharp increase in flow resistance and printing defects).

[0062] Step S343: Solve the above variational problem to obtain the cross-sectional profiles at each flow channel skeleton location, and fit the cross-sectional profiles to a continuous, smooth, non-uniform rational B-spline (NURBS) surface as the smooth flow channel topological surface. Specifically, the finite element method is used to coordinate the arc length s and polar angle. Discretize the data and solve the corresponding linear equations using the Fast March Method (FMM) to obtain the cross-sectional profile curves at each skeleton location. Subsequently, the calculated cross-sectional profile is converted into a triangular mesh representation, and then a third-order non-uniform rational B-spline is used to fit the entire flow channel surface, ultimately generating a continuous and smooth flow channel topology surface. During the fitting process, smoothness parameters are strictly controlled: the allowable geometric deviation does not exceed 0.02 mm, and the rate of change of surface curvature is limited to within 5%.

[0063] This step introduces a fast traversal method to ensure the efficiency of solving variational problems at the million-grid scale (single solution time less than 5 minutes), allowing this post-processing step to iterate rapidly with the front-end topology optimization. NURBS surface fitting not only eliminates the surface step effect caused by discrete meshes but also significantly reduces the contour complexity of 3D printed slices, thereby reducing serrations in the laser scanning path during printing and improving forming quality and surface finish. Simultaneously, the smooth and continuous surface also facilitates the generation of subsequent flow and heat transfer simulation meshes, avoiding numerical instability caused by geometric defects.

[0064] In one embodiment, the 3D-printed microchannel heat sink design method also includes a two-stage printability verification mechanism throughout the design process. This aims to systematically ensure the designed flow channel structure can be actually manufactured from the early stages of optimization iteration to the final model output stage, avoiding repeated trial and error due to manufacturability issues. Unlike the traditional approach of only performing a rough check at the end of topology optimization, this method breaks down printability verification into two levels: the first stage focuses on the initial screening of "macro-feasibility of connectivity and overhang angle" during the optimization process, while the second stage performs "high-precision slice-level" manufacturing verification on the final 3D model. The two stages form a closed-loop feedback, tightly coupling topology optimization, connectivity post-processing, and printability assurance.

[0065] The first stage of verification is executed immediately after obtaining the initial pseudo-density distribution field for topology optimization, i.e., after step S2 but before entering the connectivity-guided post-processing. At this stage, a preliminary fluid region indicator is constructed based on the pseudo-density field (e.g., a set of mesh nodes with γ > 0.5), and two checks are performed: first, a graph connectivity algorithm is used to determine whether the fluid region is fully connected to the inlet and outlet nodes, identifying any voids or islands; second, the local overhang angle distribution on the lower surface of the flow channel is calculated to assess whether there are large areas with overhang angles less than a set threshold (typically 45°). If the preliminary assessment indicates the presence of unacceptable voids or excessive overhang regions, the first stage verification is considered unsuccessful. Instead of immediately proceeding to complex post-processing, the process returns to step S2 to adjust manufacturing constraint parameters—for example, increasing the filtering radius of the minimum feature size (e.g., from 1.5 times the mesh size to 2.0 times) and increasing the weight of the overhang angle penalty term (e.g., from 0.1 to 0.3). Then, the topology optimization iteration is rerun until the first stage verification passes. This allows for the early filtering out of obviously unmanufacturable designs, saving subsequent computational resources.

[0066] The second-stage verification is performed after generating the model file (STL format) for 3D printing, i.e., after step S4 is completed. At this point, the STL model is sliced ​​(typically 30μm thick), and detailed overhang angle and minimum feature size detection are performed on the contour of each slice. The overhang angle detection method is as follows: for the current printing layer, the proportion of the area where the upper layer's contour exceeds the current layer's contour is calculated. If this proportion exceeds 30% or the structural tilt angle is less than 40°, it is marked as an unacceptable overhang area. The minimum feature size detection uses the distance field method to calculate the local thickness of the model. If the proportion of voxels with a thickness less than 0.3mm exceeds 1% of the total voxels, it is judged as a width unacceptable. When unacceptable items are detected in the second-stage detection, the process does not directly return to the topology optimization step. Instead, local adjustments are prioritized within the connectivity-guided post-processing and flow channel topology conformal approximation framework of step S3. Specifically, the regularization parameters in the conformal approximation variational problem are fine-tuned. and (For example, appropriately increasing the smoothness parameter) To enhance smoothness along the flow channel direction, or to reduce the cross-sectional curvature constraint parameters. To relax the curvature constraints of the cross section, or to add local conformal weights to strengthen the geometric constraints of the unqualified regions, the variational problem is then solved again to generate a new model.

[0067] If the model still fails the second-stage verification after at most three fine-tuning iterations, it is determined that there is a fundamental contradiction between the current topology optimization configuration and the manufacturing process. In this case, it is necessary to return to step S3 to adjust the growth parameters (such as changing the reaction rate constant, diffusion anisotropy coefficient ratio, or penalty potential function weight), and re-execute connectivity-guided region growth and approximation until the final model meets all printability requirements. Through the above two-stage closed-loop verification, it is ensured that the final 3D-printed microchannel heat sink retains the hydrodynamic performance of the topology optimization and can be successfully manufactured without repeated manual repairs or sacrifice of design performance.

[0068] Example 2 like Figure 4 As shown, this embodiment provides a 3D-printed microchannel heat sink design system to implement the method described in Embodiment 1, including: Preprocessing and modeling module 1 is used to construct a conjugate heat transfer topology optimization model. The model includes a defined design domain, boundary conditions and thermal loads. Within the model framework, the pseudo density of the material representing the fluid-solid distribution is used as the design variable, and the objective function is set with the minimization of fluid flow dissipation and heat sink temperature difference as the optimization objectives. Topology optimization solution module 2 is used to iteratively solve the model within the design domain using a variable density topology optimization method, and to apply 3D printing manufacturing constraints during the optimization process to obtain the initial pseudo density distribution field; The connectivity conformal approximation module 3 is used for post-processing and flow channel topology conformal approximation of the initial pseudo-density distribution field with connectivity guidance to obtain a flow channel topology structure with global connectivity and printability requirements. Model reconstruction output module 4 is used to reconstruct a 3D solid model based on the flow channel topology and generate a model file for 3D printing.

[0069] Example 3 This embodiment proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the method described in Embodiment 1.

[0070] Example 4 This embodiment proposes a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described in Embodiment 1.

[0071] Those skilled in the art will understand that the modules or steps described above in this application can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0072] Note that the above description is merely a preferred embodiment and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this application, and the scope of this application is determined by the scope of the appended claims.

[0073] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A microchannel heat sink design method based on 3D printing, characterized in that, include: A conjugate heat transfer topology optimization model is constructed to determine the design domain, boundary conditions and thermal load. The boundary conditions include at least the inlet and outlet. The pseudo density of the material, which represents the fluid-solid distribution, is used as the design variable, and the objective function is set with the minimization of fluid flow dissipation and heat sink temperature difference as the optimization objectives. Within the design domain, the model is iteratively solved using a variable density topology optimization method, and 3D printing manufacturing constraints are applied during the optimization process to obtain an initial pseudo-density distribution field. The initial pseudo-density distribution field is subjected to connectivity-guided post-processing and flow channel topology conformal approximation to obtain a flow channel topology structure with global connectivity and printability requirements. A three-dimensional solid model is reconstructed based on the flow channel topology, generating a model file for 3D printing.

2. The microchannel heat sink design method based on 3D printing according to claim 1, characterized in that, The post-processing and conformal approximation of the initial pseudo-density distribution field to achieve connectivity guidance, resulting in a flow channel topology with global connectivity and printability requirements, includes: The initial fluid region is extracted from the initial pseudo-density distribution field by threshold segmentation, and the symbolic distance field of the initial fluid region is constructed. Extract the flow channel skeleton from the initial fluid region, and perform connectivity analysis on the flow channel skeleton to identify regions that cannot form a complete connection with the inlet and the outlet as void islands; Using connectivity as the driving force for geometric evolution, the symbolic distance field is subjected to horizontal set evolution, and the void islands are gradually incorporated into the connected flow channels through region growth to form a connected flow channel morphology. The connected flow channel skeleton is subjected to conformal extrusion thickening and smoothing treatment to obtain a flow channel topology surface with a smooth surface that meets printability requirements; wherein, the printability requirements include at least minimum feature size constraints and overhang angle constraints.

3. The microchannel heat sink design method based on 3D printing according to claim 2, characterized in that, The level set evolution of the symbolic distance field is performed using a diffusion-reaction equation; wherein the diffusion term of the diffusion-reaction equation contains an anisotropic diffusion tensor, and the reaction term of the diffusion-reaction equation is defined on the computational grid corresponding to the connection path between the void islands and the connected channels; the evolution process is iterative until all the void islands are connected.

4. The microchannel heat sink design method based on 3D printing according to claim 2, characterized in that, The connected flow channel skeleton is subjected to conformal extrusion thickening and smoothing treatment to obtain a flow channel topology surface with a smooth surface that meets printability requirements; wherein, the printability requirements include at least minimum feature size constraints and overhang angle constraints, including: Based on the connected flow channel skeleton, the target flow channel width distribution is determined according to the distribution of the initial pseudo-density distribution field on the normal cross section at each position of the flow channel skeleton; A variational problem is constructed to minimize the deviation between the actual width determined by the flow channel cross-sectional profile and the target width distribution, while satisfying the minimum radius of curvature and overhang angle constraints, and a smoothing term is introduced to maintain surface continuity. Solving the variational problem yields the cross-sectional profiles at each of the flow channel skeleton locations, and fitting the cross-sectional profiles into a continuous, smooth, non-uniform rational B-spline surface, which serves as the flow channel topological surface.

5. The microchannel heat sink design method based on 3D printing according to claim 1, characterized in that, The process of applying 3D printing manufacturing constraints during optimization includes: By performing density filtering and projection operations on the pseudo-density field defined by the pseudo-density of the material, the optimized flow channel width is guaranteed to meet the preset minimum feature size requirement, thereby achieving the minimum feature size constraint. The local overhang angle is determined by calculating the gradient direction of the pseudo-density field. When the local overhang angle is detected to be less than a preset threshold, a penalty term is added to the objective function to suppress the overhang structure that needs to be supported, thereby achieving overhang angle constraint.

6. A microchannel heat sink design method based on 3D printing according to any one of claims 1 to 5, characterized in that, The method further includes printability verification, which includes: After obtaining the initial pseudo-density distribution field, the connectivity and overhang angle characteristics of the corresponding fluid region are initially evaluated. If the evaluation is unsatisfactory, the manufacturing constraints are adjusted or the topology optimization is returned to iterate again. And / or, after generating the model file for 3D printing, the overhang angle and minimum feature size of the model slices are detected. If the detection fails, the parameters of the connectivity-guided post-processing and the flow channel topology conformal approximation are adjusted and the model is regenerated until the printable conditions are met.

7. The microchannel heat sink design method based on 3D printing according to claim 1, characterized in that, The design domain is a three-dimensional geometric region determined according to the application scenario. The inlet and the outlet are located on opposite sides or on the same side. The inlet is set to a constant flow rate or a constant pressure drop, and the outlet is set to a pressure outlet. The heat load is applied to the designated heat dissipation surface of the design domain in the form of a volume heat source or a surface heat source. And / or, the objective function is set by combining the fluid flow dissipation term and the heat sink temperature difference term through a weighted summation method, wherein the value of the material pseudo density ranges from zero to one, where zero represents the solid region and one represents the fluid region.

8. A microchannel heat sink design system based on 3D printing, characterized in that, include: The preprocessing and modeling module is used to construct a conjugate heat transfer topology optimization model, which includes a defined design domain, boundary conditions, and thermal loads. Within the aforementioned model framework, the pseudo-density of the material, representing the fluid-solid distribution, is used as the design variable, and the objective function is set with the minimization of fluid flow dissipation and heat sink temperature difference as the optimization objectives. The topology optimization solution module is used to iteratively solve the model within the design domain using a variable density topology optimization method, and to apply 3D printing manufacturing constraints during the optimization process to obtain an initial pseudo-density distribution field. The connectivity conformal approximation module is used to perform connectivity-guided post-processing and flow channel topology conformal approximation on the initial pseudo-density distribution field to obtain a flow channel topology structure with global connectivity and meeting printability requirements. The model reconstruction output module is used to reconstruct a three-dimensional solid model based on the flow channel topology and generate a model file for three-dimensional printing.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program thereon, which, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images