Chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy

Abstract

The application discloses a kind of based on NiTi alloy's chip liquid cooling plate flow channel topological optimization design method, comprising: obtaining chip heat source distribution, determine two-dimensional design domain and establish flow-thermal coupling numerical model;Determine material, environment and boundary parameter;Based on solid isotropic material interpolation format, express physical property parameter as the function of design variable, construct topological optimization mathematical model;With minimum fluid viscosity energy dissipation rate and design domain average temperature as joint target, with fluid volume fraction and temperature interval as constraint, carry out iterative solution, obtain two-dimensional optimal flow channel topological configuration.Wherein, NiTi shape memory alloy unit is arranged on the wall surface of flow channel, adaptive flow regulation based on temperature sensing can be realized.The application realizes the collaborative optimization of heat dissipation performance and flow resistance, can effectively suppress hot spot, and design efficiency is high.

Description

Technical Field

[0001] This invention relates to the field of thermal fluid topology optimization design technology, and more specifically, to a method for optimizing the flow channel topology of a chip liquid cooling plate based on NiTi alloy. Background Technology

[0002] With the rapid development of high-performance computing chips, power semiconductor devices, and artificial intelligence servers, chip integration is constantly increasing, and the heat generation power per unit area is continuously rising. Traditional air cooling methods are no longer sufficient to meet the heat dissipation requirements under high heat flux density conditions. Liquid cooling technology, due to its advantages such as strong heat transfer capacity and high temperature control accuracy, is gradually becoming an important development direction in the field of high-end chip thermal management.

[0003] In liquid cooling systems, the liquid cooling plate is the core component for heat transfer, and its internal flow channel structure directly determines the cooling efficiency and flow resistance level. Currently, common liquid cooling plate flow channel forms in engineering include serpentine channels, parallel direct-flow channels, and porous filled structures. These structures typically rely on empirical design or parametric optimization methods for improvement. However, these methods are mostly based on fixed flow channel topologies, optimizing by adjusting the channel width, spacing, or number of branches, resulting in limited design freedom and difficulty in fully adapting to complex heat source distributions.

[0004] In recent years, topology optimization methods have been introduced into the field of thermal fluid structure design. By automatically generating flow channel configurations within a given design domain, the flow path can be adaptively adjusted according to the heat source distribution, thereby significantly improving heat dissipation performance. However, most existing topology optimization studies only focus on the minimum average temperature or minimum pressure drop as a single objective, failing to comprehensively consider the synergistic relationship between flow energy consumption and temperature distribution uniformity. Furthermore, some methods employ a complete Navier-Stokes high-fidelity model for solution, resulting in high computational complexity and poor stability of the optimization process, which is not conducive to rapid engineering design.

[0005] On the other hand, under non-uniform heat source conditions, simply reducing the average temperature often fails to effectively suppress local hot spots; while simply reducing the pressure drop may lead to a decrease in coolant utilization efficiency. Therefore, how to control flow energy consumption while ensuring heat dissipation and achieving controllable temperature range has become a key technical problem in the current topology optimization design of liquid cooling plates.

[0006] Based on the above technical background, it is necessary to propose a joint optimization method that simultaneously considers fluid viscous energy dissipation and design domain average temperature. Under the premise of satisfying fluid volume fraction and temperature range constraints, the method can realize the automatic generation of flow channel structure and comprehensive performance improvement, thereby improving the overall heat dissipation efficiency and operational stability of chip liquid cooling system. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, this invention provides a method for optimizing the topology of a chip liquid cooling plate based on NiTi alloy. By performing fluid-thermal coupling topology optimization in a planar design domain and considering the phase transformation of the alloy, the volume fraction of the flow channel is reasonably controlled during optimization, thereby obtaining a planar liquid cooling flow channel.

[0008] A method for optimizing the flow channel topology of a chip liquid cooling plate based on NiTi alloy includes the following steps:

[0009] S1. Obtain the heat source distribution information of the chip, determine the two-dimensional design domain based on the preset outer dimensions of the liquid cooling plate, and establish a coupled numerical model of flow and heat transfer within the two-dimensional design domain.

[0010] S2. Determine the physical properties of the materials used for optimization, the environmental parameters of the liquid cooling plate, and the boundary parameters of the input fluid;

[0011] S3. Based on the interpolation format of solid isotropic materials, the physical property parameters are expressed as functions of the design variable γ, so as to couple the physical property parameters, environmental parameters and boundary parameters with the density model of topology optimization to construct a mathematical model of topology optimization; wherein, the design variable γ is used to characterize the distribution of materials in the two-dimensional design domain, and the value of γ ranges from 0 to 1, with a value close to 1 indicating a fluid region and a value close to 0 indicating a solid region;

[0012] S4. Construct a multi-objective optimization model with the joint objectives of minimizing the fluid viscous energy dissipation rate and minimizing the average temperature of the design domain;

[0013] S5. Set fluid volume fraction constraints and temperature range constraints as two constraints for the optimization process;

[0014] S6. Iteratively solve the design variable γ until the optimal two-dimensional flow channel topology configuration that satisfies the fluid volume fraction constraint and the temperature range constraint is obtained;

[0015] S7. Based on the two-dimensional optimal flow channel topology, design the three-dimensional structure of the liquid cooling plate.

[0016] Further, in step S1, the design domain refers to the two-dimensional basic dimensions of the liquid cooling plate, which include the length and width of the liquid cooling plate, as well as the size and position of the working fluid inlet and outlet within the liquid cooling plate.

[0017] Establishing a coupled numerical model of flow and heat transfer includes: determining the flow channel design space within the two-dimensional design domain, discretizing the flow channel design space into multiple finite elements, and modeling based on the continuity equation, momentum equation, and energy equation.

[0018] Further, in step S2, the material includes a matrix material constituting the solid portion of the liquid cooling plate and a working fluid filled in the flow channel; the physical properties include the phase transformation characteristics of the NiTi alloy, the thermal conductivity k, density ρ, isobaric heat capacity Cp of the material, and the dynamic viscosity coefficient μ of the working fluid; the boundary parameters include at least the heat generation Q of the heat source and the pressure p of the working fluid at the inlet. in With temperature T in The environmental parameters include the initial ambient temperature.

[0019] Furthermore, in step S3, the Darcy interpolation model is used to express the reverse permeability, equivalent thermal conductivity, density, and constant-pressure heat capacity of the porous medium as functions of the design variables, thereby realizing the continuous transition of material properties between fluid and solid states.

[0020] Furthermore, in step S4, the fluid viscous energy dissipation term and the average temperature term of the two-dimensional design domain are normalized respectively, and then the two are linearly combined by weighting coefficients to form a bi-objective optimization function.

[0021] Furthermore, in step S5, the fluid volume fraction constraint is to set an upper limit for the fluid volume, and the temperature range constraint is to ensure that the upper and lower limits of the preset temperature are met within the two-dimensional design domain.

[0022] Furthermore, the characteristic feature is that step S6 specifically includes:

[0023] S61. Initialize the design variable γ;

[0024] S62. Solve for the velocity field, pressure field, and temperature field under the current design variables;

[0025] S63. Calculate the objective function value of the optimization model and the function values ​​of the fluid volume fraction constraint and the temperature range constraint;

[0026] S64. Using the adjoint sensitivity analysis method, calculate the sensitivity of the objective function and constraint function to the design variable γ;

[0027] S65. Based on the sensitivity, update the design variable γ using the moving asymptote algorithm;

[0028] S66. Repeat steps S62 to S65 until the rate of change of the objective function is lower than a preset threshold, at which point the optimization is considered to have converged, and the optimal flow channel topology is obtained.

[0029] Furthermore, in step S66, during the iteration process or after optimization convergence, the design variable γ is binarized to eliminate grayscale units and obtain a clear flow channel structure.

[0030] Furthermore, NiTi shape memory alloy units are arranged on the wall of the fluid region directly below the high heat flux density region of the chip; the NiTi shape memory alloy units are configured to undergo a phase change when they sense a local temperature change, thereby changing the local flow resistance and realizing adaptive distribution of coolant flow.

[0031] A chip liquid cooling plate, wherein the flow channel structure is designed using the method described in any one of claims 1 to 9.

[0032] The technical effects and advantages of the NiTi alloy-based chip liquid cooling plate flow channel topology optimization design method of this invention are as follows:

[0033] 1. By constructing a multi-objective optimization model with the joint objectives of minimizing fluid viscous energy dissipation rate and minimizing the average temperature of the design domain, the limitations of traditional single-objective optimization (which only pursues the lowest pressure drop or the lowest average temperature) are overcome. This method can coordinate the optimization of heat dissipation performance and flow resistance in the design stage, achieve the best balance between heat dissipation efficiency and pumping power consumption, and thus obtain a flow channel topology with better overall performance.

[0034] 2. An innovative temperature range constraint is introduced to control the temperature within the design domain within a preset range. This directly and effectively suppresses localized high-temperature points caused by uneven heat source distribution or unreasonable flow channel design, ensuring a more uniform temperature distribution on the chip surface and significantly improving the thermal reliability and long-term stability of the chip.

[0035] 3. By integrating NiTi shape memory alloy units with the flow channel walls, the liquid cooling plate is endowed with dynamic response capabilities. When the heat load distribution of the chip changes, the local temperature change can trigger the phase transformation of the NiTi alloy, thereby actively changing the flow channel shape or flow resistance in that area, realizing the adaptive redistribution of coolant flow, directing more coolant to high heat flux areas, achieving "on-demand cooling", and further improving the heat dissipation performance and energy efficiency under non-uniform and time-varying heat loads.

[0036] 4. The topology optimization method is adopted, which breaks away from the constraints of parameter optimization based on fixed flow channel topology. This method can freely "grow" the optimal flow channel layout within a given two-dimensional design domain, so that the flow channel path can adaptively match the complex heat source distribution, thereby discovering the globally optimal or near-optimal flow channel configuration with better performance that is difficult to obtain by traditional empirical design.

[0037] 5. By establishing a flow-thermal coupling model in the two-dimensional design domain for topology optimization, the huge computational cost of directly performing three-dimensional topology optimization is significantly reduced. Combined with the adjoint sensitivity analysis method and the moving asymptote (MMA) algorithm, the numerical stability and convergence efficiency of the optimization process are guaranteed, making this method more suitable for rapid design and iteration in engineering. It provides an efficient and practical tool for the innovative design of high-performance chip liquid cooling plates. Attached Figure Description

[0038] Figure 1 This is a flowchart of the present invention.

[0039] Figure 2 This is a planar design domain diagram of the present invention;

[0040] Figure 3 This is a schematic diagram of the two-dimensional design domain and the final topology optimization result of the present invention;

[0041] Figure 4 Temperature cloud map of the chip after using the liquid cooling plate prepared according to the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0043] Please see Figure 1 This invention provides a method for optimizing the topology of a chip liquid cooling plate based on NiTi alloy, the specific implementation steps of which include:

[0044] Step S1: Obtain the heat source distribution information of the chip, determine the two-dimensional design domain based on the preset dimensions of the liquid cooling plate, and establish a coupled numerical model of flow and heat transfer within the two-dimensional design domain; the two-dimensional design domain refers to the two-dimensional basic dimensions of the liquid cooling plate, which include the length and width of the liquid cooling plate, as well as the size and location of the working fluid inlet and outlet within the liquid cooling plate. In this embodiment, the established two-dimensional liquid cooling plate model is as follows: Figure 2 As shown, the inlet width L1 = 8mm, the height L2 = 1mm, and the liquid cooling plate size is 8mm × 6mm.

[0045] Step S2: The solid part of the liquid cooling plate uses steel as the main material. The physical properties of steel are: specific heat capacity 4200 [J / (kg*K)], density 7800 [kg / m^3], and thermal conductivity 44 [W / (m*K)]. The physical properties of the NiTi alloy before phase transformation are: specific heat capacity 550 [J / (kg*K)], density 6.5 [kg / m^3], and thermal conductivity 10 [W / (m*K)]. The working fluid is water, with physical properties of: specific heat capacity 4200 [J / (kg*K)], density 1000... [kg / m^3], thermal conductivity is 0.6 [W / (m*K)]; the initial ambient temperature in this embodiment is set to 20℃; the fluid temperature is 57℃; the inlet pressure is set to 2Pa; the heat dissipation rate of the heat source is set to 2500W; the maximum volume force is set to 1e7 [Pa*s / m^2].

[0046] Step S3: To achieve a continuous transition of materials between fluid and solid states, this invention employs the Darcy interpolation scheme for solid isotropic materials to interpolate the following parameters: reverse permeability α, equivalent thermal conductivity k, and c of the porous medium. p The constant-pressure heat capacity and ρ density are correlated with γ through a nonlinear interpolation function:

[0047]

[0048]

[0049]

[0050]

[0051] Where α(γ) reflects the flow resistance characteristics of porous media, k(γ) characterizes the thermal conductivity of solid-liquid two-phase composites, and c p (γ) is the isobaric heat capacity of the solid-liquid mixture, ρ(γ) is the density of the solid-liquid mixture, and α, k, c p， ρ represents permeability, thermal conductivity, constant pressure heat capacity, and density, respectively, while subscripts s and f represent fluid and solid, respectively.

[0052] In the above equation, q = 0.05. During topology optimization, the solution of the momentum equation in the fluid domain depends on the appropriate value of the damping coefficient α (γ). If q is set to 0, when γ → 0 (pure fluid domain), the damping coefficient α → 0. At this time, the momentum equation has no effective damping term, which will lead to ill-conditioned system stiffness matrix, numerical divergence, and failure to complete optimization iteration. Setting q = 0.05 can provide appropriate minimum damping for the fluid domain, effectively improve the matrix condition number, avoid numerical singularity problems, and ensure stable convergence of the solution process. This is the core premise of this value.

[0053] Topology optimization design is based on the variable density method. The design variable γ represents the material distribution state (0 represents pure solid and 1 represents pure fluid). Theoretically, it should present a binary distribution. For numerical stability considerations, a continuous variable of 0 to 1 is used to describe the relative density γ, allowing intermediate states to exist in the design domain.

[0054] Step S4: Construct a multi-objective optimization model with the joint objectives of minimizing the fluid viscous energy dissipation rate and minimizing the average temperature of the two-dimensional design domain:

[0055] Fluid flow is governed by the continuity equation and the momentum equation, resulting in the fluid control equations for ideal porous media flow:

[0056] The continuity equation for the flow field is:

[0057] ∇·u=0

[0058] In the formula: ∇ is the gradient operator, and u is the fluid velocity;

[0059] The momentum equation (Navier-Stokes equations) for the flow field is:

[0060] ρ(u·∇)u=-∇P+μ∇2u-αu

[0061] Where: P is pressure; ρ is material density; μ is the dynamic viscosity coefficient of the working fluid, which is the Brinkman friction term used in this case to penalize the fluid flowing through the solid region of the design domain in the optimization calculation. This term only applies to the design domain; α is reverse osmosis rate.

[0062] The objective function is defined as minimizing fluid dissipation power. The mathematical model for the flow distribution problem is as follows:

[0063]

[0064] The optimized model for average temperature is expressed as follows:

[0065]

[0066] Normalizing each objective and using weighted coefficients w, the two objectives are linearly combined into the objective function of a multi-objective optimization problem, which can be expressed as follows:

[0067] ,

[0068]

[0069]

[0070] In the formula, Jf is the total potential energy, is the gradient operator, α(γ) is the continuous reverse osmosis function, μ is the dynamic viscosity, w is the weighting factor of the multi-objective function, w1 and w2 are weighting coefficients, Jth is the heat generation, Jth,min is the minimum heat generation, Jth,max is the maximum heat generation, Jth is the dimensionless representation of heat generation, Jf is the dimensionless representation of potential energy, Jf,min is the minimum potential energy, Jf,max is the maximum potential energy, J is the objective function, Ω is the design domain, φ represents the thermal performance factor of the radiator, ψ represents the flow performance factor of the radiator, w is the weighting factor of the multi-objective function, φ0 and ψ0 are the heat dissipation performance and flow performance at the initial design, respectively, and the dimensionless representation of the objective function is achieved by introducing the performance of the initial design.

[0071] Step S5: This invention adopts dual constraints: the flow channel volume fraction is <0.5, the temperature is set to 350K-360K, and the temperature within the two-dimensional design domain must meet the preset upper and lower limits to prevent the generation of local hot spots or the appearance of overcooled areas.

[0072] Step S6: First, an initial solution needs to be obtained. Then, the moving asymptote algorithm (MMA) is used to solve the optimization model, update the design variable x, and enter the next round of iteration calculation. When the rate of change of the objective function is lower than the preset threshold during multiple consecutive iterations, the optimization is determined to be converged.

[0073] Step S6: After convergence, eliminate grayscale cells and design the chip liquid cooling plate structure based on the two-dimensional optimal flow channel configuration after iteration.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit its scope of protection. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading the present invention, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the scope of protection of the pending claims of the invention.

[0075] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for optimizing the topology of a chip liquid cooling plate based on NiTi alloy, characterized in that, Includes the following steps: S1. Obtain the heat source distribution information of the chip, determine the two-dimensional design domain based on the preset outer dimensions of the liquid cooling plate, and establish a coupled numerical model of flow and heat transfer within the two-dimensional design domain. S2. Determine the physical properties of the materials used for optimization, the environmental parameters of the liquid cooling plate, and the boundary parameters of the input fluid; S3. Based on the interpolation format of solid isotropic materials, the physical property parameters are expressed as functions of the design variable γ, so as to couple the physical property parameters, environmental parameters and boundary parameters with the density model of topology optimization to construct a mathematical model of topology optimization; wherein, the design variable γ is used to characterize the distribution of materials in the two-dimensional design domain, and the value of γ ranges from 0 to 1, with a value close to 1 indicating a fluid region and a value close to 0 indicating a solid region; S4. Construct a multi-objective optimization model with the joint objectives of minimizing the fluid viscous energy dissipation rate and minimizing the average temperature of the two-dimensional design domain; S5. Set fluid volume fraction constraints and temperature range constraints as two constraints for the optimization process; S6. Iteratively solve the design variable γ until the optimal two-dimensional flow channel topology configuration that satisfies the fluid volume fraction constraint and the temperature range constraint is obtained; S7. Based on the two-dimensional optimal flow channel topology, design the three-dimensional structure of the liquid cooling plate.

2. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 1, characterized in that, In step S1, the two-dimensional design domain refers to the two-dimensional basic dimensions of the liquid cooling plate, which include the length and width of the liquid cooling plate, as well as the size and position of the working fluid inlet and outlet within the liquid cooling plate. Establishing a coupled numerical model of flow and heat transfer includes: determining the flow channel design space within the two-dimensional design domain, discretizing the flow channel design space into multiple finite elements, and modeling based on the continuity equation, momentum equation, and energy equation.

3. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 1, characterized in that, In step S2, the material includes a matrix material constituting the solid portion of the liquid cooling plate and a working fluid filled in the flow channel; the physical properties include the phase transformation characteristics of the NiTi alloy, the thermal conductivity k, density ρ, isobaric heat capacity Cp of the material, and the dynamic viscosity coefficient μ of the working fluid; the boundary parameters include at least the heat generation Q of the heat source and the pressure p of the working fluid at the inlet. in With temperature T in The environmental parameters include the initial ambient temperature.

4. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 1, characterized in that, In step S3, the Darcy interpolation model is used to express the reverse permeability, equivalent thermal conductivity, density, and constant-pressure heat capacity of the porous medium as functions of the design variables, thereby realizing the continuous transition of material properties between fluid and solid states.

5. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 1, characterized in that, In step S4, the fluid viscous energy dissipation term and the average temperature term of the two-dimensional design domain are normalized respectively, and then the two are linearly combined by weighting coefficients to form a bi-objective optimization function.

6. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 1, characterized in that, In step S5, the fluid volume fraction constraint is to set an upper limit for the fluid volume, and the temperature range constraint is to set an upper and lower limit for the preset temperature of the working fluid within the two-dimensional design domain.

7. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 1, characterized in that, Step S6 specifically includes: S61. Initialize the design variable γ; S62. Solve for the velocity field, pressure field, and temperature field under the current design variables; S63. Calculate the objective function value of the optimization model and the function values ​​of the fluid volume fraction constraint and the temperature range constraint; S64. Using the adjoint sensitivity analysis method, calculate the sensitivity of the objective function and constraint function to the design variable γ; S65. Based on the sensitivity, update the design variable γ using the moving asymptote algorithm; S66. Repeat steps S62 to S65 until the rate of change of the objective function is lower than a preset threshold, which is then determined to be optimization convergence, and the optimal flow channel topology configuration is obtained.

8. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to claim 7, characterized in that, In step S66, during the iteration process or after optimization convergence, the design variable γ is binarized to eliminate grayscale cells and obtain a clear flow channel structure.

9. The chip liquid cooling plate flow channel topology optimization design method based on NiTi alloy according to any one of claims 1 to 7, characterized in that, NiTi shape memory alloy units are arranged on the wall of the fluid region directly below the high heat flux density region of the chip; the NiTi shape memory alloy units are configured to undergo a phase change when they sense a local temperature change, thereby changing the local flow resistance and realizing adaptive distribution of coolant flow.

10. A chip liquid cooling plate, characterized in that, Its flow channel structure is designed using the method described in any one of claims 1 to 9.

Images