Embedded manifold microchannel and global optimization method thereof

Abstract

The invention belongs to the technical field of heat dissipation of electronic equipment, and particularly relates to an embedded manifold microchannel and a global optimization method thereof. The invention provides a global optimization method of an embedded manifold micro-channel, and provides a micro-channel heat sink combining an embedded and manifold structure and a multi-objective optimization design method of the micro-channel heat sink in order to solve the problems that the heat dissipation heat flux density of a data center is gradually increased, and the power consumption index of a matched heat dissipation system is increased along with the heat dissipation heat flux density. A Pareto optimal solution set representing multi-target global optimum is solved by taking structure parameters and heat transfer flowing property parameters as optimization parameters and target parameters and combining means such as simulation calculation, parameter sensitivity analysis, target equation response surface method fitting and an NSGA-II optimization algorithm, so that the heat sink gives consideration to heat transfer and flowing properties, and the heat transfer and flowing properties of the heat sink are improved. And the power consumption of the heat dissipation system is the lowest on the premise of meeting the heat dissipation requirement of the data center.

Description

Technical Field

[0001] The invention belongs to the technical field of heat dissipation of electronic equipment, and in particular relates to an embedded manifold microchannel and a global optimization method thereof. Background Art

[0002] As the heat flux density of electronic equipment gradually increases, it will reach about 10 3 W / cm 2 Microchannels have become one of the main solutions to solve the problem of high heat flux density heat dissipation. However, microchannel heat dissipation enhancement mainly relies on the reduction of the hydraulic diameter of the flow channel, which is usually 10 to 10 2 At the μm level, while the heat dissipation performance is enhanced, the pressure drop index that characterizes the flow resistance characteristics increases more significantly, which greatly increases the power consumption of the supporting heat dissipation system represented by data centers, seriously hindering the application of microchannels.

[0003] In order to take into account both heat transfer and flow performance, researchers gradually introduced embedded structures and manifold structures into microchannel design, with the following advantages:

[0004] First, the advantages of the embedded structure: by designing a channel closer to the heat source surface, the thickness from the heat source surface to the fluid-solid surface is reduced and the number of interfaces is reduced, thereby reducing the thermal resistance of the heat sink and the interface thermal resistance;

[0005] Second, the advantages of the manifold structure: by designing multiple inlets and outlets at the millimeter level to divide the heat dissipation area, the effective flow length of the microchannel is reduced, thereby reducing the pressure drop;

[0006] However, there are still few related patents published on how to design an embedded manifold microchannel structure with optimal heat transfer and flow performance. Based on this, it is necessary to carry out research on structural optimization design methods suitable for embedded manifold microchannels. Summary of the invention

[0007] In order to solve the above problems existing in the prior art, the object of the present invention is to provide an embedded manifold microchannel and a global optimization method thereof.

[0008] The technical solution adopted by the present invention is:

[0009] A global optimization method for an embedded manifold microchannel comprises the following steps:

[0010] S1: Determine the design constraints of the embedded manifold microchannel: overall dimensions, heat source heat flux density, cooling medium type and volume flow rate, substrate type;

[0011] S2: Define target parameters: According to application requirements, define the indicators that affect heat transfer and flow as the target parameter set (α, β…);

[0012] S3: Define structural parameters: According to the microchannel structure, define all structural parameter sets (a, b, c...) except for the shape and size;

[0013] S4: Establish a parametric simulation model: complete the construction of a physical model with editable structural parameters, boundary condition setting, meshing and independence test, and model validity verification;

[0014] S5: Structural parameter sensitivity analysis: Under the constraints of the external dimensions, K1 sets of structural parameter sets (a, b, c…) are uniformly and randomly selected, and the corresponding target parameter sets (α, β…) are respectively obtained in the parametric simulation model. Through sensitivity analysis, K2 structural parameters (a1, b1, c1…) with the greatest impact on the target parameters are screened and retained;

[0015] S6: Define dimensionless optimization parameters: For the selected structural parameters (a1, b1, c1…), take ratios between each two, i.e. X1=a1 / b1, X2=a1 / c1, X3=b1 / c1, etc., define dimensionless parameter sets (X1, X2, X3…), and determine the value range of each dimensionless parameter;

[0016] S7: Obtaining the initial population: The initial population is the original population for multi-objective optimization, which includes information such as dimensionless optimization parameter sets and corresponding target parameter sets. To ensure the diversity of the population, the dimensionless optimization parameter sets need to be distributed as evenly as possible within the value range;

[0017] S8: Target parameter equation fitting by response surface method: Based on the correspondence between the dimensionless optimization parameter set of the initial population and its target parameter set, the calculation equation of the target parameter about the dimensionless optimization parameter is fitted by response surface method, i.e., α=f1(X1,X2,X3…), β=f2(X1,X2,X3…), etc.;

[0018] S9: NSGA-Ⅱ algorithm optimization: The fitted target parameter equation is imported into the NSGA-Ⅱ algorithm, and based on the Pareto dominance relationship, non-dominated sorting, elite retention, and crowding sorting, the crossover probability K4, mutation probability K5, crossover parameter K6, and mutation parameter K7 are defined to carry out population iteration;

[0019] S10: Obtain the Pareto optimal solution set of the target parameters: iterate and calculate n generations, and then obtain the Pareto optimal solution set that represents the optimality of multiple objectives of the target parameter set (α, β…).

[0020] Aiming at the problem that the heat flux density of data center heat dissipation is gradually increasing and the power consumption of the supporting heat dissipation system is exponentially increasing, the present invention proposes a microchannel heat sink combining embedded and manifold structures and a multi-objective optimization design method thereof, respectively taking structural parameters and heat transfer flow performance parameters as optimization parameters and target parameters, and combining simulation calculation, parameter sensitivity analysis, target equation response surface method fitting, NSGA-Ⅱ optimization algorithm and other means to obtain the Pareto optimal solution set representing the multi-objective global optimum, so that the heat sink takes into account both heat transfer and flow performance, and the power consumption of the heat dissipation system is minimized while meeting the heat dissipation needs of the data center.

[0021] As a preferred solution of the present invention, in step S7, the specific solution steps for obtaining the initial population are:

[0022] S71: uniformly randomly select K3 groups of initial dimensionless optimization parameter sets (X1, X2, X3…);

[0023] S72: Import the parameterized simulation model to obtain the corresponding target parameter sets (α, β…).

[0024] As a preferred solution of the present invention, in step S2, the target parameter characterizing the heat dissipation performance is the average convection heat transfer coefficient or thermal resistance.

[0025] As a preferred solution of the present invention, in step S2, the target for characterizing the flow performance is pressure drop or pump work.

[0026] As a preferred solution of the present invention, in step S5 and step S7, the empirical parameter value range is as follows: K1∈[10 ， 30], K2∈[2 ， 4], K3∈[20 ， 40].

[0027] As a preferred solution of the present invention, in step S9, the empirical values ​​of the iteration parameters are as follows: K4=0.9, K5=0.1, K6=15, K7=20.

[0028] As a preferred solution of the present invention, in step S10, n∈[100 ， 300].

[0029] An embedded manifold microchannel comprises a substrate layer, a microchannel layer is arranged on the substrate layer, a manifold layer is arranged on the microchannel layer, the manifold layer is shaped as a reciprocating fold line from one side to the other side, a plurality of notches on one side of the manifold layer serve as cooling medium inlets, and a plurality of notches on the other side of the manifold layer serve as cooling medium outlets.

[0030] As a preferred embodiment of the present invention, the overall structural outer dimensions of the embedded manifold microchannel are length L×width W×height H, the thickness of the substrate layer is t, the height of the microchannel layer is h1, the height of the manifold layer is h2, the long side width w1 of the cooling medium inlet and the cooling medium outlet, the short side width w2 of the cooling medium inlet and the cooling medium outlet, the wall thickness w3 of the manifold layer, the width wc of the microchannel layer, and the wall thickness ww of the microchannel layer.

[0031] As a preferred solution of the present invention, the microchannel height h1, the microchannel width wc and the microchannel wall thickness ww are used as structural parameters that affect the flow and heat transfer performance of the embedded manifold microchannel heat sink.

[0032] The beneficial effects of the present invention are:

[0033] 1. The present invention proposes an embedded manifold microchannel optimization design method that combines simulation calculation, parameter sensitivity analysis, target equation response surface method fitting, NSGA-Ⅱ optimization algorithm and other means, and based on Pareto dominance relationship, non-dominated sorting, elite retention, crowding sorting and other strategies, the global optimal solution within the design constraints can be obtained. At the same time, this method is universal and is also applicable to the optimization design of other forms of heat sink structures other than embedded manifold microchannel structures.

[0034] 2. The present invention obtains the initial population data of the correspondence between the dimensionless optimization parameter set and the target parameter set by means of simulation calculation, and introduces a parameter sensitivity analysis method to screen out the structural parameters that have a greater impact on the target parameters, and further establishes dimensionless structural parameters for optimization. The optimization results are not constrained by absolute size and are more universal. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of the embedded manifold microchannel optimization design;

[0036] Figure 2 It is a schematic diagram of the structure of the embedded manifold microchannel in the first direction;

[0037] Figure 3 This is a schematic diagram of the structure of the embedded manifold microchannel in the second direction.

[0038] Figure 4 is a top view of the embedded manifold microchannel;

[0039] Figure 5 is the left view of the embedded manifold microchannel;

[0040] Figure 6 It is a schematic diagram of the thermal resistance-pressure drop Pareto optimal solution set of the embedded manifold microchannel.

[0041] In the figure: 1 - substrate layer; 2 - microchannel layer; 3 - manifold layer; 4 - cooling medium inlet; 5 - cooling medium outlet. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0043] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present invention. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.

[0044] The present invention provides an embedded manifold microchannel, such as Figure 2 and Figure 3 As shown, the microchannel comprises a substrate layer 1, a microchannel layer 2, a manifold layer 3, a cooling medium inlet 4, a cooling medium outlet 5 and the like. The microchannel layer 2 is arranged on the substrate layer 1 of the embedded manifold microchannel, and the manifold layer 3 is arranged on the microchannel layer 2. The shape of the manifold layer 3 is a reciprocating fold line from one side to the other side. Several notches on one side of the manifold layer 3 serve as the cooling medium inlet 4, and several notches on the other side of the manifold layer 3 serve as the cooling medium outlet 5.

[0045] For this type of embedded manifold microchannel structure, in order to obtain the best heat transfer and flow performance, it is necessary to optimize the structural parameters. Among them, the heat transfer and flow performance are usually characterized by thermal resistance and pressure drop, respectively. Since it involves the optimization of multiple target parameters, this is a multi-objective optimization problem, the purpose of which is to find the Pareto optimal solution set containing multiple sets of thermal resistance and pressure drop.

[0046] like Figure 1 As shown, the specific optimization steps can be described as follows:

[0047] S1: Determine the design constraints of the embedded manifold microchannel: overall dimensions, heat source heat flux density, cooling medium type and volume flow rate, substrate type;

[0048] S2: Define target parameters: According to application requirements, define the indicators that affect heat transfer and flow as the target parameter set (α, β…);

[0049] S3: Define structural parameters: According to the microchannel structure, define all structural parameter sets (a, b, c...) except for the shape and size;

[0050] S4: Establishing a parametric simulation model: Based on the above process, complete the construction of a physical model with editable structural parameters, boundary condition setting, meshing and independence test, model validity verification, etc.;

[0051] S5: Structural parameter sensitivity analysis: Under the constraints of the external dimensions, K1 sets of structural parameter sets (a, b, c…) are uniformly and randomly selected, and the corresponding target parameter sets (α, β…) are respectively obtained in the parametric simulation model. Through sensitivity analysis, K2 structural parameters (a1, b1, c1…) with the greatest impact on the target parameters are screened and retained;

[0052] S6: Define dimensionless optimization parameters: For the selected structural parameters (a1, b1, c1…), take ratios between each two, i.e. X1=a1 / b1, X2=a1 / c1, X3=b1 / c1, etc., define dimensionless parameter sets (X1, X2, X3…), and determine the value range of each dimensionless parameter;

[0053] S7: Obtaining the initial population: The initial population is the original population for multi-objective optimization, which contains information such as dimensionless optimization parameter sets and corresponding target parameter sets. To ensure the diversity of the population, the dimensionless optimization parameter sets need to be distributed as evenly as possible within the value range. The specific solution steps are as follows:

[0054] S71: uniformly randomly select K3 groups of initial dimensionless optimization parameter sets (X1, X2, X3…);

[0055] S72: Import the parameterized simulation model to obtain corresponding target parameter sets (α, β…);

[0056] S8: Target parameter equation fitting by response surface method: Based on the correspondence between the dimensionless optimization parameter set of the initial population and its target parameter set, the calculation equation of the target parameter about the dimensionless optimization parameter is fitted by response surface method, i.e., α=f1(X1,X2,X3…), β=f2(X1,X2,X3…), etc.;

[0057] S9: NSGA-Ⅱ algorithm optimization: The fitted target parameter equation is imported into the NSGA-Ⅱ algorithm, and based on strategies such as Pareto dominance relationship, non-dominated sorting, elite retention, and crowding sorting, and iterative parameters such as crossover probability K4, mutation probability K5, crossover parameter K6, and mutation parameter K7 are defined to carry out population iteration;

[0058] S10: Obtain the Pareto optimal solution set of the target parameters: iterate and calculate n generations, and then obtain the Pareto optimal solution set that represents the optimality of multiple objectives of the target parameter set (α, β…).

[0059] In a specific implementation, in step S2, the target parameter characterizing the heat dissipation performance may be the average convection heat transfer coefficient, thermal resistance, etc., and the target parameter characterizing the flow performance may be the pressure drop, pump work, etc.

[0060] In a specific implementation scheme, in step S5 and step S7, the empirical parameter value range is as follows: K1∈[10 ， 30], K2∈[2 ， 4], K3∈[20 ， 40].

[0061] In a specific implementation, in step S9, the empirical values ​​of the iteration parameters are as follows: K4=0.9, K5=0.1, K6=15, K7=20.

[0062] In a specific embodiment, in step S10, n∈[100 ， 300].

[0063] The present invention proposes an embedded manifold microchannel optimization design method that combines simulation calculation, parameter sensitivity analysis, target equation response surface method fitting, NSGA-Ⅱ optimization algorithm and other means, and based on Pareto dominance relationship, non-dominated sorting, elite retention, crowding sorting and other strategies, the global optimal solution within the design constraints can be obtained. At the same time, this method is universal and is also applicable to the optimization design of other forms of heat sink structures other than embedded manifold microchannel structures.

[0064] The present invention obtains the initial population data of the correspondence between the dimensionless optimization parameter set and the target parameter set by means of simulation calculation, and introduces the parameter sensitivity analysis method to screen out the structural parameters that have a greater impact on the target parameters, and further establishes the dimensionless structural parameters for optimization. The optimization results are not constrained by absolute size and are more universal.

[0065] Example:

[0066] A schematic diagram of a near-space thermal management device using the present invention is shown in FIG. Figure 2 and Figure 3 As shown, its working principle is: electronic equipment generates heat during operation. In order to make it work near the optimal operating temperature, a thermal management device is required to adjust the heat dissipation and thermal insulation performance in real time according to the heat transfer effect of the external environment.

[0067] A schematic diagram of an embedded manifold microchannel using the present invention is shown in FIG. Figure 2 and Figure 3As shown, its working principle is as follows: the microchannel has multiple cooling medium inlets and outlets. The cooling medium enters the manifold layer 3 through the inlet and flows vertically downward into the microchannel layer 2, and flows to the left and right sides, and performs sufficient convection heat exchange in the process, and then flows vertically upward into the manifold layer 3 for convergence. It flows out through the outlet to complete the heat dissipation process.

[0068] By dividing the flow through the manifold layer 3, the flow length of the microchannel part under a single flow path is effectively reduced, thereby greatly reducing the pressure drop. The microchannel layer 2 is brought close to the heat source through the embedded structure, reducing the heat dissipation thermal resistance, enhancing the heat dissipation effect, and ultimately taking into account both heat transfer and flow performance.

[0069] Two-dimensional image of embedded manifold microchannels Figure 4 and Figure 5 As shown in Figure 1, the outer dimensions of the heat sink are L×W×H, the substrate thickness is t, the height of the microchannel layer 2 is h1, the height of the manifold layer 3 is h2, the long side width of the inlet and outlet is w1, the short side width is w2, the wall thickness of the inlet and outlet layer is w3, and the width and wall thickness of the microchannel are wc and ww respectively. 0 summarizes all the structural parameters and values ​​of the embedded manifold microchannel.

[0070] Table 1 Summary of embedded manifold microchannel structural parameters

[0071]

[0072]

[0073] Before carrying out the multi-objective optimization of embedded manifold microchannel thermal resistance-pressure drop, it is necessary to first determine the structural parameters to be optimized. The selection of structural parameters is based on: selecting the dominant factors that affect thermal resistance and pressure drop as much as possible, retaining the main influencing factors while minimizing the number of parameters. Based on the preliminary simulation results, it can be seen that the structural parameters that affect the flow and heat transfer performance of embedded manifold microchannel heat sinks are mainly microchannel height h1, microchannel width wc and microchannel wall thickness ww. In order to make the research conclusions universal, the above three structural parameters are dimensionless as X1 = h1 / H, X2 = wc / h1 and X3 = wc / ww.

[0074] Among them, X1 is the ratio of the microchannel height to the sum of the flow channel height, and its value range is set to (0.2, 0.8). X2 is the microchannel aspect ratio, and its value range is (0.1, 0.8). X3 is the ratio of the microchannel wall thickness to the width, and its value range is (0.5, 2.5). Then, through Latin sampling, 118 sample points were uniformly and randomly selected within the value range of the optimization parameters, that is, 118 (X1, X2, X3) optimization parameter combinations. Table 2 summarizes the optimization algorithm parameter values. Figure 6 It is the Pareto optimal solution set of thermal resistance-pressure drop of embedded manifold microchannel.

[0075] Table 2 Summary of optimization algorithm parameters

[0076]

[0077] Table 3 Pareto optimal solution set

[0078]

[0079]

[0080] The present invention is not limited to the above-mentioned optional implementation modes. Anyone can derive other various forms of products under the inspiration of the present invention. However, no matter what changes are made in the shape or structure, all technical solutions that fall within the scope defined by the claims of the present invention fall within the protection scope of the present invention.

Claims

1. A global optimization method for embedded manifold microchannels, characterized in that: The following steps are involved: S1: Determine the design constraints of the embedded manifold microchannel: overall dimensions, heat source heat flux density, cooling medium type and volume flow rate, substrate type; S2: Define target parameters: According to application requirements, define the indicators that affect heat transfer and flow as the target parameter set (α, β…); S3: Define structural parameters: According to the microchannel structure, define all structural parameter sets (a, b, c...) except the shape and size; S4: Establish a parametric simulation model: complete the construction of a physical model with editable structural parameters, boundary condition setting, meshing and independence test, and model validity verification; S5: Structural parameter sensitivity analysis: Under the constraints of the external dimensions, K1 sets of structural parameter sets (a, b, c…) are uniformly and randomly selected, and the corresponding target parameter sets (α, β…) are respectively obtained in the parametric simulation model. Through sensitivity analysis, K2 structural parameters (a1, b1, c1…) with the greatest impact on the target parameters are screened and retained; S6: Define dimensionless optimization parameters: For the selected structural parameters (a1, b1, c1…), take ratios between each two, i.e. X1=a1 / b1, X2=a1 / c1, X3=b1 / c1, etc., define dimensionless parameter sets (X1, X2, X3…), and determine the value range of each dimensionless parameter; S7: Obtaining the initial population: The initial population is the original population for multi-objective optimization, which includes information such as dimensionless optimization parameter sets and corresponding target parameter sets. To ensure the diversity of the population, the dimensionless optimization parameter sets need to be distributed as evenly as possible within the value range; S8: Target parameter equation fitting by response surface method: Based on the correspondence between the dimensionless optimization parameter set of the initial population and its target parameter set, the calculation equation of the target parameter about the dimensionless optimization parameter is fitted by response surface method, i.e., α=f1(X1,X2,X3…), β=f2(X1,X2,X3…), etc.; S9: NSGA-Ⅱ algorithm optimization: The fitted target parameter equation is imported into the NSGA-Ⅱ algorithm, and based on the Pareto dominance relationship, non-dominated sorting, elite retention, and crowding sorting, the crossover probability K4, mutation probability K5, crossover parameter K6, and mutation parameter K7 are defined to carry out population iteration; S10: Obtain the Pareto optimal solution set of the target parameters: iteratively calculate n generations, and then obtain the Pareto optimal solution set that represents the optimality of multiple objectives of the target parameter set (α, β…).

2. The global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: In step S7, the specific solution steps for obtaining the initial population are: S71: uniformly randomly select K3 groups of initial dimensionless optimization parameter sets (X1, X2, X3…); S72: Import the parameterized simulation model to obtain the corresponding target parameter sets (α, β…).

3. The global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: In step S2, the target parameter characterizing the heat dissipation performance is the average convection heat transfer coefficient or thermal resistance.

4. The global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: In step S2 , the target characterizing the flow performance is pressure drop or pump work.

5. The global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: In step S5 and step S7, the empirical parameter value range is as follows: K1∈[10 ， 30], K2∈[2 ， 4], K3∈[20 ， 40].

6. The global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: In step S9, the empirical values ​​of the iteration parameters are as follows: K4=0.9, K5=0.1, K6=15, K7=20.

7. The global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: In step S10, n∈[100 ， 300].

8. An embedded manifold microchannel, using the global optimization method of an embedded manifold microchannel according to claim 1, characterized in that: The invention comprises a substrate layer (1), a microchannel layer (2) is arranged on the substrate layer (1), a manifold layer (3) is arranged on the microchannel layer (2), the manifold layer (3) is in the shape of a reciprocating fold line from one side to the other side, a plurality of notches on one side of the manifold layer (3) serve as cooling medium inlets (4), and a plurality of notches on the other side of the manifold layer (3) serve as cooling medium outlets (5).

9. The embedded manifold microchannel according to claim 8, characterized in that: The overall structural dimensions of the embedded manifold microchannel are length L×width W×height H, the thickness of the substrate layer (1) is t, the height of the microchannel layer (2) is h1, the height of the manifold layer (3) is h2, the long side width w1 of the cooling medium inlet (4) and the cooling medium outlet (5), the short side width w2 of the cooling medium inlet (4) and the cooling medium outlet (5), the wall thickness w3 of the manifold layer (3), the width wc of the microchannel layer (2), and the wall thickness ww of the microchannel layer (2).

10. The embedded manifold microchannel according to claim 9, characterized in that: Microchannel height h1, microchannel width wc and microchannel wall thickness ww are structural parameters that affect the flow and heat transfer performance of the embedded manifold microchannel heat sink.

Images