Cooling distribution system and its operating method

Abstract

A cooling distribution system and its speed setting method are proposed. The method includes performing proportional-integral-derivative (PID) calculations based on a target temperature and the current temperature of a heat-generating component to generate a speed control parameter. The pump and fan speeds are set according to the speed control parameter, pump weight, and fan weight, calculating the first total power consumption of the pump and fan, where the pump weight equals the fan weight. The fan weight is adjusted along the numerical direction, and an optimization process is executed: the fan speed is set based on the speed control parameter and fan weight, calculating the second total power consumption of the pump and fan. The first and second total power consumptions are compared. Before reaching the preset number of optimization iterations, if the second total power consumption is lower, the fan weight is adjusted along the numerical direction; otherwise, it is adjusted in the opposite direction, and the optimization process is re-executed.

Description

Technical Field

[0001] This invention relates to direct wafer liquid cooling, and more particularly to a cooling distribution system and its operation method. Prior Technology

[0002] In server liquid cooling technology, Direct Liquid to Chip (DLC) has become the mainstream solution. DLC primarily uses a cold plate to directly contact high-wattage heat-generating chips (such as CPUs and GPUs) and removes heat using coolant. Based on the type of heat exchanger (HX), DLC can be divided into water-cooled (L2L HX) and air-cooled (L2A HX). Water-cooled DLC relies on chilled water and plate heat exchangers for cooling, while air-cooled DLC uses a liquid-to-air heat exchanger (L2A HX) and a fan for heat dissipation. DLC cooling systems typically include two independent cooling loops. For example, water-cooled DLC relies on two pump motors to maintain coolant circulation, while air-cooled DLC uses a pump motor to drive the coolant and a fan to promote heat exchange. The core cooling device of air-cooled DLC is the Cooling Distribution Unit (CDU), which mainly consists of an air-cooled heat exchanger, a coolant pressurization pump, and a fan. Its operation involves a pump providing coolant circulation, delivering chilled water to the water-cooled plate for heat exchange, and then returning the water after absorbing heat to exchange heat with air via a liquid-to-air heat exchanger, ultimately continuously outputting chilled water to supply the water-cooled plate.

[0003] Existing DLC ​​cooling technologies primarily focus on optimizing the water-cooled plate and heat exchanger, while paying less attention to optimizing the energy consumption of the pump motor and fan. Currently, most L2A CDU control methods employ a fixed maximum flow strategy. Under maximum system load, the pump motor maintains high speed, while the fan speed adjusts based on the chip temperature or heat exchanger outlet water temperature. When the temperature rises, the fan accelerates to improve cooling efficiency; when the temperature decreases, the fan decelerates to save energy. However, this control method does not consider the energy distribution between the pump and fan, potentially leading to excessive energy consumption by the cooling equipment. Furthermore, the fixed maximum flow operating mode cannot adapt to changes in demand under different load conditions, further reducing overall energy efficiency. Summary of the Invention

[0004] In view of this, the present invention proposes a cooling distribution system and its operation method, which uses an optimization algorithm to reduce total power consumption and determines the speed ratio of the pump motor and the fan.

[0005] An operation method for a cooling distribution system according to an embodiment of the present invention is applicable to a cooling distribution system, wherein the cooling distribution system includes at least a temperature sensor, a fan, a pump, and a control device. The method is executed by the control device and includes: obtaining a current temperature of a heat-generating element from the temperature sensor; performing a proportional-integral-derivative control-based calculation based on a target temperature and the current temperature to generate a speed control parameter; setting a speed of the pump based on the speed control parameter and a pump weight; setting a speed of the fan based on the speed control parameter and a fan weight; and calculating a first total power consumption of the pump and fan, wherein the pump weight is equal to the fan weight; The fan weight is adjusted along a numerical direction, and an optimization program is executed. The optimization program includes: setting the fan speed according to the speed control parameters and fan weight, and calculating a second total power consumption of the pump and fan; comparing the second total power consumption with the first total power consumption; when the second total power consumption is less than the first total power consumption and the number of times the optimization program is executed is less than a preset number, replacing the first total power consumption with the second total power consumption, adjusting the fan weight along the numerical direction, and executing the optimization program again; and when the second total power consumption is not less than the first total power consumption and the number of times the optimization program is executed is less than the preset number, adjusting the fan weight in the opposite direction to the numerical direction, and executing the optimization program again.

[0006] A cooling distribution system according to an embodiment of the present invention includes: a temperature sensor, a heat exchanger, a fan, a pump, and a control device. The temperature sensor is used to obtain a current temperature of a heating element. The heat exchanger is used to lower the temperature of coolant in an input line. The fan is used to lower the temperature of the heat exchanger. The pump is used to push coolant in an output line. The control device is electrically connected to the fan and the pump. The control device is used to: perform proportional-integral-derivative control-based calculations based on a target temperature (current temperature) to generate a speed control parameter; set a pump speed based on the speed control parameter and a pump weight; set a fan speed based on the speed control parameter and a fan weight, and calculate a first total power consumption of the pump and fan, wherein the pump weight is equal to the fan weight; adjust the fan weight along a numerical direction, and execute an optimization program, the optimization program including: setting the fan speed based on the speed control parameter and the fan weight, and calculating a second total power consumption of the pump and fan; comparing the second total power consumption with the first total power consumption; when the second total power consumption is less than the first total power consumption and the number of times the optimization program is executed is less than a preset number, replacing the first total power consumption with the second total power consumption, adjusting the fan weight along the numerical direction, and executing the optimization program again; and when the second total power consumption is not less than the first total power consumption and the number of times the optimization program is executed is less than a preset number, adjusting the fan weight in the opposite direction to the numerical direction, and executing the optimization program again.

[0007] In summary, this invention uses chip temperature as sensing information and employs a single controller to optimize the control of the liquid-to-air cooling distribution unit, focusing on the speed distribution mechanism between the pump motor and the fan. This method can improve the cooling efficiency of the liquid-to-air cooling distribution unit and reduce the energy consumption of the cooling equipment.

[0008] The foregoing description of the disclosure and the following description of the embodiments are intended to demonstrate and explain the spirit and principles of the present invention, and to provide a further explanation of the scope of the patent application of the present invention. Simple Explanation of the Diagram

[0009] Figure 1 is a schematic diagram of a cooling distribution system and its operating environment according to an embodiment of the present invention;

[0010] Figure 2 is a flowchart illustrating the operation method of a cooling distribution system according to an embodiment of the present invention; and

[0011] Figure 3 is a detailed flowchart of one step in Figure 2. Implementation

[0012] The following detailed description of the features and advantages of the present invention is sufficient to enable anyone skilled in the art to understand the technical content of the present invention and implement it accordingly. Based on the disclosure, patent claims, and drawings in this specification, anyone skilled in the art can easily understand the relevant objectives and advantages of the present invention. The following embodiments further illustrate the points of the present invention, but are not intended to limit the scope of the present invention in any way.

[0013] Figure 1 is a schematic diagram of a cooling distribution system and its operating environment according to an embodiment of the present invention. As shown in Figure 1, the cooling distribution system 1 includes: a heat exchanger 11, a fan 13, a pump 15, a control device 17, and a temperature sensor 19. The cooling distribution system 1 is an air-cooled DLC, wherein the heat exchanger 11, fan 13, and pump 15 constitute an L2A CDU. The present invention aims to optimize the cooling and heat dissipation drivers, namely the fan 13 and the pump 15. Furthermore, the present invention does not limit the number of fans 13 and pumps 15.

[0014] As shown in Figure 1, the heat exchanger 11 delivers cooled coolant to the water-cooled plate 21 through the output pipeline P1. The water-cooled plate 21 exchanges heat with the heating element 23, and the heated coolant is sent to the heat exchanger 11 through the input pipeline P2.

[0015] Heat exchanger 11 is used to reduce the temperature of the coolant in input line P2. Fan 13 is used to reduce the temperature of heat exchanger 11. Pump 15 is used to push coolant in output line P1. Temperature sensor 19 is used to sense the current temperature of heating element 23. Control device 17 is electrically connected to fan 13, pump 15, and temperature sensor 19. In one embodiment, control device 17 may include one or more processing units with data receiving, recording, processing, storage, and output functions, such as a microcontroller, central processing unit, graphics processor, programmable logic controller, or any combination thereof. Control device 17 is used to execute the operation method of cooling distribution system 1 according to an embodiment of the present invention, as shown in FIG2.

[0016] Figure 2 is a flowchart illustrating the operation method of a cooling distribution system 1 according to an embodiment of the present invention. The method is executed by a control device 17 and includes steps S0 to S8.

[0017] In step S0, the control device 17 obtains the current temperature of the heating element 23 from the temperature sensor 19.

[0018] In step S1, the control device 17 performs proportional-integral-derivative (PID) control-based calculations based on the target temperature and the current temperature to generate speed control parameters. The target temperature is an upper limit temperature value set by the user to ensure the normal operation of the heating element 23. The current temperature is taken from a temperature sensor 19 located adjacent to the heating element 23 and electrically connected to the control device 17. In one embodiment, step S1 includes steps S11 to S13 as shown in FIG3.

[0019] In step S11, the control device 17 calculates the temperature difference between the target temperature and the current temperature.

[0020] In step S12, the control device 17 obtains a set of target control parameters corresponding to the temperature difference based on a pre-stored parameter table. The pre-stored parameter table records the correspondence between multiple preset temperature ranges and multiple sets of target control parameters. Each of these target control parameters includes a proportional coefficient, an integral coefficient, and a derivative coefficient, as shown in Table (1) below:

[0021] Table (1) is an example of a pre-stored parameter table. Preset temperature range Temperature difference Target control parameters I 1 [K p1K i1K d1] I 2 [K p2K i2K d2] I 3 [K p3K i3K d3] I 4 [K p4K i4K d4] Non-reactive region [K p5K i5K d5]

[0022] In Table (1), I1 to I4 represent the first to fourth temperature ranges. This represents the temperature difference (target temperature minus current temperature). to These represent the first to fourth temperature differences, respectively. Kp1 to Kp5 represent proportional coefficients, Ki1 to Ki5 represent integral coefficients, and Kd1 to Kd5 represent differential coefficients.

[0023] In step S13, the control device 17 generates speed control parameters using the target control parameters and the temperature difference. Following the previous example, after obtaining a set of target control parameters corresponding to the temperature difference according to Table (1), the control device 17 can generate speed control parameters using this set of target control parameters, the temperature difference, and the following relationship (A):

[0024] Relation (A): ,in For speed control parameters, This represents the temperature difference.

[0025] In step S2, the control device 17 sets the speed of pump 15 according to the speed control parameters and pump weight, sets the speed of fan 13 according to the speed control parameters and fan weight, and calculates the first total power consumption of pump 15 and fan 13, wherein the pump weight is equal to the fan weight. In one embodiment, the value of the pump weight is 1, and the value of the fan weight is 0 to 1.

[0026] In step S3, the control device 17 adjusts the fan weight along a numerical direction. Next, the control device 17 executes an optimization procedure including steps S4 to S8. In step S4, the control device 17 sets the fan speed of 13 according to the speed control parameters and the fan weight, and calculates the second total power consumption of the pump 15 and the fan 13.

[0027] In step S5, the control device 17 compares the second total power consumption with the first total power consumption, for example, determining whether the second total power consumption is less than the first total power consumption. If the second total power consumption is less than the first total power consumption, then proceed to step S6. If the second total power consumption is not less than or equal to the first total power consumption, then proceed to step S7.

[0028] In step S6, the control device 17 replaces the first total power consumption with the second total power consumption and adjusts the fan weight along the original numerical direction.

[0029] In step S7, the control device 17 adjusts the fan weight in the direction opposite to the direction of the numerical value. After steps S6 and S7 are completed, step S8 is executed.

[0030] In step S8, the control device 17 determines whether the number of times the optimization program has been executed is less than a preset number. In one embodiment, the preset number is 5 times. If the determination is yes, the process returns to step S4 to execute the optimization program again. If the determination is no, the operation method of the cooling distribution system according to an embodiment of the present invention ends.

[0031] In steps S3 to S8, the Unidirectional Hill Climbing (UHC) algorithm from the optimization algorithm is used. UHC is a single-objective random search algorithm. This algorithm uses the current point's position as a basis to randomly search for any nearby point. When the fitness of the search point is better than that of the current point (i.e., in step S5, the second total power consumption is less than the first total power consumption), the position is moved to the search point. In this way, the algorithm continuously moves towards positions with lower fitness until no better solution can be found.

[0032] This invention applies UHC to optimize the energy consumption of a server water-cooled plate cooling system. Taking an air-cooled DLC as an example, the water-cooled plate 21 dissipates heat by removing the heat generated by the heat-generating element 23 through the action of the pump 15 and the fan 13. The total power consumption of the pump 15 and the fan 13 is the target of optimization in this invention. In different systems, sometimes the power consumption of the pump 15 is greater than that of the fan 13, and sometimes the power consumption of the fan 13 is greater than that of the pump 15, depending on the design considerations of the cooling distribution system 1. For example, when the impedance of the water circuit is high, the power consumption of the pump 15 will be greater. Theoretically, there is an optimal ratio between the operation of the pump 15 and the fan 13. In other words, under the premise of maintaining the current temperature of the heat-generating element 23 not exceeding the target temperature, the speed of each of the pump 15 and the fan 13 has an optimal setting value to minimize the total cooling energy consumption. Based on this concept, this invention defines the fitness of the cooling distribution system as the total power consumption of the pump 15 and the fan 13, as shown below:

[0033] ,in Represents fitness. Represents the power consumption of a single fan. This represents the power consumption of a single pump, therefore This represents the total power consumption of all fans. This represents the sum of the power consumption of all pumps.

[0034] The optimal control combination of pump 15 and fan 13, determined by an optimization algorithm, will be adjusted based on adaptability. In air-cooled DLCs, the heat dissipation response is greater for pump 15 than for fan 13. This means that when the load on the heat-generating element 23 changes, pump 15 must react quickly to accommodate the temperature change of the heat-generating element 23, preventing the current temperature from becoming too high. In one embodiment of this invention, the following control ratio is defined:

[0035] ,in Represents pump weight. Represents fan weight. This represents the rotational speed of pump 15. This represents the rotational speed of fan 13. The meaning of this proportional control relationship is: the control device 17 outputs the rotational speed control parameter based on the temperature difference between the target temperature and the current temperature. At that time, the rotational speed of pump 15 And the fan speed of fan 13 ,in The value is between 0 and 1. This is achieved by using a fixed pump 15 with a distribution ratio of 1. The algorithm then only needs to search for the ratio of fan 13, whose control ratio is between 0 and 1.

[0036] UHC assigns fan weights based on a fitness formula. The optimization process is shown in Figure 2. In the initialization phase, specifically step S2, the initial state is the fan weight. Furthermore, fan 13 and pump 15 have the same control parameters, namely: .

[0037] After initialization, the operation method of the cooling distribution system 1 according to an embodiment of the present invention is executed. After the temperature is balanced and it is confirmed that the current temperature of the heating element does not exceed the target temperature, the fitness is calculated according to the formula to obtain the first total power consumption as the power consumption reference value. Then, the UHC starts to search for the proportion of fan 13 to make the fitness value smaller. The temperature balance means that under the action of the control device 17, the measured temperature stably approaches the target temperature.

[0038] Step S3 begins searching for solutions with lower values, which means starting to optimize the fan weights. For example, setting fan weight. Control device 17 adjusts according to fan weight The CDU control test is re-executed until temperature equilibrium is reached. If the specifications are met and there are no concerns about overheating, a new fitness is calculated, as shown in step S4. If the new fitness (second total power consumption) is smaller than the baseline value (first total power consumption), i.e., if step S5 determines this to be true, then this fitness is updated to the baseline value, as shown in step S6, and the fan weights are adjusted. Update to 0.8, and then continue changing in the same direction, that is, searching along the original numerical value direction, such as fan weight. The search continues until the total power consumption no longer decreases. If the new fitness is greater than the baseline value (i.e., step S5 determines otherwise), the search proceeds in the opposite direction, as shown in step S7. This process continues until the number of searches exceeds a preset number, as shown in step S8.

[0039] Table 2, Test Examples. Power consumption unit: Watt Heat dissipation of heating element First total power consumption Second total power consumption Power saving ratio 150 11.54 [0.8, 1.0] 10.44 9.51% 175 17.01 [0.8, 1.0] 15.82 6.99% 200 22.84 [0.8, 1.0] 20.41 10.65% 225 37.55 [0.6, 1.0] 34.27 8.73% 250 60.31 [0.8, 1.0] 54.29 9.98% 275 100.39 [0.8, 1.0] 68.95 31.32%

[0040] Table 2 shows the optimization of fan weights using the cooling distribution system operation method according to an embodiment of the present invention. The results, as shown in Table 2, indicate that under high power loads, the method proposed in this invention yields an optimal ratio of [0.8, 1.0], resulting in power savings of up to 31.32%. This demonstrates that significant energy savings can be achieved by properly allocating the control ratio between the pump 15 and the fan.

[0041] In summary, this invention uses chip temperature as sensing information and employs a single controller to optimize the control of the liquid-to-air cooling distribution unit, focusing on the speed distribution mechanism between the pump motor and the fan. This method can improve the cooling efficiency of the liquid-to-air cooling distribution unit and reduce the energy consumption of the cooling equipment.

[0042] In one embodiment of the present invention, the cooling distribution system and its operation method can be used on a server, and the server can be used for artificial intelligence (AI) computing, edge computing, or as a 5G server, cloud server, or vehicle networking server.

[0043] While the present invention has been disclosed above with reference to the foregoing embodiments, it is not intended to limit the invention. Any modifications and refinements made without departing from the spirit and scope of the invention are within the scope of patent protection of the present invention. For details regarding the scope of protection defined in the present invention, please refer to the appended claims.

[0044] 1: Cooling distribution system 11: Heat Exchanger 13: Fan 15: Pump 17: Control device 19: Temperature sensor P1: Output pipeline P2: Input pipeline 21: Water-cooled plate 23: Heating element S0~S8: Steps S11~S13: Steps

Claims

1. A method of operating a cooling distribution system, applicable to a cooling distribution system, wherein the cooling distribution system includes at least a temperature sensor, a fan, a pump, and a control device, the method being executed by the control device and comprising: Obtain a current temperature of a heating element from the temperature sensor; A speed control parameter is generated by performing a proportional-integral-derivative (PI-DE) control operation based on a target temperature and the current temperature; a pump speed is set based on the speed control parameter and a pump weight, a fan speed is set based on the speed control parameter and a fan weight, and a first total power consumption of the pump and the fan is calculated, wherein the pump weight is equal to the fan weight; the fan weight is adjusted along a numerical direction, and an optimization program is executed, the optimization program including: setting the fan speed based on the speed control parameter and the fan weight, calculating a second total power consumption of the pump and the fan; and comparing the second total power consumption with the first total power consumption. When the second total power consumption is less than the first total power consumption and the number of times the optimization program is executed is less than a preset number, the second total power consumption replaces the first total power consumption, the fan weight is adjusted along the direction of the value magnitude, and the optimization program is executed again; and when the second total power consumption is not less than the first total power consumption and the number of times the optimization program is executed is less than the preset number, the fan weight is adjusted along the direction opposite to the direction of the value magnitude, and the optimization program is executed again.

2. The method of operating the cooling distribution system as described in claim 1, wherein the optimization procedure further includes: After setting the fan speed according to the speed control parameters and the fan weight, it is confirmed that the current temperature of the heating element does not exceed the target temperature.

3. The method of operating the cooling distribution system as described in claim 1, wherein the pump weight is 1 and the fan weight is 0 to 1.

4. The method of operating the cooling distribution system as described in claim 1, wherein the preset number of times is 5.

5. The method of operating the cooling distribution system as claimed in claim 1, wherein performing calculations based on proportional-integral-derivative control to generate the speed control parameters according to the target temperature and the current temperature of the heating element includes: Calculate a temperature difference between the target temperature and the current temperature; Based on a pre-stored parameter table, a set of target control parameters corresponding to the temperature difference is obtained. The pre-stored parameter table records the correspondence between multiple preset temperature ranges and multiple sets of target control parameters. Each of these target control parameters includes a proportional coefficient, an integral coefficient, and a derivative coefficient. The speed control parameters are generated using the set of target control parameters and the temperature difference.

6. A cooling distribution system, comprising at least: A temperature sensor for obtaining a current temperature of a heating element; A heat exchanger for reducing the temperature of a coolant in an input line; a fan for reducing the temperature of the heat exchanger; and a pump for propelling the coolant in an output line. The system includes a control device electrically connected to the fan and the pump. The control device is configured to: perform proportional-integral-derivative (PI-DI) control-based calculations based on a target temperature and the current temperature to generate a speed control parameter; set a pump speed based on the speed control parameter and a pump weight; set a fan speed based on the speed control parameter and a fan weight, and calculate a first total power consumption of the pump and the fan, wherein the pump weight is equal to the fan weight; adjust the fan weight along a numerical direction, and execute an optimization program, the optimization program including: adjusting the speed control parameter and the current temperature based on the target temperature and the current temperature; and performing calculations based on the target temperature and the current temperature to generate a speed control parameter. The fan weight is set to the fan speed, and a second total power consumption of the pump and the fan is calculated; the second total power consumption is compared with the first total power consumption; when the second total power consumption is less than the first total power consumption and the number of times the optimization program is executed is less than a preset number, the second total power consumption is used to replace the first total power consumption, the fan weight is adjusted in the direction of the value magnitude, and the optimization program is executed again; when the second total power consumption is not less than the first total power consumption and the number of times the optimization program is executed is less than the preset number, the fan weight is adjusted in the direction opposite to the value magnitude, and the optimization program is executed again.

7. The cooling distribution system as described in claim 6, wherein the optimization procedure further includes: After setting the fan speed according to the speed control parameters and the fan weight, it is confirmed that the current temperature of the heating element does not exceed the target temperature.

8. The cooling distribution system as described in claim 6, wherein the pump weight is 1 and the fan weight is between 0 and 1.

9. The cooling distribution system as described in claim 6, wherein the preset number of times is 5.

10. The cooling distribution system as claimed in claim 6, wherein the control device is further configured to calculate a temperature difference between the target temperature and the current temperature; obtain a set of target control parameters corresponding to the temperature difference based on a pre-stored parameter table, wherein the pre-stored parameter table records the correspondence between multiple preset temperature ranges and multiple sets of target control parameters, each of the target control parameters including a proportional coefficient, an integral coefficient and a derivative coefficient; and generate the speed control parameters using the set of target control parameters and the temperature difference.

Images