A method for controlling flow field uniformity of an immersion liquid cooling cabinet

Abstract

The application discloses a kind of control methods of flow field uniformity of immersion liquid cooling cabinet, to solve the problem that temperature uniformity in the liquid pool of immersion liquid cooling cabinet in prior art cannot be accurately and energy-saving controlled, a kind of control methods of flow field uniformity of immersion liquid cooling cabinet, comprising the following steps: S1, detect the temperature of each point in the cabinet, and calculate the average temperature and variance Tf in the cabinet, and obtain the cabinet setting temperature Ts;S2, whether to improve the liquid pump frequency is determined by judging the cabinet setting temperature Ts and the cabinet liquid temperature Tc;S3, judge the temperature difference ΔT and variance Tf in the cabinet, determine the overall uniformity of temperature in the cabinet and whether there is local hot spot;The application accurately identifies local hot spot and the uniformity of temperature in the cabinet by the temperature change in the cabinet, and then judges the operating load of IT equipment, according to the feedback of temperature, carries out accurate heat dissipation control and energy-saving control, improves the stability and reliability of equipment in data center.

Description

Technical Field

[0001] This invention relates to the field of immersion liquid-cooled cabinet technology, and more specifically to a method for controlling the flow field uniformity of immersion liquid-cooled cabinets. Background Technology

[0002] With the advancement of technology, the performance requirements of electronic information equipment are becoming increasingly demanding, leading to a significant increase in the heat generation and heat flux density of electronic components. Consequently, the power consumption required to cool these components will inevitably increase exponentially. Currently, the power consumption of a single CPU chip has reached 300-400W, and that of a single GPU chip is as high as 700-800W. The power consumption of a single server rack is projected to exceed 30KW by 2025. This excessive heat dissipation concentrated in the confined space of the server rack presents a significant challenge. How to effectively dissipate heat while simultaneously ensuring temperature uniformity across servers and even individual chips are problems that immersion liquid-cooled server racks must address urgently.

[0003] Traditional air cooling cannot meet heat dissipation requirements, necessitating immersion cooling. Immersion cooling achieves high-power heat dissipation, improves energy efficiency, and effectively enhances server computing efficiency and extends chip lifespan. However, due to variations in server load and computing power demands, peak computing power varies significantly, requiring multi-scenario solutions to heat dissipation and flow field optimization. Existing technologies address these issues by adjusting liquid pump flow and simply regulating inlet and outlet water temperatures, often at the expense of energy consumption to ensure temperature uniformity within the liquid tank. Current technologies often rely on single-point threshold judgments to adjust liquid or water pumps, increasing flow rate and flow field disturbance to improve temperature uniformity, but this process results in significant fluctuations and is not energy-efficient.

[0004] The Chinese patent document "An Immersion Liquid-Cooled Cabinet and Data Center", with publication number CN115087315A and publication date of September 20, 2022, provides an immersion liquid-cooled cabinet and data center, including: an immersion tank configured as a box-shaped structure with an opening at the top, the immersion tank being inserted into the cabinet through the opening, the immersion tank containing coolant, and slots for placing servers being provided in the immersion tank; however, this Chinese patent does not address the issue of solving the problem of temperature uniformity within the liquid tank. Summary of the Invention

[0005] This invention solves the problems of inaccurate temperature uniformity and energy-saving control within the liquid pool of immersion liquid-cooled cabinets in existing technologies. It proposes a method for controlling the flow field uniformity of immersion liquid-cooled cabinets. By analyzing the temperature changes within the cabinet, it accurately identifies local hot spots and the temperature uniformity within the cabinet, thereby determining the operating load of IT equipment. Based on the temperature feedback, it performs precise heat dissipation and energy-saving control, improving the stability and reliability of data center equipment.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for controlling the flow field uniformity of an immersion liquid-cooled cabinet, comprising the following steps:

[0007] S1, detect the temperature at various points inside the cabinet, calculate the average temperature and variance Tf inside the cabinet, and obtain the cabinet set temperature Ts; S2, determine whether to increase the liquid pump frequency by comparing the cabinet set temperature Ts with the cabinet liquid outlet temperature Tc.

[0008] S3, determine the temperature difference ΔT and variance Tf inside the cabinet to determine the overall temperature uniformity inside the cabinet and whether there are local hot spots;

[0009] If the maximum value of the temperature difference ΔT inside the cabinet is greater than the first preset parameter B and the variance Tf is greater than or equal to the second preset parameter C, then all impellers in the area corresponding to the maximum value of the temperature difference ΔT inside the cabinet will be activated; and the trend of the overall temperature uniformity inside the cabinet will be determined by judging the first derivative of the temperature variance Tf inside the cabinet.

[0010] In this invention, the average temperature and variance of the cabinet are first calculated based on the temperature at various points within the cabinet. Then, a judgment is made regarding the cabinet set temperature Ts and the cabinet outlet liquid temperature Tc to determine whether to increase the liquid pump frequency. Next, a judgment is made regarding the temperature difference ΔT and variance Tf within the cabinet. After the judgment, there are four possible scenarios, each requiring different processing. For one scenario, where the maximum value of the temperature difference ΔT is greater than the first preset parameter B and the variance Tf is greater than or equal to the second preset parameter C, local hot spots exist, and the overall uniformity within the cabinet is poor. In this case, all impellers in the area corresponding to the maximum value of the temperature difference ΔT need to be activated to enhance disturbance and improve heat exchange performance. Furthermore, the trend of overall temperature uniformity within the cabinet is further determined by judging the first derivative of the temperature variance Tf, achieving prediction and precise control.

[0011] Preferably, step S2 includes the following two cases:

[0012] Q1. If Tc≥Ts+A, the outlet temperature is too high. Increase the pump frequency. The pump frequency increase is Tc / Ts*1.5. The value of A is from 1 to 8℃.

[0013] Q2, if Tc < Ts + A, then the outlet temperature is normal.

[0014] In this invention, if Tc ≥ Ts + A, it indicates that the outlet temperature is too high and there is an abnormality, and the pump frequency needs to be increased; if Tc < Ts + A, no adjustment is needed; after step S2, the temperature difference ΔT and variance Tf inside the cabinet are judged.

[0015] Preferably, step S3 includes the following cases:

[0016] D1. If the variance Tf inside the cabinet is greater than or equal to C and the maximum value of the temperature difference ΔT inside the cabinet is greater than B, then there are local hot spots and the overall uniformity inside the cabinet is poor. The value of B is in the range of 5 to 10℃ and the value of C is in the range of 1 to 5℃.

[0017] D2. If the variance Tf inside the cabinet is greater than or equal to C and the maximum value of the temperature difference ΔT inside the cabinet is less than or equal to B, then there are no local hot spots, but the overall uniformity inside the cabinet is poor.

[0018] D3. If the variance Tf inside the cabinet is less than C and the maximum value of the temperature difference ΔT inside the cabinet is greater than B, then the overall uniformity inside the cabinet is good, but there are local hot spots.

[0019] D4. If the variance Tf within the cabinet is less than C and the maximum value of the temperature difference ΔT within the cabinet is less than or equal to B, then the overall uniformity within the cabinet is good and there are no local hot spots.

[0020] In this invention, there are four cases regarding the determination of the temperature difference ΔT and variance Tf inside the cabinet. Except for the case where the variance Tf inside the cabinet is less than C and the maximum value of the temperature difference ΔT inside the cabinet is less than or equal to B, adjustments are required to improve heat exchange performance.

[0021] Preferably, the average temperature inside the cabinet is as follows:

[0022] T0 = ​​(T1 + T2 + T3 + T4 + T5 + T6) / 6

[0023] Among them, T1, T2, T3, T4, T5 and T6 represent the temperature values ​​at various points inside the cabinet;

[0024] The variance Tf within the cabinet is specifically:

[0025] Tf=[(T1-T0)^2+(2-T0)^2+(T3-T0)^2+(T41-T0)^2+(T5-T0)^2+(T6-T0)^2] / 5.

[0026] In this invention, after calculating the average temperature inside the cabinet, the variance value Tf of the temperature inside the cabinet is also calculated, which facilitates the calculation and judgment of subsequent steps.

[0027] Preferably, the step of determining the overall temperature uniformity trend within the cabinet by judging the first derivative of the temperature variance value Tf within the cabinet specifically includes:

[0028] If f(Tf)≤D, then the temperature variance inside the cabinet is decreasing, and the overall temperature uniformity inside the cabinet is showing a trend of improvement. The value of D ranges from -0.05 to -0.2.

[0029] If f(Tf) > D, the temperature variance inside the cabinet is increasing, and the overall temperature uniformity inside the cabinet tends to deteriorate. Therefore, the pump frequency should be increased by (Tf)*MaxΔT / T0*50.

[0030] In this invention, when f(Tf)≤D, the temperature variance inside the cabinet is continuously decreasing, meaning the overall temperature uniformity inside the cabinet is continuously improving. Conversely, when f(Tf)≤D, it indicates that the overall temperature uniformity inside the cabinet is continuously deteriorating, requiring an increase in the pump frequency for adjustment.

[0031] Preferably, in step S1, temperature sensors are installed at various points inside the cabinet, and the temperature at each point inside the cabinet is collected by the temperature sensors.

[0032] In this invention, a temperature sensor is used to collect the temperature at various points, and the collected data is stored and calculated.

[0033] As a preferred option, for scenario D2, activate some impellers in the central area of ​​the cabinet; for scenario D3, activate some impellers in the central area of ​​the cabinet; for scenario D4, no action is required.

[0034] In this invention, some impellers in the central area of ​​the cabinet are impellers near the center of the cabinet. A spatial coordinate system is established with the center of the cabinet as the origin and divided into multiple equally sized areas. The impellers are set at the bottom of the cabinet and are evenly arranged horizontally. The impellers in the corresponding areas are activated according to the temperature difference inside the cabinet.

[0035] The present invention has the following beneficial effects:

[0036] 1. The present invention relates to a method for controlling the flow field uniformity of an immersion liquid-cooled cabinet. By measuring the temperature changes inside the cabinet, local hot spots and the temperature uniformity inside the cabinet are accurately identified, thereby determining the operating load of the IT equipment. Based on the temperature feedback, precise heat dissipation control and energy-saving control are performed to improve the stability and reliability of data center equipment.

[0037] 2. In this invention, by determining the temperature threshold inside the cabinet and the first derivative of the temperature variance inside the cabinet, the temperature and its change trend can be detected in real time, thereby enabling prediction and precise control. Attached Figure Description

[0038] Figure 1 This is a flowchart of a method for controlling the flow field uniformity of an immersion liquid-cooled cabinet according to the present invention;

[0039] Figure 2This is a schematic diagram of the temperature sensor position in Embodiment 2 of the method for controlling the flow field uniformity of an immersion liquid-cooled cabinet in this invention;

[0040] CG1-CG6 all refer to temperature sensors. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely one preferred embodiment of the invention and are only used to explain the invention. They do not limit the scope of protection of the invention. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0042] Example 1:

[0043] This embodiment proposes a method for controlling the flow field uniformity of an immersion liquid-cooled cabinet, referencing... Figure 1 The process includes the following steps.

[0044] Step S1: Detect the temperature at various points within the server rack, calculate the average temperature and variance Tf within the server rack, and obtain the set server rack temperature Ts. Specifically, the temperature at each point is collected using temperature sensors, which are positioned at corresponding locations within the server rack. In this embodiment, six temperature sensors are used, and their specific locations are as follows: Figure 2 As shown, Figure 2 This is a plan view of the immersion liquid-cooled cabinet from one perspective.

[0045] Specifically, the average temperature inside the server rack is as follows:

[0046] T0 = ​​(T1 + T2 + T3 + T4 + T5 + T6) / 6

[0047] Among them, T1, T2, T3, T4, T5 and T6 represent the temperature values ​​at various points inside the cabinet;

[0048] The variance Tf within the rack is specifically:

[0049] Tf = [(T1-T0)^2 + (2-T0)^2 + (T3-T0)^2 + (T41-T0)^2 + (T5-T0)^2 + (T6-T0)^2] / 5. In this embodiment, after calculating the average temperature inside the cabinet, the variance Tf of the temperature inside the cabinet is also calculated, which facilitates the calculation and judgment in subsequent steps.

[0050] For the temperature difference ΔT between various points, please refer to Table 1:

[0051] Table 1: Temperature difference ΔT between various points

[0052]

[0053]

[0054] Step S2 involves determining whether to increase the liquid pump frequency by comparing the cabinet set temperature Ts with the cabinet outlet liquid temperature Tc. Specifically, step S2 includes the following two scenarios.

[0055] In case Q1, if Tc ≥ Ts + A, the outlet temperature is too high. Therefore, the pump frequency is increased by Tc / Ts * 1.5, and the value of A ranges from 1 to 8℃.

[0056] In case Q2, if Tc < Ts + A, the outlet temperature is normal. In this embodiment, if Tc ≥ Ts + A, it indicates that the outlet temperature is too high and there is an abnormality, and the pump frequency needs to be increased; if Tc < Ts + A, no adjustment is needed; after step S2, the temperature difference ΔT and variance Tf inside the cabinet are determined.

[0057] Step S3: Determine the temperature difference ΔT and variance Tf within the cabinet to ascertain the overall temperature uniformity within the cabinet and whether there are any local hot spots; specifically, this includes the following situations.

[0058] If the maximum value of the temperature difference ΔT inside the cabinet is greater than the first preset parameter B and the variance Tf is greater than or equal to the second preset parameter C, then all impellers in the area corresponding to the maximum value of the temperature difference ΔT inside the cabinet are activated, i.e., all impellers in the area corresponding to MaxΔT are activated; and the trend of the overall temperature uniformity inside the cabinet is determined by judging the first derivative of the temperature variance Tf inside the cabinet; this situation corresponds to case D1, in which there are local hot spots and the overall uniformity inside the cabinet is poor, the value of B ranges from 5 to 10℃, and the value of C ranges from 1 to 5℃; in this case, the trend of the overall temperature uniformity inside the cabinet is also determined by judging the first derivative of the temperature variance Tf inside the cabinet. The temperature variation trend is obtained by taking the first derivative of the temperature variance Tf. Specifically, when f(Tf) ≤ D, the temperature variance in the cabinet is decreasing, and the overall temperature uniformity in the cabinet tends to improve. The value of D ranges from -0.05 to -0.2. When f(Tf) > D, the temperature variance in the cabinet is increasing, and the overall temperature uniformity in the cabinet tends to worsen. In this case, the pump frequency is increased by (Tf)*MaxΔT / T0*50. After the judgment is completed, the cabinet temperatures T1, T2, T3, T4, T5, T6 and the cabinet set temperature Ts are re-detected.

[0059] For case D2, if the variance Tf inside the cabinet is greater than or equal to C and the maximum value of the temperature difference ΔT inside the cabinet is less than or equal to B, i.e., MaxΔT≤B and Tf≥C, then there are no local hot spots, but the overall uniformity inside the cabinet is poor. In this case, some impellers in the central area of ​​the cabinet are activated to enhance the disturbance and improve the heat exchange performance.

[0060] For case D3, if the variance Tf inside the cabinet is less than C and the maximum value of the temperature difference ΔT inside the cabinet is greater than B, i.e. MaxΔT>B and Tf<C, then the overall uniformity inside the cabinet is good, but there are local hot spots. In this case, some impellers in the central area of ​​the cabinet are activated to enhance the disturbance and improve the heat exchange performance.

[0061] For case D4, if the variance Tf within the cabinet is less than C and the maximum value of the temperature difference ΔT within the cabinet is less than or equal to B, i.e., MaxΔT≤B and Tf<C, then the overall uniformity within the cabinet is good and there are no local hot spots. Therefore, no treatment is required for case D4.

[0062] After step S3 is completed, the temperature at each point inside the rack and the rack setting temperature Ts are re-detected and periodically monitored and judged to achieve online and timely detection and space management of the data center, thereby improving the effectiveness and timely feedback of intelligent control.

[0063] In this embodiment, the average temperature and variance of the cabinet are first calculated based on the temperature at various points within the cabinet. Then, a judgment is made regarding the cabinet set temperature Ts and the cabinet outlet liquid temperature Tc to determine whether to increase the liquid pump frequency. Next, a judgment is made regarding the temperature difference ΔT and variance Tf within the cabinet. After the judgment, there are four possible scenarios, each requiring different processing. For one scenario, where the maximum value of the temperature difference ΔT is greater than the first preset parameter B and the variance Tf is greater than or equal to the second preset parameter C, local hot spots exist, and the overall uniformity within the cabinet is poor. In this case, all impellers in the area corresponding to the maximum value of the temperature difference ΔT need to be activated to enhance disturbance and improve heat exchange performance. Furthermore, the trend of overall temperature uniformity within the cabinet is further determined by judging the first derivative of the temperature variance Tf, achieving prediction and precise control.

[0064] In this embodiment, there are four cases regarding the determination of the temperature difference ΔT and variance Tf inside the cabinet. Except for the case where the variance Tf inside the cabinet is less than C and the maximum value of the temperature difference ΔT inside the cabinet is less than or equal to B, adjustments are required to improve heat exchange performance.

[0065] In this embodiment, when f(Tf)≤D, the temperature variance value inside the cabinet is continuously decreasing, that is, the overall temperature uniformity inside the cabinet is continuously improving. Conversely, when f(Tf)≤D, it indicates that the overall temperature uniformity inside the cabinet is continuously deteriorating, and the liquid pump frequency needs to be increased for adjustment.

[0066] In this embodiment, some impellers in the central area of ​​the cabinet are those near the center of the cabinet. A spatial coordinate system is established with the center of the cabinet as the origin, and multiple equally sized areas are divided. By identifying the temperature of the liquid pool in the cabinet, local hot spots can be accurately identified and controlled. The impellers are located at the bottom of the cabinet and are evenly distributed horizontally. The impellers in the corresponding areas are activated according to the temperature difference in the cabinet. In addition, low-power impellers are used to generate disturbances, which enhances heat exchange performance, thereby improving the overall heat dissipation performance of the cabinet and reducing the PUE value.

[0067] This embodiment achieves both accuracy and cost-effectiveness in judgment and adjustment by making multiple threshold judgments. Specifically, it determines the temperature inside the cabinet, the set temperature of the cabinet, and the variance of the temperature inside the cabinet to confirm the load and flow field conditions, thereby enabling precise control during the process and reducing fluctuations.

[0068] In this embodiment, the temperature variation and uniformity within the cabinet are identified by performing a first derivative calculation on the variance of the liquid pool temperature change.

[0069] Example 2:

[0070] refer to Figure 2 , Figure 2 This is a plan view of an immersion liquid-cooled cabinet from one perspective. CG1-CG6 represent temperature sensors. Specifically, the total height of the cabinet is b, the total length is a, and the total width is c (not shown in the figure). The height b1 of temperature sensor CG1 ranges from 100-500mm, and the height b2 of temperature sensor CG2 satisfies:

[0071] B2 = B1 + (B - B1) / 3,

[0072] The height b3 of the temperature sensor CG3 satisfies:

[0073] B3 = BM,

[0074] The value of M ranges from 30 to 300 mm, and in this embodiment, it is 150 mm.

[0075] The width c1 of temperature sensor CG1 ranges from 0 to 0.5 cm, and the width c2 of temperature sensor CG2 ranges from 1 to 0.5 cm. The above dimensions are only one example of implementing the present invention, and the position dimensions of the temperature sensors are not limited to the above dimensions.

Claims

1. A method of controlling the flow field uniformity of an immersion liquid-cooled cabinet, The method comprises the steps of: S1, detecting the temperature of each point in the cabinet, calculating the average temperature and variance Tf of the cabinet, and obtaining the cabinet setting temperature Ts; S2, determining whether to increase the frequency of the liquid pump by judging the cabinet setting temperature Ts and the cabinet liquid outlet temperature Tc; S3, judging the temperature difference ΔT and variance Tf in the cabinet to determine the overall uniformity of the temperature in the cabinet and whether there is a local hot spot; If the maximum value of the temperature difference ΔT in the cabinet is greater than the first preset parameter B and the variance Tf is greater than or equal to the second preset parameter C, all the impellers in the region corresponding to the maximum value of the temperature difference ΔT in the cabinet are started; And the change trend of the overall uniformity of the temperature in the cabinet is determined by judging the first derivative of the variance Tf of the temperature in the cabinet; S3 comprises: D1, if Tf≥C and the maximum value of ΔT is greater than B, there is a local hot spot and the overall uniformity of the temperature in the cabinet is poor; D2, if Tf≥C and the maximum value of ΔT is less than or equal to B, there is no local hot spot, but the overall uniformity of the temperature in the cabinet is poor; D3, if Tf is less than C and the maximum value of ΔT is greater than B, the overall uniformity of the temperature in the cabinet is good, but there is a local hot spot; D4, if Tf is less than C and the maximum value of ΔT is less than or equal to B, the overall uniformity of the temperature in the cabinet is good and there is no local hot spot; For D2 and D3, part of the impellers in the central region of the cabinet are started; for D4, no processing is required; If f(Tf)≤D, the variance of the temperature in the cabinet is decreasing, and the overall uniformity of the temperature in the cabinet is improving; If f(Tf)>D, the variance of the temperature in the cabinet is increasing, and the overall uniformity of the temperature in the cabinet is deteriorating, then the frequency of the liquid pump is increased.

2. The method of claim 1, wherein the method further comprises: The step S2 comprises the following two cases: Q1, if Tc≥Ts+A, the liquid outlet temperature is too high, the frequency of the liquid pump is increased, the frequency of the liquid pump is increased by Tc / Ts*1.5, and the value range of A is 1-8℃; Q2, if Tc<Ts+A, the liquid outlet temperature is normal.

3. The method of claim 1 or 2, wherein the method further comprises: In D1, the value range of B is 5-10℃, and the value range of C is 1-5℃.

4. The method of claim 3, wherein the method further comprises: The average temperature in the cabinet is specifically: T0=(T1+T2+T3+T4+T5+T6) / 6 Wherein, T1, T2, T3, T4, T5 and T6 represent the temperature values of each point in the cabinet; The variance Tf of the temperature in the cabinet is specifically: Tf=[(T1-T0)^2+(T2-T0)^2+(T3-T0)^2+(T4-T0)^2+(T5-T0)^2+(T6-T0)^2] / 5.

5. The method of claim 1 or 2 or 4, wherein, The value range of D is -0.05 to -0.2, and the increase range of the frequency of the liquid pump is (Tf)*MaxΔT / T0*50.

6. The method of claim 1 or 2 or 4, wherein, In the step S1, temperature sensors are arranged at each point in the cabinet, and the temperature of each point in the cabinet is collected by the temperature sensors.

7. The method of claim 1, wherein the method further comprises: The part of the impellers in the central region of the cabinet is the impeller near the center of the cabinet, a space coordinate system is established with the center of the cabinet as the origin, and a plurality of equal regions are divided; the impellers are arranged on the bottom of the cabinet and evenly arranged along the horizontal direction, and the impellers in the corresponding region are started according to the temperature difference in the cabinet.

Images