Single-phase immersion liquid cooling system based on air floating fluid technology and control method thereof

Abstract

The application relates to a single-phase immersion liquid cooling system based on air floating fluid technology and a control method thereof. Micron-sized bubbles are injected in the high heat flux density area of a server through an air floating fluid strengthening module, the strong disturbance characteristics of gas-liquid two-phase flow are utilized, local fluid flow velocity is significantly improved, and the laminar boundary layer of high-viscosity insulating cooling liquid on the surface of a chip is destroyed. Under the same operating conditions, compared with a traditional liquid cooling scheme, the heat dissipation efficiency is greatly improved, the chip junction temperature is significantly reduced, the local heat accumulation and uneven heat dissipation problem caused by the weak flowability of high-viscosity synthetic oil is solved. Moreover, a multi-scene adaptive control strategy is adopted, the control parameter benchmark and the adjustment weight of each actuator are dynamically adjusted according to different seasonal ambient temperatures and real-time thermal load working conditions. Extremely fast response can be realized for load mutation, the back-liquid temperature is maintained in a very high temperature control precision range, and the control delay and thermal hysteresis phenomenon caused by the heat conduction characteristics of high-viscosity insulating cooling liquid are overcome.

Description

Technical Field

[0001] This application relates to the field of heat dissipation for high heat flux density servers in data centers, and in particular to a single-phase immersion liquid cooling system based on air flotation fluid technology and its control method. Background Technology

[0002] With the rapid development of artificial intelligence (AI), big data, and cloud computing technologies, the demand for computing power in data centers is growing exponentially. The widespread adoption of high-density computing nodes and AI servers has driven the thermal design power (TDP) of a single chip to exceed 1000W and is evolving towards 1500W. Against this backdrop, traditional air-cooling technology has reached its limits, and single-phase immersion liquid cooling technology, with its high power efficiency ratio (PUE), environmental friendliness, and excellent insulation performance, is gradually becoming the core solution for cooling high heat flux density servers.

[0003] Currently, mainstream single-phase immersion liquid cooling systems commonly use hydrocarbon-based synthetic oils (such as PAO series) or silicon-based synthetic oils (such as dimethyl silicone oil 50cs series) as cooling media. However, these insulating coolants generally have high viscosity (typically ≥10mm² / s at 25℃) and high specific heat capacity. In natural convection or circulation modes driven solely by external pumps, high-viscosity fluids easily form a laminar boundary layer on the chip surface, resulting in slow flow rates (typically only about 10mm / s), leading to uneven local heat dissipation and low thermal convection efficiency. Furthermore, the cooling capacity of hydrocarbon-based insulating coolants is often limited to 400W / U, making it difficult to meet the needs of next-generation chips. In addition, due to the high viscosity and large specific heat capacity of the insulating coolant, the liquid temperature rise often lags significantly behind the actual heat generation changes of the chip, producing a significant thermal hysteresis effect, making it difficult for the heat dissipation system to respond instantly to millisecond-level fluctuations in chip load.

[0004] Furthermore, in terms of control strategies, existing single-phase immersion liquid cooling systems mostly employ fixed-parameter PID control algorithms. Because data centers experience significant temperature variations throughout the year (spanning winter, transitional seasons, and summer), and server load rates dynamically fluctuate between 30% and 100%, fixed parameters struggle to adapt to the differentiated heat dissipation needs under varying seasons and load rates. This easily leads to overshooting or undershooting, resulting in wasted energy or the risk of chip overheating. Simultaneously, sensors in existing systems typically lack redundancy and automatic calibration mechanisms, leading to data drift and distortion after long-term operation, further deteriorating control accuracy. Therefore, there is an urgent need for a single-phase immersion liquid cooling system that can overcome the heat dissipation bottleneck of high-viscosity fluids, possesses full-condition adaptive control capabilities, and exhibits high system robustness. Summary of the Invention

[0005] This application provides a single-phase immersion liquid cooling system and its control method based on air flotation fluid technology, so as to at least solve the problem that the heat dissipation performance of liquid cooling systems in related technologies cannot meet the heat dissipation requirements of data centers.

[0006] In a first aspect, embodiments of this application provide a single-phase immersion liquid cooling system based on air flotation fluid technology, the system comprising: A liquid-cooled cabinet includes a first liquid chamber and an inner liquid pool for immersing and accommodating servers. A liquid distribution device is provided between the first liquid chamber and the inner liquid pool, and the liquid distribution device has a plurality of overflow holes. An air-float fluid enhancement module, whose output end passes through the first liquid cavity and extends to the inner liquid pool, is used to inject microbubbles into the insulating coolant in the liquid-cooled cabinet to form an air-liquid enhanced heat transfer flow field in the server heat dissipation module. The cooling distribution device is connected to the liquid cooling cabinet through a secondary side pipeline to form a secondary side closed loop. Its input end and output end are respectively connected to the liquid inlet and liquid outlet of the liquid cooling cabinet, which is used to drive the insulating coolant to circulate and perform heat exchange. The primary-side circulation module is connected to the cooling distribution device through a primary-side pipeline to form a primary-side heat exchange loop, which is used to provide cooling water with adjustable flow rate to the cooling distribution device; The control module is communicatively connected to the air flotation fluid enhancement module, the cooling distribution device, and the primary side circulation module, respectively, and the control module is configured to execute the system adaptive control method.

[0007] In some embodiments, the air flotation fluid enhancement module includes an air source supply unit and an array-type air outlet assembly; The gas supply unit includes a gas generator, a pressure stabilizing tank, and a flow regulating valve group, used to output clean gas with adjustable pressure and flow rate; The array-type air outlet assembly is connected to the air source supply unit via a gas delivery pipeline and is located at the bottom of the liquid cooling cabinet. The array-type air outlet component adopts a microporous aeration structure, and the projection area of ​​its air outlet covers the high heat flux density area of ​​each server, which is used to generate microbubbles with a diameter within a preset range and inject them into the insulating coolant.

[0008] In some embodiments, the liquid cooling cabinet is vertically divided into a liquid inlet rectification zone, a heat exchange zone, and a gas-liquid separation zone; The liquid inlet rectification area is located at the bottom of the liquid cooling cabinet, and the liquid distribution device is installed inside it. The array-type air outlet assembly is integrated below the liquid distribution device or is installed through the liquid distribution device. The heat exchange zone includes multiple movable fluid baffles for dividing the inner liquid pool into multiple independent vertical fluid domains for individually accommodating servers, thereby constraining the rising path of the microbubbles. The gas-liquid separation zone is located at the top of the liquid cooling cabinet. It is connected to the heat exchange zone through a gas-liquid separation filter and to the cooling distribution device through a secondary side pipeline. The gas-liquid separation filter is used to intercept and eliminate microbubbles rising to the liquid surface.

[0009] In some embodiments, the fluid baffle is disposed between adjacent servers within the heat exchange zone to adjust the cross-sectional area of ​​the vertical fluid domain according to the physical dimensions of the servers, thereby constraining the insulating coolant to flow vertically upward along the server cross-section. A filling block is also provided in the non-heating area of ​​the inner liquid pool. The filling block is a solid block that does not chemically react with the insulating coolant and is used to occupy the unused space in the inner liquid pool to guide the insulating coolant to the heat-generating parts of the server.

[0010] In some embodiments, the system further includes a sensor network electrically connected to the control module for collecting system operating status data. The sensing network adopts a redundant architecture, including a main sensor group and a backup sensor group. The main sensor group and the backup sensor group are configured in pairs to monitor the return liquid temperature of the liquid-cooled cabinet, the server power consumption, and the external ambient temperature, respectively. The control module is configured to execute data verification logic to compare the collected data of the main sensor group and the backup sensor group in real time, and to execute fault switching control logic when the deviation between the two sets of data exceeds a safety threshold.

[0011] Secondly, this application also provides a control method, which is applied to the single-phase immersion liquid cooling system based on air flotation fluid technology and executed by the control module, including the following steps: Acquire system operating status data, including ambient temperature, server load rate, return liquid temperature and its rate of change; Based on the ambient temperature and the server load rate, determine the current operating scenario and call the PID baseline parameters and weight allocation coefficients corresponding to the current operating scenario; Based on the deviation between the return liquid temperature and the target threshold and its rate of change, the PID baseline parameters are corrected, and the total adjustment requirement is calculated using the corrected PID parameters. Based on the weighting coefficients, the total adjustment demand is calculated into air flotation flow rate adjustment, primary side water pump frequency adjustment, and CDU circulation pressure adjustment. According to the calculation results, control commands are sent in parallel to the air flotation fluid enhancement module, the primary side circulation module, and the cooling distribution device.

[0012] In some embodiments, the method further includes: The control module pre-sets a mapping strategy between different operating scenarios and the weight allocation coefficients; The weight allocation coefficients are dynamically assigned based on the heat dissipation requirement level of the operating scenario: In operating scenarios where the heat dissipation demand is determined to be greater than the first preset threshold or an extreme load, the weight coefficient of the air flotation fluid enhancement module is adjusted to be greater than the weight coefficients of the primary side circulation module and the cooling distribution device, so as to give priority to the use of air flotation fluid to enhance heat exchange efficiency. In operating scenarios where the load is determined to be below the second preset threshold or where energy saving is prioritized, the weight coefficient of the air flotation fluid enhancement module is reduced, and the weight coefficient of the primary side circulation module or the cooling distribution device is increased, so as to give priority to the use of natural cold sources or low power consumption operating modes.

[0013] In some embodiments, the method further includes a parameter correction step for a specific operating condition: When the deviation between the return liquid temperature and the target threshold is detected to be positive and greater than the first preset threshold, the proportional coefficient of the PID algorithm is increased and the integral coefficient is decreased to speed up the response. When the deviation is detected to be negative and the absolute value is greater than the second preset threshold, the proportional coefficient is decreased and the integral coefficient is increased to suppress overshoot oscillation. When the server load rate is detected to rise beyond the mutation threshold within a preset time window, the integral coefficient is kept unchanged, and the proportional coefficient and derivative coefficient are simultaneously increased to enhance the ability to predict and compensate for temperature mutations. The real-time dynamic viscosity of the insulating coolant is calculated based on the return temperature. When the real-time dynamic viscosity is higher than the preset viscosity threshold, the integral time constant of the PID is extended and the differential coefficient is increased to counteract the thermal convection hysteresis effect caused by the high viscosity insulating coolant.

[0014] In some embodiments, the control commands sent by the air flotation fluid enhancement module include air flotation activation logic, wherein: When any one of the server load rate, the return liquid temperature, or the chip junction temperature exceeds its corresponding static threshold, the air flotation fluid enhancement module is activated. When the calculated temperature deviation and temperature change rate are greater than their respective preset trend thresholds, the air flotation fluid enhancement module is activated at a preset time before the actual temperature exceeds the limit, and the differential control action in the PID algorithm is increased simultaneously to counteract the thermal hysteresis effect of the insulating coolant.

[0015] In some embodiments, sending control commands to the air flotation fluid enhancement module further includes air flotation shutdown logic, wherein: When the return liquid temperature falls below the shut-off threshold and the rate of change remains stable for a preset duration, the graded shut-off process is initiated. The air flotation flow rate is reduced in stages according to a preset step size. During the reduction process, the proportional coefficient of the PID algorithm is reduced and the integral coefficient is increased. If the rebound of the return liquid temperature exceeds the safety threshold during the staged shutdown process, the shutdown process is terminated and the air flotation flow rate of the previous stage is restored.

[0016] Compared to related technologies, the single-phase immersion liquid cooling system and its control method based on air flotation fluid technology provided in this application have at least the following beneficial effects: By directionally injecting micron-sized bubbles into high heat flux density areas of the server using an air-float fluid enhancement module, the strong disturbance characteristics of the gas-liquid two-phase flow significantly increase the local fluid velocity, thereby effectively disrupting the laminar boundary layer of the high-viscosity insulating coolant on the chip surface. This greatly expands the heat transfer limit of a single chip. Under the same operating conditions, compared with traditional immersion liquid cooling solutions, the overall heat dissipation performance is significantly improved, the chip junction temperature is significantly reduced, and the problems of local heat accumulation and uneven heat dissipation caused by the poor fluidity of high-viscosity synthetic oil are solved.

[0017] The solution employs a multi-scenario adaptive control strategy, dynamically adjusting the baseline control parameters and the adjustment weights of various actuators (such as the air flotation module, circulating water pump, and CDU) based on different seasonal ambient temperatures and real-time heat load conditions. Through real-time calculation and online self-tuning of temperature deviations and their rates of change, the system can achieve extremely rapid response to sudden load changes and maintain the return liquid temperature within a very high temperature control accuracy range. This effectively overcomes the control delay and thermal hysteresis caused by the thermal conductivity characteristics of high-viscosity insulating coolants at the control logic level. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1a This is an architectural diagram of the secondary-side liquid-cooled cabinet and the secondary-side CDU of a single-phase immersion system according to an embodiment of this application; Figure 1bThis is a schematic diagram of the primary circulation module and constant pressure water supply device SP of a single-phase immersion system according to an embodiment of this application. Figure 1c This is a schematic diagram of the primary side outdoor cold source of a single-phase immersion system according to an embodiment of this application; Figure 2 This is a schematic diagram illustrating the various valves, sensors, and pumps in a single-phase immersion system architecture. Figure 3 This is a schematic diagram of a liquid-cooled cabinet based on the air-float fluid technology according to an embodiment of this application; Figure 4 This is an isometric side view of a liquid-cooled cabinet according to an embodiment of this application; Figure 5 This is an isometric side view of the liquid-cooled cabinet according to an embodiment of this application; Figure 6 This is a perspective view of a liquid-cooled cabinet without a server, according to an embodiment of this application. Figure 7 This is a schematic diagram of local micro-disturbance of fluid inside a liquid-cooled cabinet according to an embodiment of this application; Figure 8 This is a schematic diagram illustrating the implementation effect of a single-phase immersion liquid cooling system based on air flotation fluid technology according to an embodiment of this application; Figure 9 This is a schematic diagram of the control logic of the control method according to an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0020] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0021] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0022] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0023] Example 1

[0024] This embodiment provides a single-phase immersion liquid cooling system based on air flotation fluid technology, which aims to solve the problems of low heat dissipation efficiency and control lag when high-viscosity hydrocarbon or silicon-based synthetic oils are used to cool servers with high heat flux density.

[0025] Figure 1a This is an architectural diagram of the secondary-side liquid-cooled cabinet and the secondary-side CDU of a single-phase immersion system according to an embodiment of this application. Figure 1b This is a schematic diagram of the primary circulation module and constant pressure water supply device SP of a single-phase submersible system according to an embodiment of this application. Figure 1c This is a diagram illustrating the architecture of the primary-side outdoor cold source of a single-phase immersion system according to an embodiment of this application. Figure 1a , Figure 1b and Figure 1cAs shown, the system mainly includes a primary-side circulation module, a secondary-side cooling distribution unit (CDU), a liquid-cooled cabinet, and an air flotation fluid enhancement module in its overall hardware architecture. The modules and the control module are connected by pipelines to form a complete heat exchange loop. Figure 2 This is a schematic diagram illustrating the various valves, sensors, and pumps in a single-phase submerged system architecture.

[0026] like Figure 1a , Figure 1b and Figure 1c As shown, specifically, in the primary-side circulation module, the system is configured with an outdoor cold source and a primary-side variable frequency water pump. The outdoor cold source adopts a combined design, including a dry cooler (cold source 1), a cooling tower (cold source 2), and a chiller unit (cold source 3), and is equipped with a constant pressure water supply device to maintain stable water pressure. This multi-cold source configuration allows the control system to flexibly switch operating modes according to the outdoor wet-bulb temperature. For example, in low-temperature conditions in winter, only the dry cooler operates to utilize natural cold source heat dissipation; in transitional seasons, the dry cooler and cooling tower operate together; in extreme high-temperature conditions in summer, the chiller unit is turned on for auxiliary active cooling. In addition, a constant pressure water supply device is connected in parallel in the primary-side pipeline to compensate for the loss of circulating water and maintain stable primary-side water pressure.

[0027] Figure 3 This is a schematic diagram of a liquid-cooled cabinet based on the air-float fluid technology according to an embodiment of this application. Figure 4 This is an isometric side view of a liquid-cooled cabinet according to an embodiment of this application. Figure 5 This is an isometric side view of the liquid-cooled cabinet according to an embodiment of this application. Figure 3 , Figure 4 and Figure 5 As shown, the cabinet has an integrated sealed internal liquid pool for containing insulating coolant. This insulating coolant can be, but is not limited to, hydrocarbon synthetic oil (such as PAO series), silicone-based synthetic oil (such as dimethyl silicone oil 50cs series), or fluorinated liquid.

[0028] Furthermore, the cooling distribution unit (CDU) serves as a heat exchange hub connecting the primary and secondary sides, such as... Figure 1a , Figure 1b and Figure 1c As shown, the secondary-side piping connects to the liquid-cooled cabinet to form a closed-loop circuit. The CDU integrates a plate heat exchanger, a secondary-side variable frequency liquid pump, a pressure relief valve, manual valves, and an electric regulating valve. High-precision pressure and temperature sensors are installed on the CDU's inlet and outlet pipes to monitor pipeline parameters in real time and provide feedback to the control system.

[0029] In the secondary circulation, the insulating coolant, driven by a variable frequency pump inside the CDU, enters the liquid inlet rectification area at the bottom of the liquid-cooled cabinet via the supply pipeline. After rectification by the liquid distribution device, the low-temperature insulating coolant flows vertically upward through the server immersed in the inner liquid pool, absorbing the heat generated by the server's heat-generating components, thus raising its temperature. The insulating coolant carrying heat (and microbubbles generated by the air flotation module) rises to the top of the cabinet, where it passes through a gas-liquid separator to remove gas before flowing into the return liquid chamber. Subsequently, the high-temperature insulating coolant flows back to the CDU via the return liquid pipeline and enters the hot-side channel of the plate heat exchanger.

[0030] Furthermore, in the primary circulation, the primary circulation pump drives the cooling water to flow through the outdoor cold source for cooling before entering the cold-side channel of the plate heat exchanger inside the CDU. Finally, during the heat exchange process, the high-temperature insulating coolant and the low-temperature cooling water on both sides of the plate heat exchanger undergo non-contact heat transfer. After releasing heat, the insulating coolant cools down and is pumped back into the liquid-cooled cabinet for the next cycle; while the cooling water that has absorbed heat flows back to the outdoor cold source for heat dissipation, thus achieving optimized closed-loop heat dissipation for energy efficiency throughout the seasons and the entire process.

[0031] In this embodiment, the liquid-cooled cabinet is the core heat dissipation unit. Its internal assembly space adopts a universal design, which can adapt to the installation of servers of various specifications and support server nodes of different heights and physical forms. At the same time, by optimizing the internal flow field distribution and circulation capability, the heat dissipation capacity of the cabinet can be flexibly configured according to the power density of the assembled equipment, thereby meeting the diverse heat dissipation load requirements from conventional density to ultra-high density. In a specific embodiment, the liquid-cooled cabinet can be adapted to the assembly of 19-inch standard servers, supports mainstream specifications such as 2U / 4U / 8U, and can be expanded to meet the heat dissipation load of 100kW per cabinet.

[0032] Specifically, the liquid-cooled cabinet is precisely functionally partitioned in the vertical direction. For example... Figure 3 As shown, the bottom of the cabinet is designed with a first liquid chamber (i.e., the liquid inlet rectification area), and a liquid distribution device is installed between the first liquid chamber and the inner liquid pool above. As a fluid dynamics component, the core function of this liquid distribution device is to perform secondary rectification and pressure equalization on the insulating coolant that rapidly flows into the first liquid chamber from the liquid inlet, eliminate fluid dynamic pressure fluctuations, and ensure that the fluid can enter the inner liquid pool vertically upward with a uniform velocity field and a stable pressure field.

[0033] In the embodiments of this application, to adapt to different fluid characteristics and heat dissipation requirements, the liquid distribution device can be configured in various structural forms: First, it can adopt a pipe network distribution structure, which consists of multiple parallel branch pipes, with a series of precision nozzles or linear slits opened along the axial direction on the branch pipes, and achieves uniform flow through pipe network pressure balance; Second, it can adopt a labyrinth-type rectifier grid, which increases fluid resistance through internal guiding channels and forces uniform fluid diffusion using the pressure drop principle; Third, it can also adopt a microporous dielectric plate, such as a stainless steel sintered mesh plate or a porous ceramic plate, which utilizes its micron-level pore structure to achieve high-precision laminar flow output.

[0034] As a preferred implementation method, such as Figure 3 As shown, the liquid distribution device in this embodiment uses a liquid distribution plate. This liquid distribution plate is horizontally placed and sealed to the inner wall of the cabinet, separating the bottom first liquid chamber from the upper inner liquid pool. Several overflow holes are distributed on the liquid distribution plate. In its specific design, a non-uniform, on-demand distribution pattern is adopted. Specifically, the distribution density or diameter of the overflow holes is increased in the vertical projection area corresponding to high-power server chips (such as CPUs and GPUs), while they are sparsely distributed in areas without heat-generating components. Through this design, the flowing insulating coolant, after being expanded and buffered by the first liquid chamber, is forcibly regulated by the liquid distribution plate, forming a uniform, directional, and turbulent upward flow, efficiently and precisely flushing the internal servers immersed in the inner liquid pool.

[0035] More importantly, in order to overcome the flow resistance of high-viscosity fluids and enhance heat transfer, this embodiment has installed an air flotation fluid enhancement module in the mounting slot at the bottom of the cabinet. Figure 6 This is a perspective view of a liquid-cooled cabinet without servers according to an embodiment of this application. Figure 7 This is a schematic diagram of local micro-disturbances of the fluid inside the liquid-cooled cabinet according to an embodiment of this application, such as... Figure 6 and Figure 7 As shown, the module includes an air supply unit and an array-type air outlet assembly.

[0036] Specifically, the gas supply unit includes a gas generator, a pressure stabilizing tank, and a flow regulating valve assembly connected in sequence via pipelines. The gas generator is equipped with an oil and water removal filter to output clean gas; specifically, the purity of the clean gas is preferably greater than or equal to 99.99%. The pressure stabilizing tank has a volume set to greater than or equal to 10L to eliminate pressure pulsations during gas generator operation; the flow regulating valve assembly is electrically connected to the control module and is used to adjust the gas injection rate within a flow range of 5-20L / min / server.

[0037] Preferably, the array-type air outlet assembly is disposed in the mounting groove at the bottom of the liquid-cooled cabinet, and adopts a microporous aeration structure. Its material can be a microporous sintered material, specifically, the micropore diameter ranges from 0.1 to 100 μm. For example... Figure 6As shown, the array-type air outlet component projects onto the horizontal plane, covering the area directly below the server's heat-generating components (such as CPU or GPU cooling modules), thus generating a gas-liquid two-phase flow that directly acts on the chip surface. Furthermore, the air-float fluid enhancement module is detachably connected to the bottom of the liquid-cooled cabinet using a slot-type or sliding rail design to support independent maintenance without draining the insulating coolant.

[0038] In the specific execution process, when the air flotation function is activated, microbubbles are injected into the insulating coolant and rise rapidly under the drive of buoyancy, such as... Figure 7 As indicated by the middle arrow, a gas-liquid two-phase enhanced heat transfer flow field is formed. Utilizing the drainage effect and wake effect during the bubble's ascent, intense unsteady micro-perturbations are induced on the surface of the heating element, thereby disrupting the flow equilibrium of the high-viscosity fluid. Experiments have shown that, using this structure, local micro-perturbations can increase the fluid velocity from approximately 10 mm / s in natural convection to the order of 1000 mm / s, significantly disrupting the laminar boundary layer on the heat dissipation surface.

[0039] Figure 8 This is a schematic diagram illustrating the implementation effect of a single-phase immersion liquid cooling system based on air flotation fluid technology according to an embodiment of this application, such as... Figure 8 As shown, under liquid inlet conditions of 40°C, this technical solution reduces the chip junction temperature by about 13%, and the heat dissipation capacity of a single chip exceeds 1000W.

[0040] Optionally, to further constrain the flow field and improve heat exchange efficiency, multiple movable fluid baffles are installed in the heat exchange zone of the inner liquid pool. Combined with... Figure 3 and Figure 4 As shown, the partition can be mounted on the Tank via lugs, located between adjacent servers, to divide the inner liquid pool into multiple independent vertical fluid domains.

[0041] Specifically, the relative position of the fluid baffle can be flexibly adjusted according to the U-number (thickness) of the high-density server. This baffle allows the gas-liquid mixture to flow vertically upwards along the server surface, preventing air bubbles from escaping to ineffective areas. Meanwhile, as... Figure 3 As shown, for the non-heating idle space in the inner liquid pool, a solid filling block can be set to occupy the ineffective liquid level volume and guide the insulating coolant to flow to the heating component. The material of the solid filling block can be, but is not limited to, bakelite and stainless steel.

[0042] In addition, to ensure control accuracy and system reliability, this system has constructed a fully redundant sensing and detection network.

[0043] Table 1 is an example table of sensor redundancy configurations according to embodiments of this application:

[0044] In the system's sensing and detection layer, key monitoring points such as the cabinet inlet (Tin), outlet (Tout), ambient temperature (Ta), air flotation flow rate (Qf), and CDU circulating pressure (Pc) are all equipped with paired main and backup sensors. Taking the outlet temperature (Tout) as an example, two sets of immersion-type PT100 temperature sensors are used. The main sensor is located in the center of the liquid path, and the backup sensor is located on the same axis 10cm away from the main sensor.

[0045] Furthermore, the control module integrates a data verification and automatic calibration unit, with the following specific execution logic: Firstly, real-time fault diagnosis and redundancy switching: The control module collects data from the primary and backup sensors in real time and calculates their absolute deviation values. When the deviation between the primary and backup return liquid temperature is greater than 0.5℃, the flow rate deviation is greater than 5%, or the pressure deviation is greater than 0.03MPa, and any of these deviations lasts for more than 3 seconds, the system determines that the primary sensor has experienced data drift or a physical fault. At this time, the control module performs a millisecond-level switching action, calling the backup sensor data as a feedback signal and recording the fault code. At the moment of switching, the control module uses the moving average of the three sampling cycles before the switching to perform data smoothing processing to eliminate the impact of signal jumps on PID calculations.

[0046] Secondly, the periodic self-calibration logic: When the system is in a quasi-steady-state operation phase (e.g., when the load fluctuation rate is ≤5% and lasts for 10 minutes), the self-calibration unit performs online comparison between the main and backup sensors. If the deviation is within the allowable range (temperature deviation ≤0.3℃, flow or pressure deviation ≤3%), the system uses the main sensor reading as a reference to calculate the real-time offset of the backup sensor and stores this offset in non-volatile memory for dynamic compensation of the backup data.

[0047] This system uses an air-float fluid enhancement module to directionally inject micron-sized bubbles into high heat flux density areas of the server. Utilizing the strong disturbance characteristics of the gas-liquid two-phase flow, it significantly increases the local fluid velocity, effectively disrupting the laminar boundary layer of the high-viscosity insulating coolant on the chip surface. This greatly expands the heat transfer limit of a single chip. Under the same operating conditions, compared to traditional immersion liquid cooling solutions, its overall heat dissipation performance is significantly improved, and the chip junction temperature is significantly reduced. It effectively solves the problem of localized heat accumulation and uneven heat dissipation caused by the poor fluidity of high-viscosity synthetic oil.

[0048] Furthermore, the system constructs a redundant detection network based on primary and backup dual-channel sensors, combined with periodic automatic calibration logic and a real-time fault switching strategy, ensuring the accuracy of key data acquisition such as temperature, flow rate, and pressure. This effectively avoids system misjudgments caused by drift or failure of a single sensor, significantly improving the long-term operational stability and safety of the system.

[0049] Example 2

[0050] This embodiment provides a control method for the aforementioned single-phase immersion liquid cooling system based on air flotation fluid technology. This method is executed by a control module (such as a PLC control unit). Figure 9 This is a schematic diagram of the control logic of the control method according to an embodiment of this application, such as... Figure 9 As shown, this method aims to solve the problems of large temperature fluctuations and slow response caused by the thermal hysteresis of high-viscosity insulating coolants. Through a closed-loop logic of "operating status acquisition—scene recognition—parameter self-tuning—execution coordination—closed-loop correction," precise temperature control under all operating conditions is achieved. Specifically, it includes the following steps: S1, Obtaining Running Status and Verifying Data First, real-time operating data of the system is collected through a redundant sensor network, including ambient temperature (Ta), cabinet return liquid temperature (Tout), server load rate (L), air flotation flow rate (Qf), and CDU circulating pressure (Pc). During the acquisition process, data verification and automatic calibration logic are executed. Through dual-path redundant acquisition and dynamic calibration mechanisms, data errors caused by sensor drift are effectively eliminated, ensuring the absolute accuracy of the input source for the control algorithm. This provides a reliable data foundation for subsequent accurate calculations and avoids system malfunctions caused by missampling.

[0051] S2, Scenario Determination and Baseline Parameter Calling like Figure 9 As shown, based on the collected ambient temperature (Ta) and server load rate (L), the control module divides the current operating state into several typical scenarios.

[0052] Specifically, the season is determined first: winter (Ta≤10℃), transition season (10℃<Ta≤25℃), summer (Ta>25℃); the load is determined second: low load (L≤60%), medium load (60%<L≤90%), high load (90%<L≤100%), extreme load (L=100% and Tout exceeds the limit).

[0053] After determining the scenario, the system calls the corresponding PID baseline parameters (Kp, Ki, Kd) and the weight allocation coefficients (α, β, γ) of the actuator. To prevent repeated switching at critical points, the system introduces a hysteresis control mechanism: for example, the threshold for switching from low load to medium load is set to L≥65%, while the cutback threshold is set to L≤55%.

[0054] Table 2 is an example table of PID baseline parameters and weight allocation for different scenarios according to embodiments of this application:

[0055] This step utilizes multi-dimensional scenario segmentation and hysteresis control strategies to achieve precise matching between the control strategy and the external environment and internal load. This avoids the rigidity of control caused by a one-size-fits-all approach and prevents system oscillations caused by operating condition fluctuations, thereby significantly improving control stability.

[0056] S3, PID parameter self-tuning and adjustment calculation Based on the baseline parameters, the system dynamically corrects the PID parameters at the millisecond level based on the deviation (e) between the return liquid temperature Tout and the target threshold and its rate of change (de / dt).

[0057] Table 3 is an example table of parameter self-tuning rules according to embodiments of this application:

[0058] Specifically, when a large deviation (e > 3℃) is detected, Kp is increased by about 20%-50% and Ki is decreased by about 30% to accelerate the response; when a steady-state small deviation (0.5℃ < e ≤ 1℃, no fluctuation) is detected, Kp is decreased by about 20% and Ki is increased by about 50% to eliminate steady-state error; when a sudden load change is detected (e.g., L increases by ≥ 30% within 10s), Ki is kept unchanged, and Kp (about 30%) and Kd (about 40%) are significantly increased to offset thermal hysteresis using the predictive ability of the differential term.

[0059] In one specific implementation, to address the poor fluidity and significant lag in thermal sensing of high-viscosity synthetic oils at low temperatures, this embodiment also incorporates an adaptive compensation mechanism based on fluid physical properties. The control module has a pre-installed viscosity-temperature mapping model for this type of coolant, and uses redundant sensors to calculate fluid dynamics in real time.

[0060] When the coolant is detected to be in a high viscosity range, the system does not passively respond to the return temperature. Instead, it increases the differential coefficient Kd in the PID algorithm to enhance the system's sensitivity to small temperature fluctuation trends. This changes the control logic from reactive adjustment to trend warning, effectively offsetting the time constant delay caused by slow flow rate by utilizing the proactive adjustment characteristics of the differential term.

[0061] Furthermore, considering the slow heat exchange and persistent deviations under high viscosity conditions, the integral time constant Ti is synchronously extended to weaken the intensity of the integral action. This prevents excessive accumulation of the integral term during the lag period (integral saturation), thereby avoiding significant temperature overshoot oscillations caused by excessive command after fluid flow resumes. Through this compensation mechanism, the system achieves smooth switching between different viscosity states, ensuring a deep coupling between control accuracy and physical response characteristics.

[0062] Furthermore, the total regulation demand ΔU is calculated based on the corrected PID parameters and then converted into physical regulation quantities. Specifically, the total regulation demand is converted into air flotation flow rate regulation, primary pump frequency regulation, and CDU circulation pressure regulation using the following control model:

[0063] in, For the total adjustment demand, It is the air flotation flow rate adjustment component. This is the frequency regulation component of the primary water pump. This is the CDU cycle pressure regulation component; These are the dynamic weighting coefficients for the air flotation fluid enhancement module, the primary side circulation module, and the cooling distribution device, respectively.

[0064] Furthermore, the dimensionless adjustment quantity output is converted into physical parameters that the actuator can recognize, and the mapping relationship is as follows: Table 4 is an example table of the mapping between adjustment amount and physical parameters according to embodiments of this application:

[0065] This step overcomes the thermal hysteresis effect of high-viscosity insulating coolant by using PID parameter self-tuning and dynamic weight allocation based on deviation characteristics, achieving millisecond-level algorithm response. This ensures the system can quickly track load changes while effectively suppressing temperature overshoot.

[0066] To address the thermal hysteresis characteristics of high-viscosity hydrocarbon or silicon-based synthetic oils under large load fluctuations, this embodiment demonstrates the system's dynamic adaptive capability through the parameter evolution and physical action process under the following specific operating conditions. During the adjustment process, the adjustment step size of all physical variables follows a smooth transition principle to avoid fluid impact and maintain flow field stability. The following, combined with... Figure 9 The logical flow shown details the execution details in each typical scenario.

[0067] 1. Winter operating conditions (Ta≤10℃, full load rate) 1.1 Low load in winter (L≤60%) Triggering conditions: Ta=5℃, L=50%, Tout=38℃ (threshold 37℃, e=1℃), de / dt=0.2℃ / min; The parameter tuning process is as follows: For steady-state small deviations (0.5℃ < e = 1℃ ≤ 1℃), de / dt = 0.2℃ / min (no fluctuations / abrupt changes), the matching scheme is "parameter self-tuning rule," and the corresponding adjustment strategy is: Kp reduced by 20%, Ki increased by 50%, and Kd reduced by 10%. Calculation process (adjustment step size 0.05, results rounded to two decimal places): Final Kp value = Baseline value × (1 - Downward adjustment ratio) = 0.5 × (1 - 20%) = 0.5 × 0.8 = 0.40 (in line with the adjustment range of 0.5~2.0); Final Ki value = Baseline value × (1 + upward adjustment ratio) = 0.01 × (1 + 50%) = 0.01 × 1.5 = 0.015 (in line with the adjustment range of 0.01~0.1). Final Kd value = Base value × (1 - Downward adjustment ratio) = 0.1 × (1 - 10%) = 0.1 × 0.9 = 0.09 (in line with the adjustment range of 0.1~0.8).

[0068] Adjustment execution: a. Weight allocation: α=0, β=0.5, γ=0.5; b. Total regulation demand ΔU=4→ΔUp=2 (water pump), ΔUc=2 (CDU); c. Physical actions: Water pump frequency +4Hz (25→29Hz), CDU pressure +0.1MPa (0.3→0.4MPa), air flotation standby; CDU heat exchanger power reduced to 30%, utilizing natural cold source for energy saving; Closed-loop correction: Tout is collected every 500ms. If e drops to 0.5℃, Kp is adjusted back to 0.45, the pump frequency is adjusted back to 27Hz, and the CDU pressure is adjusted back to 0.35MPa.

[0069] 1.2 Winter load (60% < L ≤ 90%) Triggering conditions: Ta=7℃, L=75%, Tout=45℃ (threshold 43℃, e=2℃), de / dt=0.4℃ / min (no fluctuations / abrupt changes, consistent with steady-state small deviation characteristics); Parameter tuning: Baseline Kp=0.8 → down by 20% (0.64), Ki=0.03 → up by 50% (0.045), Kd=0.2 → down by 10% (0.18) (strictly matching the user-given steady-state small deviation adjustment strategy); Adjustment execution: Weighting: α=0.1, β=0.4, γ=0.5; Total regulation demand ΔU=6→ΔUf=0.6 (air flotation), ΔUp=2.4 (water pump), ΔUc=3 (CDU); Physical parameters: Air flotation flow rate +1.2L / min (from 5→6.2L / min, smoothly increasing within 0.6s), water pump frequency +4.8Hz (32→36.8Hz), CDU pressure +0.15MPa (0.35→0.5MPa), CDU heat exchanger power 60%; Closed-loop correction: Tout is collected every 500ms. If e drops to 0.8℃ (still meets the steady-state small deviation) → Kp is maintained at 0.64, the pump frequency is adjusted back to 35Hz, and the CDU pressure is adjusted back to 0.45MPa.

[0070] 1.3 High load in winter (90% < L ≤ 100%) Triggering conditions: Ta=8℃, L=95%, Tout=52℃ (threshold 50℃, e=2℃), de / dt=0.7℃ / min; Parameter tuning: Baseline Kp=1.2 → Increase by 10% (1.32), Ki=0.05 → Decrease by 20% (0.04), Kd=0.4 → Increase by 30% (0.52); Adjustment execution: a. Weight allocation: α=0.4, β=0.4, γ=0.2; b. Total regulation demand ΔU=8→ΔUf=3.2 (air flotation), ΔUp=3.2 (water pump), ΔUc=1.6 (CDU); c. Physical action: Air flotation flow rate +6.4L / min (from 8→14.4L / min, smoothly increased within 3.2s), water pump frequency +6.4Hz (40→46.4Hz), CDU pressure +0.08MPa (0.5→0.58MPa), CDU heat exchanger operates at full power; Closed-loop correction: Data is collected every 300ms. If Tout drops to 48℃, Kp is adjusted back to 1.2, the air flotation flow rate drops to 12L / min, and the water pump frequency is adjusted back to 43Hz.

[0071] 1.4 Winter extreme load (L=100%≥10min, Tout>50℃) Triggering conditions: Ta=10℃, L=100%, Tout=53℃ (threshold 50℃, e=3℃), de / dt=0.8℃ / min, Tj=86℃; Parameter tuning: Baseline Kp=1.8 → Increase by 30% (2.34), Ki=0.02 → Decrease by 30% (0.014), Kd=0.6 → Increase by 20% (0.72); Adjustment execution: a. Weight allocation: α=0.6, β=0.3, γ=0.1; b. Total regulation demand ΔU=10→ΔUf=6 (air flotation), ΔUp=3 (water pump), ΔUc=1 (CDU); c. Physical action: Air flotation flow rate +12L / min (from 10→20L / min, smoothly increased within 6s), water pump frequency +6Hz (45→51Hz→cut off to 50Hz, safety boundary), CDU pressure +0.05MPa (0.8→0.85MPa). Closed-loop correction: Data is collected every 200ms. If Tout drops to 48℃ and Tj≤80℃, Kp is adjusted back to 1.8, the air flotation flow rate is reduced to 15L / min, and the water pump frequency is adjusted back to 45Hz.

[0072] 2. Transitional season operating conditions (10℃<Ta≤25℃, full load rate) 2.1 Low load during the transition season (L≤60%) Triggering conditions: Ta=15℃, L=55%, Tout=41℃ (threshold 40℃, e=1℃), de / dt=0.2℃ / min (meets steady-state small deviation characteristics); Parameter tuning: Baseline Kp=0.6 → down by 20% (0.48), Ki=0.02 → up by 50% (0.03), Kd=0.15 → down by 10% (0.135); Adjustment execution: Weighting: α=0, β=0.4, γ=0.6; Total regulation demand ΔU=3→ΔUp=1.2 (water pump), ΔUc=1.8 (CDU); Physical actions: Water pump frequency +2.4Hz (28→30.4Hz), CDU pressure +0.09MPa (0.3→0.39MPa), air flotation standby; CDU heat exchanger power 40%, utilizing ambient cold source for energy saving; Closed-loop correction: If Tout drops to 39.5℃, Kp will be adjusted back to 0.55, the pump frequency will be adjusted back to 29Hz, and the CDU pressure will be adjusted back to 0.35MPa.

[0073] 2.2 Transitional load (60% < L ≤ 90%) Triggering conditions: Ta=20℃, L=80%, Tout repeats between 43℃ and 45℃ (|de / dt|=±0.6℃ / min). Parameter tuning: Baseline Kp=1.0 → down by 10% (0.9), Ki=0.06 → down by 20% (0.048), Kd=0.3 → up by 30% (0.39); Adjustment execution: a. Weight allocation: α=0.1, β=0.4, γ=0.5; b. Total regulation demand ΔU=5→ΔUf=0.5 (air flotation), ΔUp=2 (water pump), ΔUc=2.5 (CDU); c. Physical operation: The air flotation flow rate is maintained at 8L / min (without frequent start-stop), the water pump frequency is finely adjusted at 35±2Hz, and the CDU pressure is finely adjusted at 0.4±0.02MPa; Closed-loop correction: After the Tout fluctuation range narrows to ±0.7℃, Kd pulls back to 0.3, restoring the normal adjustment parameters.

[0074] 2.3 High load during the transition season (90% < L ≤ 100%) Triggering conditions: Ta=22℃, L=92%, Tout=49℃ (threshold 48℃, e=1℃), de / dt=0.3℃ / min; Parameter tuning: Baseline Kp=1.5 → remain unchanged (1.5), Ki=0.04 → increase by 20% (0.048), Kd=0.5 → increase by 10% (0.55); Adjustment execution: Weighting: α=0.5, β=0.35, γ=0.15; Total regulation demand ΔU=7→ΔUf=3.5 (air flotation), ΔUp=2.45 (water pump), ΔUc=1.05 (CDU); Physical action: Air flotation flow rate +7L / min (from 8→15L / min, smoothly increased within 3.5s), water pump frequency +4.9Hz (40→44.9Hz), CDU pressure +0.0525MPa (0.6→0.6525MPa). Closed-loop correction: If Tout drops to 47℃, Kp will be adjusted back to 1.4, the air flotation flow rate will be reduced to 12L / min, and the pump frequency will be adjusted back to 42Hz.

[0075] 2.4 Transitional extreme load (L=100%≥10min, Tj≥85℃) Triggering conditions: Ta=25℃, L=100%, Tout=51℃ (threshold 48℃, e=3℃), de / dt=0.9℃ / min, Tj=87℃; Parameter tuning: Baseline Kp=1.9 → Increase by 20% (2.28), Ki=0.02 → Decrease by 30% (0.014), Kd=0.7 → Increase by 20% (0.84); Adjustment execution: a. Weight allocation: α=0.6, β=0.3, γ=0.1; b. Total regulation demand ΔU=12→ΔUf=7.2 (air flotation), ΔUp=3.6 (water pump), ΔUc=1.2 (CDU); c. Physical action: The air flotation flow rate is increased to 20L / min (from 12 to 20L / min, smoothly increasing within 4 seconds), the water pump frequency is increased by 7.2Hz (42 to 49.2Hz), the CDU pressure is increased by 0.06MPa (0.7 to 0.76MPa), and the cooling tower operates at full power; Closed-loop correction: If Tj drops to 80℃, Kp will be adjusted back to 1.9, the air flotation flow rate will be reduced to 16L / min, and the pump frequency will be adjusted back to 46Hz.

[0076] 3. Summer operating conditions (Ta > 25℃, full load rate) 3.1 Low load in summer (L≤60%) Triggering conditions: Ta=28℃, L=50%, Tout=44℃ (threshold 43℃, e=1℃), de / dt=0.2℃ / min (meets steady-state small deviation characteristics); Parameter tuning: Baseline Kp=0.7 → down by 20% (0.56), Ki=0.02 → up by 50% (0.03), Kd=0.2 → down by 10% (0.18); Adjustment execution: Weighting: α=0.1, β=0.4, γ=0.5; Total regulation demand ΔU=4→ΔUf=0.4 (air flotation), ΔUp=1.6 (water pump), ΔUc=2 (CDU); Physical actions: Air flotation flow rate +0.8L / min (from 3→3.8L / min, smoothly increasing within 0.4s), water pump frequency +3.2Hz (30→33.2Hz), CDU pressure +0.1MPa (0.35→0.45MPa), cooling tower low-frequency operation; Closed-loop correction: If Tout drops to 42.5℃ → Kp returns to 0.65, the air flotation flow rate drops to 3L / min, and the water pump frequency returns to 31Hz.

[0077] 3.2 Summer load (60% < L ≤ 90%) Triggering conditions: Ta=30℃, L=85%, Tout=47℃ (threshold 45℃, e=2℃), de / dt=0.6℃ / min; Parameter tuning: Baseline Kp=1.1 → Increase by 20% (1.32), Ki=0.05 → Decrease by 10% (0.045), Kd=0.4 → Increase by 20% (0.48); Adjustment execution: a. Weight allocation: α=0.2, β=0.4, γ=0.4; b. Total regulation demand ΔU=7→ΔUf=1.4 (air flotation), ΔUp=2.8 (water pump), ΔUc=2.8 (CDU); c. Physical action: Air flotation flow rate +2.8L / min (from 5→7.8L / min, smoothly increasing within 1.4s), water pump frequency +5.6Hz (35→40.6Hz), CDU pressure +0.14MPa (0.4→0.54MPa), cooling tower in normal operation; Closed-loop correction: If Tout drops to 44℃, Kp will be adjusted back to 1.1, the air flotation flow rate will be reduced to 6L / min, and the pump frequency will be adjusted back to 38Hz.

[0078] 3.3 High load in summer (90% < L ≤ 100%) Triggering conditions: Ta=32℃, L=98%, Tout=50℃ (threshold 50℃, e=0℃), de / dt=0.3℃ / min; Parameter tuning: Baseline Kp=1.6 → remain unchanged (1.6), Ki=0.03 → increase by 20% (0.036), Kd=0.6 → remain unchanged (0.6); Adjustment execution: a. Weight allocation: α=0.5, β=0.35, γ=0.15; b. Total regulation demand ΔU=4→ΔUf=2 (air flotation), ΔUp=1.4 (water pump), ΔUc=0.6 (CDU); c. Physical actions: Maintain air flotation flow rate at 15L / min, water pump frequency +2.8Hz (45→47.8Hz), CDU pressure +0.03MPa (0.6→0.63MPa), and maintain stable temperature; Closed-loop correction: If Tout rises to 51℃, Kp is adjusted to 1.8, the air flotation flow rate is increased to 18L / min, and the water pump frequency is increased to 49Hz.

[0079] 3.4 Summer extreme load (L=100%≥10min, Tout>50℃ or Tj≥85℃) Triggering conditions: Ta=35℃, L=100%, Tout=55℃ (threshold 50℃, e=5℃), de / dt=0.9℃ / min, Tj=88℃; Parameter tuning: Baseline Kp=2.0 → Increase by 50% (3.0), Ki=0.02 → Decrease by 30% (0.014), Kd=0.8 → Increase by 20% (0.96); Adjustment execution: a. Weight allocation: α=0.6, β=0.3, γ=0.1; b. Total regulation demand ΔU=12→ΔUf=7.2 (air flotation), ΔUp=3.6 (water pump), ΔUc=1.2 (CDU); c. Physical actions: The air flotation flow rate is increased to 20L / min (from 12 to 20L / min, smoothly increasing within 4 seconds), the water pump frequency is increased to 50Hz (maximum flow rate 20m³ / h), the CDU pressure is increased to 1.0MPa (rated upper limit), and the cooling tower / dry cooler operates at full power; an audible and visual alarm is triggered, and the maximum heat dissipation parameters are locked. Closed-loop correction: Data is collected every 100ms. If Tout drops to 48℃ and Tj≤80℃, Kp will be reverted to 2.0, the air flotation flow rate will be reduced to 15L / min, the water pump frequency will be reduced to 45Hz, and the CDU pressure will be reduced to 0.8MPa. If Tj≥90℃, the server frequency reduction protection will be activated.

[0080] The above embodiments, through the refined switching of twelve scenarios, achieve a balance between high reliability and low energy consumption. They cover a complete logical closed loop from data acquisition and scenario recognition to dynamic execution, providing data support and solution verification for the stable operation of a single-phase immersion liquid cooling system based on air flotation fluid technology under complex operating conditions.

[0081] Step S4: Execution control and air flotation coordination The control module sends the calculated physical parameters to each actuator in parallel. For the core air flotation fluid enhancement module, there is independent start / stop logic, specifically: 1) Enable logic: Force enable when L>90%, Tout>50℃ or chip junction temperature Tj≥85℃.

[0082] Additionally, if L increases by ≥30% within 10 seconds (load sudden change) or Tout change rate is ≥ the corresponding operating condition threshold (0.3℃ / min in winter, 0.4℃ / min in transition, and 0.5℃ / min in summer). Specifically, when the algorithm predicts that e≥2℃ and de / dt≥0.5℃ / min, it triggers "prediction start" 0.5s in advance and simultaneously increases the Kd value by 30%.

[0083] 2) Shutdown Logic: This system adopts a tiered, decreasing strategy and sets differentiated trigger conditions based on real-time scenarios: Under normal circumstances, the return liquid temperature Tout must be ≤ the corresponding operating condition threshold -2℃, de / dt ≤ 0.2℃ / min, and stable for 3min. Under extreme circumstances, Tout ≤ 48℃ and Tj ≤ 80℃, de / dt ≤ 0.2℃ / min, and stable for 5min.

[0084] After the above conditions are met, the flow control valve group reduces the air flotation flow rate in steps of 5L / min / 30s. The control module simultaneously performs PID parameter retuning, that is, reducing Kp (proportional coefficient) by 40% and increasing Ki (integral coefficient) by 30%, to suppress thermal fluctuations caused by the reduction of fluid disturbance through parameter compensation.

[0085] During the shutdown process, if a Tout rise of ≥3℃ or de / dt ≥0.4℃ / min is detected, the system immediately pauses the current shutdown process and restores the initial gas injection flow rate. Under extreme conditions, the shutdown logic has a lower priority than the over-temperature alarm. If Tj ≥88℃ is detected, the system directly terminates the shutdown command and switches the air flotation flow rate to the maximum range to perform forced cooling.

[0086] This step utilizes the predictive start-up and phased shutdown strategy of the air flotation module to achieve on-demand release and smooth recovery of heat dissipation capacity. This provides explosive heat dissipation capacity under extreme operating conditions while avoiding energy waste and flow field impact caused by frequent start-stop, thus significantly improving the overall energy efficiency of the system.

[0087] Step S5: Closed-loop correction and smooth scene transition The system continuously collects feedback data and performs closed-loop correction at a high frequency (normally 100ms / time, extreme 50ms / time). During scene switching or parameter adjustments, the system executes a smooth parameter transition logic, with all PID parameters and actuator commands gradually changing according to a preset step size (e.g., PID parameter 0.05 step / 100ms, flow rate 2L / min·s). For example: Mid-season load → High load → Kp changes from 0.8 to 1.2 (transition completed in 800ms), air flotation flow rate changes from 8 to 12L / min (transition completed in 2s). Specifically, the system also avoids repeated scene switching caused by small parameter changes by setting a scene switching hysteresis threshold. Table 5 is an example table of scene switching hysteresis thresholds according to embodiments of this application:

[0088] Furthermore, the system also fine-tunes the previously calculated adjustment amount based on Tout's real-time feedback. For example, in the low-load mode during winter, if the deviation is eliminated, the water pump frequency is adjusted back and the air flotation is kept off to maximize the use of natural cold source.

[0089] To address the slow control response caused by thermal hysteresis of high-viscosity coolant and to balance the computing load on the controller, this embodiment employs a dynamic data cycle adaptation strategy.

[0090] When the system is in a steady state or under normal load variations, the control module executes a standard cycle configuration: the sensor data acquisition interval is set to 100ms, and the PID algorithm parameter self-tuning cycle is set to 500ms. This frequency maintains a temperature control accuracy of ±0.8℃ while avoiding excessive consumption of PLC processing resources.

[0091] When the system detects a sudden load change (e.g., a load rate increase of ≥30% within 10 seconds) or enters extreme operating conditions (e.g., chip junction temperature ≥85℃ or liquid return temperature exceeding 50℃), it triggers a rapid response mode. At this time, the data acquisition interval is shortened to 50ms, and the PID self-tuning cycle is compressed to 200ms. By doubling the sampling and calculation frequency, the system captures minute temperature changes (de / dt), and with the rapid intervention of the air flotation module, the system's response delay is reduced from 3s with traditional fixed parameters to less than 0.6s, effectively preventing instantaneous heat buildup on the chip.

[0092] Table 6 is an example table of dynamic adaptation parameters for data acquisition and control cycles according to embodiments of this application:

[0093] Through the above steps, a multi-scenario adaptive control strategy is adopted to dynamically adjust the control parameter benchmark and the adjustment weights of each actuator (such as the air flotation module, circulating water pump, and CDU) according to different seasonal ambient temperatures and real-time heat load conditions. By real-time calculation and online self-tuning of temperature deviation and its rate of change, the system can achieve rapid response to sudden load changes and maintain the return liquid temperature within a very high temperature control accuracy range. This effectively overcomes the control delay and thermal hysteresis caused by the thermal conductivity characteristics of high-viscosity insulating coolant at the control logic level. In addition, the weights of each actuator are intelligently allocated according to the coupling relationship between ambient temperature and load rate. In winter and low-load conditions, natural cold sources are prioritized and the power consumption of air flotation and water pumps is reduced; in extreme summer conditions, air flotation enhancement and active cooling are combined to achieve rapid cooling. This strategy significantly reduces the PUE value while meeting the heat dissipation requirements of high heat flux density, achieving the best balance between heat dissipation efficiency and operating energy consumption.

[0094] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0095] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A single-phase immersion liquid cooling system based on air flotation fluid technology, characterized in that, The system includes: The liquid-cooled cabinet includes a first liquid chamber and an inner liquid pool for immersing and accommodating servers. A liquid distribution device is provided between the first liquid chamber and the inner liquid pool, and the liquid distribution device has a plurality of overflow holes. The liquid-cooled cabinet is vertically divided into a liquid inlet rectification area, a heat exchange area and a gas-liquid separation area. An air-float fluid enhancement module, whose output end passes through the first liquid cavity and extends to the inner liquid pool, is used to inject microbubbles into the insulating coolant in the liquid-cooled cabinet to form an air-liquid enhanced heat transfer flow field on the surface of the server heat dissipation module. The air-float fluid enhancement module uses an array-type air outlet component with a microporous aeration structure to generate the microbubbles. The projection area of ​​the air outlet holes of the array-type air outlet component covers the high heat flux density area of ​​each server. By utilizing the liquid drainage effect and wake effect during the rise of the microbubbles, unsteady micro-disturbances are induced on the surface of the server heat dissipation module to disrupt the laminar boundary layer on the surface of the server heat dissipation module and break the flow balance of the high viscosity fluid. The cooling distribution device is connected to the liquid-cooled cabinet via a secondary side pipeline to form a secondary side closed loop. Its input and output ends are respectively connected to the liquid inlet and outlet of the liquid-cooled cabinet, which is used to drive the circulation of the insulating coolant and perform heat exchange. The gas-liquid separation zone is located at the top of the liquid-cooled cabinet. It is connected to the heat exchange zone via a gas-liquid separation filter and is also connected to the cooling distribution device via a secondary side pipeline. The gas-liquid separation filter is used to intercept and eliminate microbubbles rising to the liquid surface to prevent microbubbles from entering the cooling distribution device. The primary-side circulation module is connected to the cooling distribution device through a primary-side pipeline to form a primary-side heat exchange loop, which is used to provide cooling water with adjustable flow rate to the cooling distribution device; The control module is communicatively connected to the air flotation fluid enhancement module, the cooling distribution device, and the primary side circulation module, respectively. The control module is configured to execute a system adaptive control method, wherein the system adaptive control method dynamically adjusts the control parameter reference and the adjustment weights of the air flotation fluid enhancement module, the cooling distribution device, and the primary side circulation module according to different seasonal ambient temperatures and real-time heat load conditions.

2. The system according to claim 1, characterized in that, The air flotation fluid enhancement module includes an air source supply unit and an array-type air outlet assembly. The gas supply unit includes a gas generator, a pressure stabilizing tank, and a flow regulating valve group, used to output clean gas with adjustable pressure and flow rate; The array-type air outlet assembly is connected to the air supply unit via a gas delivery pipeline and is located at the bottom of the liquid cooling cabinet.

3. The system according to claim 2, characterized in that, The liquid inlet rectification area is located at the bottom of the liquid cooling cabinet, and the liquid distribution device is installed inside it. The array-type air outlet assembly is integrated below the liquid distribution device or is installed through the liquid distribution device. The heat exchange zone includes multiple movable fluid baffles for dividing the inner liquid pool into multiple independent vertical fluid domains for individually accommodating servers, thereby constraining the rising path of the microbubbles.

4. The system according to claim 3, characterized in that, The fluid baffle is disposed between adjacent servers in the heat exchange zone and is used to adjust the cross-sectional area of ​​the vertical fluid domain according to the physical dimensions of the server, so as to constrain the insulating coolant to flow vertically upward along the cross-section of the server. A filling block is also provided in the non-heating area of ​​the inner liquid pool. The filling block is a solid block that does not chemically react with the insulating coolant and is used to occupy the unused space in the inner liquid pool to guide the insulating coolant to the heat-generating parts of the server.

5. The system according to claim 1, characterized in that, The system further includes a sensor network electrically connected to the control module, used to collect system operating status data. The sensing network adopts a redundant architecture, including a main sensor group and a backup sensor group. The main sensor group and the backup sensor group are configured in pairs to monitor the return liquid temperature of the liquid-cooled cabinet, the server power consumption, and the external ambient temperature, respectively. The control module is configured to execute data verification logic to compare the collected data of the main sensor group and the backup sensor group in real time, and to execute fault switching control logic when the deviation between the two sets of data exceeds a safety threshold.

6. A control method, characterized in that, The method is applied to the single-phase immersion liquid cooling system based on air flotation fluid technology as described in any one of claims 1 to 5, and is executed by the control module, comprising the following steps: Acquire system operating status data, including ambient temperature, server load rate, return liquid temperature and its rate of change; Based on the ambient temperature and the server load rate, determine the current operating scenario and call the PID baseline parameters and weight allocation coefficients corresponding to the current operating scenario; Based on the deviation between the return liquid temperature and the target threshold and its rate of change, the PID baseline parameters are corrected, and the total adjustment requirement is calculated using the corrected PID parameters. Based on the weighting coefficients, the total adjustment demand is calculated into air flotation flow rate adjustment, primary side water pump frequency adjustment, and CDU circulation pressure adjustment. According to the calculation results, control commands are sent in parallel to the air flotation fluid enhancement module, the primary side circulation module, and the cooling distribution device.

7. The method according to claim 6, characterized in that, The method further includes: The control module pre-sets a mapping strategy between different operating scenarios and the weight allocation coefficients; The weight allocation coefficients are dynamically assigned based on the heat dissipation requirements of the operating scenario: In operating scenarios where the heat dissipation demand is determined to be greater than the first preset threshold or an extreme load, the weight coefficient of the air flotation fluid enhancement module is adjusted to be greater than the weight coefficients of the primary side circulation module and the cooling distribution device, so as to give priority to the use of air flotation fluid to enhance heat exchange efficiency. In operating scenarios where the load is determined to be below the second preset threshold or where energy saving is prioritized, the weight coefficient of the air flotation fluid enhancement module is reduced, and the weight coefficient of the primary side circulation module or the cooling distribution device is increased, so as to give priority to the use of natural cold sources or low power consumption operating modes.

8. The method according to claim 6, characterized in that, The method also includes a parameter correction step for specific operating conditions: When the deviation between the return liquid temperature and the target threshold is detected to be positive and greater than the first preset threshold, the proportional coefficient of the PID algorithm is increased and the integral coefficient is decreased to speed up the response. When the deviation is detected to be negative and the absolute value is greater than the second preset threshold, the proportional coefficient is decreased and the integral coefficient is increased to suppress overshoot oscillation. When the server load rate is detected to rise beyond the mutation threshold within a preset time window, the integral coefficient is kept unchanged, and the proportional coefficient and derivative coefficient are simultaneously increased to enhance the ability to predict and compensate for temperature mutations. The real-time dynamic viscosity of the insulating coolant is calculated based on the return temperature. When the real-time dynamic viscosity is higher than the preset viscosity threshold, the integral time constant of the PID is extended and the differential coefficient is increased to counteract the thermal convection hysteresis effect caused by the high viscosity insulating coolant.

9. The method according to claim 7, characterized in that, The control commands sent by the air flotation fluid enhancement module include air flotation activation logic, wherein: When any one of the server load rate, the return liquid temperature, or the chip junction temperature exceeds its corresponding static threshold, the air flotation fluid enhancement module is activated. When the calculated temperature deviation and temperature change rate are greater than their respective preset trend thresholds, the air flotation fluid enhancement module is activated at a preset time before the actual temperature exceeds the limit, and the differential control action in the PID algorithm is increased simultaneously to counteract the thermal hysteresis effect of the insulating coolant.

10. The method according to claim 7, characterized in that, Sending control commands to the air flotation fluid enhancement module further includes air flotation shutdown logic, wherein: When the return liquid temperature falls below the shut-off threshold and the rate of change remains stable for a preset duration, the graded shut-off process is initiated. The air flotation flow rate is reduced in stages according to a preset step size. During the reduction process, the proportional coefficient of the PID algorithm is reduced and the integral coefficient is increased. If the rebound of the return liquid temperature exceeds the safety threshold during the staged shutdown process, the shutdown process is terminated and the air flotation flow rate of the previous stage is restored.

Images