A server and a server heat dissipation method

Abstract

The application relates to a server and a server heat dissipation method. The server comprises a server shell, a graphic processing module, a processor and an extended function module are sequentially arranged in the server shell along a wind flow conveying direction, a first cooler is arranged in front of the graphic processing module, a second cooler is arranged on the processor, a third cooler is arranged between the processor and the extended function module, two ends of the third cooler respectively extend to opposite inner walls of the server shell, and the center line of the third cooler is arranged to intersect the wind flow conveying direction. The third cooler is used for dissipating heat of a space where the processor and the extended function module are located, and reducing heat transmission from a space where the processor is located to a space where the extended function module is located. The device can improve the heat dissipation efficiency and dynamically control the heat dissipation temperature.

Description

Technical Field

[0001] This application relates to the field of computer hardware heat dissipation technology, and in particular to a server and a server heat dissipation method. Background Technology

[0002] With the rapid growth in demand for high-density computing, the heterogeneous component layout within servers (such as processors, graphics processing modules, and high-speed network cards) is leading to increasingly prominent heat dissipation challenges. In traditional front-intake architectures, hard drives and processors share limited cooling air channels, causing rear-mounted network cards and graphics processing modules to operate in high-temperature wake zones, resulting in generally higher operating temperatures and a risk of overheating. Meanwhile, blade servers, by integrating multiple computing units within a limited chassis space, significantly increase power density, making heat dissipation a critical factor restricting system stability and reliability.

[0003] Currently, data centers generally use central air conditioning for unified cooling. Cool air must pass through the server room and racks before reaching the servers, resulting in low cooling capacity utilization. Some manufacturers are attempting to improve airflow by optimizing duct design and adding redundant fans. For example, IBM uses dual redundant high-flow fans to enhance cool air supply. Furthermore, liquid cooling technology has been introduced into high-power chip cooling scenarios, including board-level cold plate liquid cooling and immersion liquid cooling solutions. The former reduces reliance on air cooling by transferring heat through cooling water, while the latter significantly improves heat dissipation efficiency, but usually requires modifications to the server room infrastructure.

[0004] While the aforementioned solutions improve heat dissipation performance to some extent, significant problems remain. In air cooling systems, the temperature of the cold air rises considerably during multi-stage transmission, reaching critical chips at elevated temperatures, leading to increased energy consumption. In a series layout of the graphics processing module-processor-network card, the front-end graphics processing module blocks most of the incoming airflow, causing the back-end processor and network card to maintain consistently high temperatures, even increasing the bit error rate. Although liquid cooling technology can significantly improve heat exchange capacity, its inlet water temperature is limited by the risk of condensation, reducing its heat dissipation efficiency. Immersion liquid cooling, while offering significant performance improvements, brings the problem of excessively high infrastructure modification and maintenance costs, hindering large-scale deployment. Therefore, achieving precise chip-level temperature control while balancing energy efficiency and deployment costs remains a pressing technical challenge that needs to be addressed. Summary of the Invention

[0005] Therefore, it is necessary to provide a server and server cooling method that can improve heat dissipation efficiency and dynamically control heat dissipation temperature to address the above-mentioned technical problems.

[0006] In a first aspect, a server is provided, comprising: a server housing, wherein a graphics processing module 1, a processor 2, and an expansion function module 3 are sequentially arranged in the server housing along the airflow direction; a first cooler 4 is provided in front of the graphics processing module 1, the first cooler 4 being used to cool the airflow entering the space where the graphics processing module 1 is located; a second cooler 5 is provided on the processor 2, the second cooler 5 being used to cool the surface of the processor 2; a third cooler 6 is provided between the processor 2 and the expansion function module, the two ends of the third cooler 6 extending to the opposite inner walls of the server housing, and the centerline of the third cooler 6 intersecting the airflow direction; the third cooler 6 being used to dissipate heat from the space where the processor 2 and the expansion function module 3 are located, and to reduce the transfer of heat from the space where the processor 2 is located to the space where the expansion function module 3 is located.

[0007] Secondly, a server heat dissipation method for servers is provided, including:

[0008] Obtain temperature and humidity information for each cooler;

[0009] Based on temperature and humidity information, the dew point temperature of the corresponding cooler surface is determined, and the condensation risk index is determined in combination with the cooler surface temperature.

[0010] Rolling predictions of the condensation risk index are made and the prediction results are compared with the preset dew point temperature threshold. The comparison results are obtained by introducing hysteresis judgment and time holding conditions.

[0011] The operating status of the cooler is controlled based on the comparison results.

[0012] By implementing the aforementioned server and server heat dissipation method, a graphics processing module, a processor, and an expansion function module are sequentially arranged inside the server casing along the airflow direction. Corresponding coolers are placed at the locations of different functional modules, achieving targeted heat dissipation for each critical heat-generating component. Specifically, the first cooler prioritizes reducing the temperature of the incoming airflow before reaching the graphics processing module, ensuring the graphics processing module operates in a lower-temperature airflow environment. The second cooler directly acts on the processor, making heat dissipation of the core computing unit more thorough and efficient. The third cooler spans the area between the processor and the expansion function module, with its centerline intersecting the airflow direction. This not only creates airflow cutting and temperature difference isolation between the processor and expansion function module spaces but also actively removes heat from both sides, effectively reducing the transfer of processor heat to the expansion function module space. The overall design achieves modular, zoned cooling, allowing airflow to remove heat step-by-step as it passes through different coolers. The isolation effect of the third cooler reduces thermal interference between different functional modules, ensuring stable operation of each component in a relatively independent thermal environment, thereby improving the overall heat dissipation efficiency and operational reliability of the server. Attached Figure Description

[0013] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A top view of a server provided in an embodiment of this application;

[0015] Figure 2 A front view of a third cooler for a server provided in an embodiment of this application;

[0016] Figure 3 A front view of a second cooler for a server provided in an embodiment of this application;

[0017] Figure 4 A flowchart illustrating a server heat dissipation method provided in this application embodiment;

[0018] In the diagram: 1. Graphics processing module; 2. Processor; 3. Extended function module; 4. First cooler; 5. Second cooler; 6. Third cooler; 7. Guide fins; 8. Guide channel; 9. Condensate collector; 10. Cooling pipe; 11. Fan module. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0020] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0021] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] In one embodiment, such as Figure 1 As shown, a server is provided, including: a server housing, in which a graphics processing module 1, a processor 2, and an expansion function module 3 are sequentially arranged along the airflow direction. A first cooler 4 is provided in front of the graphics processing module 1, which is used to cool the airflow entering the space where the graphics processing module 1 is located. A second cooler 5 is provided on the processor 2, which is used to cool the surface of the processor 2. A third cooler 6 is provided between the processor 2 and the expansion function module. The two ends of the third cooler 6 extend to the opposite inner walls of the server housing, and the center line of the third cooler 6 intersects the airflow direction. The third cooler 6 is used to dissipate heat from the space where the processor 2 and the expansion function module 3 are located, and to reduce the transfer of heat from the space where the processor 2 is located to the space where the expansion function module 3 is located.

[0023] Wherein, the first cooler 4 refers to the cooler arranged on one side of the graphics processing module 1, used to dissipate heat on that side; the second cooler 5 refers to the cooler arranged on the other side of the graphics processing module 1 and attached to the processor 2, used to dissipate heat on both the graphics processing module 1 and the processor 2 simultaneously; the third cooler 6 refers to the cooler arranged on the other side of the processor 2 and adjacent to the expansion function module 3, used to dissipate heat on the other side of the processor 2 and the expansion function module 3; the graphics processing module 1 refers to the graphics processing module 1 component used to perform high-performance parallel computing; the processor 2 refers to the central processing unit processor 2 used to perform general computing tasks; and the expansion function module 3 refers to the modular circuit unit used to provide network expansion, storage expansion or other functional expansion.

[0024] Based on the airflow direction formed by the front fan or chassis duct, the upstream side of the airflow from the air inlet to the air outlet is called the front. The arrangement of the first cooler, located before the airflow reaches the graphics processing module and upstream of the main airflow channel, allows the air entering the graphics processing module to exchange heat or be conditioned by the first cooler before contacting the module.

[0025] Specifically, the airflow direction is determined within the server casing, preferably using a through-flow straight duct structure with front-to-back airflow, and a stable, directional main airflow is formed by a fan array or air wall. Subsequently, the graphics processing module 1, processor 2, and expansion function module 3 are sequentially fixed and installed along this airflow direction, ensuring that the heat dissipation areas of all three face the same main channel to reduce additional pressure drop and bypass flow caused by bends in the airflow duct. The first cooler 4 is located in front of the graphics processing module 1, i.e., upstream of the graphics processing module. The first cooler can be a finned heat exchanger, a liquid-cooled pre-cooling radiator, or a high-heat-capacity phase change plate. Its air-facing surface is arranged at an equal angle or orthogonal to the main airflow. A guide shroud forces the main airflow through its heat exchange surface, thereby pre-cooling and homogenizing the airflow about to pass through the graphics processing module 1 without adding additional local fans, reducing temperature stratification and hot spot streaks at the inlet. A second cooler 5 is provided on the processor 2. The second cooler is preferably a direct contact cooling unit (such as a heat pipe / vapor chamber + high-density finned heat sink or liquid cooling plate). It is reliably coupled to the top cover of the processor 2 through a thermally conductive interface material. The heat exchange surface of the cooling unit is directly exposed to the main airflow or connected to the circulating cooling loop to extract heat from the processor core through a low thermal resistance path. A third cooler 6 is provided between the processor 2 and the expansion function module 3. The two ends of the cooler extend to the opposite inner walls of the server housing and are fitted to the inner walls by sealing strips or foam gaskets to limit the unorganized bypass of airflow in this area. The centerline of the third cooler 6 is intersecting the airflow direction. The "intersection" can be approximately perpendicular or inclined at a certain angle. The purpose is to cut the main airflow profile in a non-collinear manner, so that the cooler body can simultaneously dissipate heat from the two sides of the space where the processor 2 and the expansion function module 3 are located as a heat exchange component, and also act as an aerodynamic barrier to separate and guide the heat plumes on the processor side and the expansion function module side. To improve airflow convergence, the third cooler 6, in conjunction with upper and lower edge guide vanes, directs the segmented airflow to the high heat flux regions of their respective modules. Simultaneously, small gaps for pressure relief or micro-perforations are incorporated at the cooler's tail to reduce separation vortices and recirculation zones, thereby minimizing heat backflow caused by secondary mixing. All three types of coolers can share the main airflow of the same fan system or exchange heat through a shared cooling circuit within a liquid-cooled architecture. In the gas-liquid mixing scheme, the first cooler handles pre-cooling and temperature equalization, the second cooler handles high heat flux point source cooling, and the third cooler handles inter-area heat exchange and aerodynamic isolation. These three coolers form a continuous, low-bypass heat dissipation link through guide vanes and sealing strategies.

[0026] The first cooler, located in front of the graphics processing module, pre-cools and homogenizes the upstream airflow, resulting in a lower and more uniform air temperature distribution at the GPU inlet. This suppresses localized fin failure and uneven heat transfer caused by temperature stratification at the inlet boundary conditions. This not only reduces the peak junction temperature of the GPU but also decreases the temperature difference between the chip and the heatsink, thereby increasing the heat transfer driving force and reducing the need for the fan to increase its speed to overcome hot spots. The second cooler is directly coupled to the processor, forming a short-path, low-thermal-resistance heat transfer chain. Core heat is quickly drawn away to a location where it can be efficiently carried away by the main airflow or cooling loop, preventing the processor's heat plume from spreading widely downstream and accumulating on the air intake surface of subsequent modules. The third cooler, with its ends sealed to the inner walls of the casing and positioned intersecting the airflow direction, naturally constitutes a "heat-exchangeable barrier." On the one hand, it exchanges heat with the through-flow on its own surface, directly reducing the average volume temperature of the space where the processor and the extended function module are located; on the other hand, it geometrically interrupts the unidirectional thermal plume of the main airflow, limiting the intrusion and entrainment of high-temperature air from the processor side to the extended function module side, thereby weakening the convective coupling and thermal crosstalk between the two functional areas from the source. Since the third cooler plays a dual role of thermal and aerodynamic isolation, the transient thermal shock generated by the processor is no longer directly transmitted to the space of the extended function module. The thermal environment fluctuation amplitude of the latter is reduced, and the device life and long-term reliability are improved. Overall, the thermal field distribution inside the server is more controllable, and the temperature gradient and temperature cycle of key components are suppressed, thereby reducing thermal mismatch stress.

[0027] It should be understood that the cooler, graphics processing module 1, processor 2 and expansion function module 3 are all installed in the server housing by common mechanical fixing structures such as screws, clips or welding to ensure their stability and relative position during server operation. Those skilled in the art can choose a suitable fixing method according to actual needs, and the present invention does not limit this.

[0028] In one embodiment, such as Figure 2 As shown, the third cooler 6 includes multiple guide fins 7 arranged at intervals, with airflow passages formed between adjacent guide fins 7. The multiple guide fins 7 are interconnected, and the cross-sectional width of the guide fins 7 gradually decreases from bottom to top. The centerline of the third cooler 6 is set perpendicular to the airflow delivery direction.

[0029] Among them, the cross-sectional width refers to the guide fin 7, which is a thin metal sheet fixed inside the third cooler. It exchanges heat with the airflow through its surface and helps to remove the heat generated by the electronic components. The width of the fin gradually decreases in order to increase the airflow velocity and enhance the heat exchange efficiency. The airflow direction refers to the direction of airflow in the server, which is usually generated by the server fan. The airflow enters the heat sink from the fan and transfers heat through the guide fins.

[0030] The cross-sectional width of the guide fins 7 gradually decreases from bottom to top, forming an approximately triangular geometric structure. Specifically, when the base of the guide fins 7 is wider and the top is narrower, it can form an equilateral triangle or trapezoidal arrangement that approximates an equilateral triangle.

[0031] Specifically, the third cooler 6 is composed of multiple spaced-apart guide fins 7, forming airflow channels between them. When airflow enters this area along the airflow direction, it is first guided into these airflow channels. Heat exchange is completed during the process of the airflow fully contacting the guide fins 7, thereby achieving temperature control of the space between the processor and the extended function module. The multiple guide fins 7 are interconnected to form a stable overall structure, avoiding deformation of individual fins due to airflow impact, while improving the overall mechanical strength and heat conduction continuity of the heat sink. The centerline of the third cooler 6 is set perpendicular to the airflow direction, and the airflow passes through the guide fins 7 in a vertical scouring manner, which not only increases the coverage of the heat dissipation surface by the airflow, but also reduces the insufficient heat exchange caused by the straight flow of airflow. Through this design, the third cooler 6 achieves the dual functions of enhanced heat dissipation and aerodynamic separation in a limited space. It can not only effectively dissipate heat from the processor and extended function module areas, but also block the direct transfer of heat between the two sides, reduce thermal interference, effectively improve heat dissipation efficiency, reduce airflow pressure loss, improve airflow organization, avoid local overheating and improve system reliability. At the same time, it can ensure efficient cooling while reducing fan energy consumption and noise.

[0032] In one embodiment, the third cooler 6 further includes a recovery tank disposed below the guide fins 7 and inclined relative to the processor 2.

[0033] The recovery tank is a collection structure located below the guide fins 7. It is trough-shaped and is used to collect and guide the condensate generated by heat exchange, so as to prevent the liquid from dripping onto sensitive components such as the processor 2.

[0034] Specifically, the third cooler 6 incorporates a recovery tank below the guide fins 7. Utilizing its inclined arrangement with the processor 2, condensate that may condense or adhere during the cooling heat exchange process due to airflow and temperature differences is naturally collected and guided into the tank along the direction of gravity and the inclined surface of the recovery tank. This prevents droplets from remaining on the surface of the guide fins 7 or irregularly dripping onto the processor 2 area. This structure ensures concentrated collection of condensate below the guide fins 7, guaranteeing a continuous and stable cooling process while preventing electrical short circuits or localized temperature fluctuations on the processor 2 surface caused by direct droplet scouring. Simultaneously, the inclined design of the recovery tank with the processor 2 utilizes gravitational potential energy to achieve unidirectional discharge and guidance of the condensate, preventing repeated retention or backflow of condensate within the cooler and ensuring dry and unobstructed airflow throughout the cooler. This effectively improves the reliability of the third cooler 6 in complex operating environments, reduces the risk of hardware damage due to condensate accumulation, maintains stable heat dissipation performance, and further extends the service life of core components such as the processor 2 and expansion function modules.

[0035] In one embodiment, such as Figure 3 As shown, the second cooler 5 rests against the top of the processor 2, and the distance from the second cooler 5 to the third cooler 6 is less than the distance from the third cooler 6 to the extended function module 3.

[0036] Specifically, the short distance between the second cooler 5 and the third cooler 6 facilitates airflow optimization, allowing airflow to pass efficiently between the coolers during the cooling process. This ensures a smooth airflow path and minimal pressure loss throughout the cooling system. The third cooler 6, located between the second cooler 5 and the expansion module 3, effectively isolates the processor 2 and the expansion module 3 from thermal interference through its optimized layout, while providing dual cooling protection for both modules during airflow. Because the second cooler 5 is directly adjacent to the processor 2 and has a short distance along the airflow path, it maximizes the heat dissipation efficiency for the processor 2, reducing system temperature fluctuations caused by heat accumulation. The close proximity between the second cooler 5 and the third cooler 6 ensures effective airflow transfer, improving the overall cooling system efficiency and avoiding the risk of airflow attenuation and reduced heat dissipation due to excessive cooler spacing. The greater distance between the third cooler 6 and the expansion module 3 allows for a more stable airflow through the heat dissipation area, preventing excessive heat concentration and achieving effective temperature balance. This results in a more stable cooling effect for the entire system, reduces thermal interference between different components, and ensures stable operation of each module. This structural configuration effectively improves the heat dissipation capacity of the cooling system while ensuring the system's reliability and long-term stable operation.

[0037] In one embodiment, the second cooler 5 includes a guide channel 8 and a condensate collector 9. The guide channel 8 is disposed on the side of the processor 2 facing the first cooler 4, and the condensate collector 9 is disposed between the guide channel 8 and the processor 2, attached to the processor 2 and connected to the condensate collector 9.

[0038] The flow channel 8 refers to the groove structure formed inside the server housing to guide the flow of condensate. Its shape can be straight, arc, or inclined to achieve efficient flow guidance under different layout conditions. The condensate collector 9 refers to the receiving structure connected to the flow channel 8, which is used to collect the water condensed on the surface of the cooler. It can be an independent container or a groove structure integrally formed on the bracket, and may include a drain port, a sealing valve, or an interface for connecting to an external drain pipe, thereby realizing the orderly discharge of condensate and system maintenance.

[0039] Specifically, by setting a guide channel 8 on one side of the processor 2, the droplets can be guided and collected into the guide channel 8 in the initial stage of condensation generation, preventing condensation from dripping directly onto the surface of the processor 2 or the circuit board area. Guided by the guide channel 8, the condensation further flows into the condensate collector 9, where it is centrally stored and can be discharged through the drain port or guided to a unified drainage system through an external conduit. This effectively prevents short circuit risks or corrosion hazards caused by condensation accumulation during server operation. Through this structural design, the device achieves efficient heat dissipation from the cooler while avoiding the safety risks caused by condensation dripping from the cooler surface in traditional liquid cooling technologies, ensuring the electrical reliability of the processor 2 and surrounding sensitive components. The close fit between the guide channel 8 and the processor 2 not only improves the timeliness of condensation collection but also avoids the possibility of local short circuits caused by condensation stagnation between the cooler and the processor 2. Simultaneously, the condensate collector 9 ensures centralized management of the condensate, giving the cooling system better long-term operational stability and ease of maintenance. The above measures can improve heat dissipation efficiency while ensuring security, achieving a balance between high performance and reliability, and meeting the dual requirements of high power density servers for heat dissipation and protection.

[0040] In one embodiment, a protective layer is provided on the inner side of the guide channel 8 to reduce the adhesion of condensate.

[0041] The protective layer refers to a functional material covering the inner wall surface of the guide channel 8, which has hydrophobic or hydrophilic properties. The protective coating can be hydrophobic, allowing condensate to quickly roll off the surface of the guide channel 8 and preventing adhesion; or it can be hydrophilic, allowing condensate to quickly form a continuous water film on the surface and flow smoothly, preventing irregular water droplet retention. The protective coating material can be fluorinated polymers, nano-ceramic coatings, or other corrosion-resistant and moisture-resistant materials, and can be selected as a single-layer coating or a multi-layer composite coating structure to enhance its durability and stability, depending on different environmental requirements.

[0042] Specifically, a protective layer is provided on the surface of the guide channel 8 inside the server housing, covering the inner wall of the guide channel 8. When the surface temperature of the cooler is lower than the dew point temperature during heat exchange, water vapor in the air condenses into droplets. These droplets move downwards along the inner wall of the guide channel 8 under the influence of gravity. By forming a protective coating on the inner wall of the guide channel 8, the adhesion of condensate to the surface of the guide channel 8 is effectively reduced, making it less likely for condensate to remain on the surface of the guide channel 8. Instead, it is guided to the condensate collector 9 at a faster speed, thereby improving the timeliness and continuity of drainage and preventing condensate from accumulating or flowing back in the guide channel 8. The smooth discharge of condensate not only improves collection efficiency but also prevents secondary cooling problems caused by droplets remaining in the guide channel 8 for a long time, thus ensuring that condensate can be collected and discharged in an orderly manner through the drainage structure.

[0043] In one embodiment, the server housing further includes a plurality of cooling pipes 10 for conveying cooling medium. The first cooler 4, the second cooler 5, and the third cooler 6 are connected through the cooling pipes 10. The cooling pipes 10 contain cooling medium. The cross-sectional dimension of the cooling pipe 10 corresponding to the first cooler 4 is larger than the cross-sectional dimension of the cooling pipe 10 corresponding to the second cooler 5, and the cross-sectional dimension of the cooling pipe 10 corresponding to the second cooler 5 is larger than the cross-sectional dimension of the cooling pipe 10 corresponding to the third cooler 6.

[0044] The cross-sectional dimension of the cooling pipe 10 refers to the lateral dimension of the cooling pipe 10 that connects each cooler to the cooling medium circulation system. The size of this dimension directly affects the flow rate and velocity of the cooling medium, thus determining the heat dissipation efficiency.

[0045] Specifically, the cross-sectional area of ​​the cooling pipe 10 corresponding to the first cooler 4 is 12mm², the cross-sectional area of ​​the cooling pipe 10 corresponding to the second cooler 5 is 8mm², and the cross-sectional area of ​​the cooling pipe 10 corresponding to the third cooler 6 is 5mm². Through this gradual reduction design, the cooling medium gradually forms a pressure difference as it flows through the cooling pipe 10, enabling the three coolers to achieve automatic flow balance in the same loop, avoiding flow imbalance caused by excessively low or high flow resistance in a single cooler pipe. Specifically, the first cooler 4 corresponding to the graphics processing module 1 requires a higher flow rate of cooling medium due to its larger heat load, so it is paired with a larger cross-sectional area cooling pipe 10 to ensure its heat exchange capacity; the second cooler 5 corresponding to the processor 2 is supplied by a medium cross-sectional area cooling pipe 10, taking into account both flow rate and pressure drop matching; the third cooler 6 corresponding to the extended function module 3 has a lower heat load, and its cooling requirements can be met by a smaller cross-sectional area cooling pipe 10. The stepped flow channel's pressure differential balance design enables the three coolers to distribute heat as needed within a single loop while avoiding excessive flow deviation, thereby improving the overall system's heat dissipation stability and cooling efficiency. Through this structure, the cooling pipe 10 achieves graded control from large to small in terms of physical dimensions. It should be noted that the cross-sectional dimensions of the cooling pipe 10 described above are merely example parameters, and these parameters can be adjusted according to actual application requirements. For example, the pressure drop characteristics can be modified by changing the cross-sectional area, the length of the cooling pipe 10, or the inner wall roughness, ensuring optimal matching of refrigerant flow distribution under different application scenarios.

[0046] In one embodiment, the cooling pipe 10 is provided with a spray section corresponding to the second cooler 5, the spray section facing the surface of the processor 2, for spraying cooling medium.

[0047] The spray section includes spray holes, which are tiny jet channels opened on the cooling pipe 10 to spray the cooling medium onto the surface of the processor 2 in the form of mist or fine stream, forming a direct cooling path. The spacing between them is a preset value to ensure spray uniformity. The spray direction refers to the main flow direction of the cooling medium sprayed from the spray holes. In this invention, it is designed to be towards the core heat-generating area of ​​the processor 2. The cooling medium refers to the medium that flows in the cooling pipe 10 and achieves heat transfer through cooling, condensation, or forced convection. It can be a liquid coolant, a cryogenic coolant, or a phase change working fluid.

[0048] Specifically, by providing at least one spray hole on the cooling pipe 10 corresponding to the second cooler 5, and distributing the spray holes at preset intervals on the surface of the cooling pipe 10 with the spray direction facing the processor 2, directional cooling of the high heat flux density area of ​​the processor 2 is achieved. When the cooling medium flows in the cooling circuit, when it passes through the area of ​​the cooling pipe 10 with the spray holes, part of the cooling medium is sprayed onto the surface of the processor 2 in the form of atomization or fine stream, directly contacting the heat-generating area of ​​the processor 2, forming phase change heat transfer or forced convection heat transfer, thereby quickly removing the heat generated during the operation of the processor 2. This design allows the processor 2 to not only rely on the contact cooling of the cooler, but also to obtain additional direct spray cooling, enhancing the diversity and efficiency of the heat dissipation path. When the processor 2 is under high load and the temperature rises sharply, the spray structure can activate the spray function to achieve rapid cooling; when the temperature drops to a reasonable range, the spray function is turned off to avoid condensate accumulation caused by excessive spraying. To ensure the long-term stable operation of the system, the distribution spacing of the spray holes is optimized through preset values ​​to make the spray coverage uniform, and the spray direction is corrected by fluid dynamics to ensure that the cooling medium can maximize the coverage of the processor 2 surface without the occurrence of off-center spray or liquid backflow and accumulation.

[0049] In one embodiment, a fan module 11 is provided between the graphics processor 2 and the processor 2, and the fan module 11 is used to generate airflow in the server housing.

[0050] Among them, fan module 11 refers to one or more fan assemblies installed in the server housing. The fans can be axial fans, centrifugal fans or thin fans, and their direction and number can be flexibly set according to the internal layout of the server chassis to achieve airflow coverage of the cooler and adjacent functional modules.

[0051] Specifically, the fan module 11 generates a directional airflow during operation, which flows between the surface of the second cooler 5 and the surface of the graphics processing module 1, enhancing local convective heat transfer. As the second cooler 5 absorbs heat from the processor 2 and the graphics processing module 1, its surface temperature gradually increases. The airflow generated by the fan module 11 carries away the heat adhering to the surface, accelerating the heat exchange process between the cooler and the cooling medium and preventing localized overheating of the cooler surface. Through this structural design, the liquid cooling of the cooler and the air cooling of the fan module 11 complement each other, enabling rapid cooling not only in high heat flux density environments but also maintaining the system temperature solely through the fan module 11 in certain operating scenarios, thereby improving the operational flexibility and redundancy of the entire cooling system. When the cooling medium flow rate is insufficient or the circulation efficiency decreases, the fan module 11 can act as an auxiliary cooling unit, ensuring that the processor 2 and the graphics processing module 1 do not overheat and fail due to insufficient cooling. In addition, the airflow driven by the fan module 11 can also form an overall airflow inside the server, which helps to remove the residual heat generated by the extended function module 3, power supply components or other auxiliary circuits, further improving the overall heat dissipation performance and stability of the system.

[0052] In one embodiment, a liquid inlet is provided at the end of the cooling pipe 10 for replenishing the cooling medium when it is insufficient or needs to be replaced. The liquid inlet can be a threaded interface, quick-connect fitting, or other structure that can be connected to a liquid replenishment device, facilitating the injection or discharge of cooling medium during maintenance. A sealing valve is provided at the liquid inlet, which is closed when not replenishing liquid to prevent cooling medium leakage and ensure the sealing and pressure stability of the cooling pipe 10 during operation. When replenishment is required, the sealing valve is opened and the liquid replenishment device is connected to inject cooling medium into the cooling pipe 10. After replenishment, the valve is closed again to ensure continuous sealing of the pipeline. In this embodiment, the liquid inlet can be located upstream, downstream, or at the end of the loop of the cooling pipe 10 according to the spatial layout of the server chassis and the convenience of maintenance. The specific location can be adjusted according to the actual design of the heat dissipation system. Through the above settings, the liquid inlet not only improves the maintainability of the cooling medium circulation system but also avoids the decrease in system heat exchange efficiency due to the gradual loss of cooling medium, thereby ensuring the heat dissipation stability of the server under long-term high-load operation.

[0053] Preferably, the arrangement of the guide fins 7 above the vertical space of the third cooler 6 is not limited to a simple parallel arrangement. Various spacing methods can be used to adapt to different condensate distribution patterns and the internal structure of the server. For example, the guide fins 7 can be arranged in a staggered manner, with adjacent rows of guide fins 7 offset horizontally, thereby increasing the coverage and interception range of droplets and preventing condensate from falling directly through the gaps between the fins. Alternatively, a staggered arrangement can be used, creating a height difference in the vertical direction, so that condensate is intercepted and guided by fins of different heights during its descent, forming a multi-stage interception and slow-release structure. An inclined angle arrangement can also be used, setting a certain angle between the guide fins 7 and the horizontal plane, allowing droplets to quickly slide into the recovery tank after contacting the fins, thus shortening the residence time of droplets on the fins and preventing accumulation. These different spacing methods can be used individually or in combination within the same bracket to achieve optimal matching based on the condensate generation and distribution of the third cooler 6 under different operating conditions.

[0054] By incorporating diverse spacing methods such as staggered, layered, and angled arrangements, the guide fins 7 achieve more efficient droplet interception and guidance within a limited space. This not only improves the collection efficiency of condensate but also effectively reduces the probability of condensate dripping freely onto the electronic components below, thereby enhancing the overall protection capability of the device and the reliability of the heat dissipation system. This design ensures that the server maintains good safety and long-term stability even under extreme operating conditions such as high humidity and high power density.

[0055] In this application, the spacing of the guide fins 7 refers to their arrangement and angle in space. It is not limited to equidistant parallel arrangement, but can be staggered, layered, or inclined at an angle, etc., and can be combined and applied according to different needs.

[0056] Preferably, the first cooler 4 is optimized for the high heat flux density characteristics of the graphics processing module 1. Its main body adopts a copper-nickel alloy manifold structure, with 8 to 12 sets of parallel microchannels internally. The cross-sectional dimensions of the microchannels are preferably 0.6mm × 2mm, conformally fitting the shape of the heat sink of the graphics processing module 1. A phase change material with a thermal conductivity of 8W / mK is pre-coated on the contact surface to reduce contact thermal resistance and achieve efficient heat transfer. To achieve stable circulation of the cooling medium, the first cooler 4 is fixed by a flow-guiding bracket. This bracket provides a pre-reserved channel for the cooling pipe 10 during installation and leads out the cooling medium circulation pipeline. The cooling pipe 10 is equipped with a φ3mm dual-loop capillary network to transport the cooling medium to and from the cooler, and finally connects to the central air conditioning branch via a quick-connect interface for rapid connection to an external cold source. The second cooler 5 adopts a spray-cooling synergistic design to meet the high transient heat flux requirements generated during high-frequency operation of the processor 2. Specifically, a porous cooling plate is installed on the processor 2 heatsink. Liquid coolant is sprayed onto the surface of the processor 2 heatsink through micro-holes with a diameter of approximately 0.1 mm and atomized, absorbing a large amount of heat through latent heat during the phase change process. To avoid the problem of dry burning on the surface of the processor 2 after the coolant has been completely cooled, an array of 12 heat pipes with a diameter of 5 mm is embedded at the bottom of the cooling plate. This array can simultaneously exhaust uncooled liquid coolant and the gas generated during cooling into the circulation system, thereby ensuring the continuous and stable operation of the cooling system. The upper cavity of the processor 2 heatsink is a sealed cavity, ensuring that the coolant maintains a closed circulation path during spraying, cooling, and recirculation. The third cooler 6 is located behind the processor 2 and installed at an angle of 30° to 60°, preferably 30°. This cooler uses inverted V-shaped fins to construct an airflow channel, guiding the airflow inside the chassis along the inclined direction, thereby enhancing the air heat exchange effect on the area behind the processor 2 and the expansion module 3 area without increasing the additional fan load. To further prevent condensate from accumulating on the fin surface and causing electrical conductivity risks or airflow blockage, a nano-hydrophilic coating is applied to the fin surface, making its contact angle less than 10°. The condensate spreads rapidly on the surface and flows into the guide channel 8 below along the direction of gravity. It is then guided into the condensate collector 9 for centralized storage and finally discharged through the drain port of the collector or by connecting to an external drain pipe, ensuring the long-term operational safety of the server.

[0057] The structural form, size range, material selection and quantity allocation parameters mentioned in the above embodiments are only preferred examples. Those skilled in the art can adjust or replace them according to specific application needs, and should not be construed as limiting the scope of protection of this application.

[0058] In one embodiment, such as Figure 4 As shown, a server heat dissipation method applied to servers includes:

[0059] S100: Acquires temperature and humidity information for each cooler;

[0060] S200: Based on temperature and humidity information, determine the dew point temperature of the corresponding cooler surface and combine it with the cooler surface temperature to determine the condensation risk index.

[0061] S300: Performs rolling prediction of condensation risk index and compares the prediction results with the preset dew point temperature threshold. The comparison results are obtained by introducing hysteresis judgment and time holding conditions.

[0062] S400: Control the working state of the cooler based on the comparison results.

[0063] In one embodiment, controlling the temperature of the cooler based on the comparison results includes:

[0064] In response to the dew point temperature of the cooler surface being lower than a preset dew point temperature threshold, the flow rate of the cooling medium in the corresponding cooling pipe of the cooler is reduced.

[0065] In response to the dew point temperature of the cooler surface being higher than a preset dew point temperature threshold, the flow rate of the cooling medium in the corresponding cooling pipe of the cooler is increased.

[0066] Among them, the dew point temperature refers to the critical temperature at which air condenses under certain humidity and temperature conditions. It is used in conjunction with the surface temperature of the cooler to determine whether condensation will form. The preset dew point temperature threshold is a threshold temperature set during the system design phase based on the equipment operating environment and allowable humidity level. It is used to determine whether there is a risk of condensation formation.

[0067] Specifically, sensors collect real-time temperature and humidity information around the cooler. Temperature information reflects the actual temperature of the air and the cooler surface, while humidity information indicates the water vapor content in the air. Based on these two types of information, the system calculates the ambient dew point temperature and compares it with the cooler surface temperature to obtain the Condensation Risk Index (CRI). This index is a quantitative risk indicator; as the cooler surface temperature gradually approaches or falls below the dew point temperature, the CRI value increases, indicating a significant possibility of condensation on the cooler surface. In this way, the system can transform the risk of condensation on the cooler surface into measurable data, providing a basis for subsequent prediction and control. Based on this, the system further performs rolling predictions of the condensation risk index. The prediction process comprehensively utilizes historical data and current environmental conditions to assess the changing trends of cooler surface temperature and dew point temperature over several future periods, thereby predicting future condensation risk. The prediction model continuously updates the input data; each time new temperature and humidity information is acquired, the future risk situation is recalculated, forming a dynamic prediction result. To avoid frequent switching due to instantaneous environmental fluctuations or measurement errors, the system incorporates hysteresis judgment and time-holding conditions during the prediction process. A high-risk state is only identified when the risk index consistently exceeds a threshold and meets the time condition. Subsequently, the system compares the prediction results with preset thresholds. If condensation risk exists in the future, the system adjusts the cooler's operating status in advance, such as changing cooling power or airflow to reduce the proximity of surface temperature to dew point temperature. In this way, the cooler can intervene before condensation actually occurs, effectively preventing surface water formation while maintaining cooling efficiency. In summary, this method, through a closed-loop process of "data acquisition—determining dew point—determining risk index—rolling prediction—threshold judgment—control execution," can not only promptly detect potential condensation risks to the cooler but also achieve early prevention. Compared to traditional methods that rely solely on the current state, this method more accurately reflects future risks, avoids energy waste caused by frequent switching, ensures the stability and reliability of cooler operation, thereby extending equipment life and improving the overall energy efficiency of the system.

[0068] In one embodiment, the spray holes on the cooling pipe corresponding to the second cooler are designed as controllable micro-valve structures. Each spray hole is opened or closed individually by an electrically controlled micro-valve. The micro-valve is electrically connected to the control system. The control system performs regional adjustment based on the processor surface temperature distribution data collected by the sensor. When the temperature of a certain area exceeds a preset threshold, the spray hole located in that area is opened first to spray the cooling medium. In areas with lower temperatures, the spray hole remains closed.

[0069] This enables on-demand spraying and zoned heat dissipation. The spray nozzles can provide two operating modes during operation: continuous spraying and pulse spraying. Continuous spraying is suitable for scenarios where the processor runs at full load for a long time to maintain a stable temperature environment; pulse spraying is suitable for short-term peak load scenarios, which can quickly remove heat through intermittent spraying while reducing the consumption of cooling medium.

[0070] To ensure the long-term stable operation of the spray cooling system, the spray nozzles work in coordination with the cooling pipe inlet and condensate recovery channel. When the sensor detects insufficient cooling medium or low condensate recovery, the medium can be replenished through the inlet, maintaining a stable cooling loop. In this way, spray cooling not only provides uniform cooling across the entire surface but also intelligently adjusts based on the localized heat intensity of the processor, achieving regionalized and differentiated cooling modes. This ensures optimal heat dissipation under various load conditions. This design enhances the flexibility and intelligence of the spray cooling system, significantly improving its heat dissipation efficiency and reliability.

[0071] In one embodiment, the condensate collector 9 is provided with a return port at the bottom, which is connected to the cooling pipe 10 through a conduit and works with a check valve or a return pump to achieve directional recovery of condensate.

[0072] A filtration device or micro-purification module can be further installed inside the condensate collector 9 to remove any dust particles, metal ions, or organic impurities that may be mixed in, ensuring that the recovered condensate meets the requirements for use with the cooling medium. Once a certain amount of condensate has been collected, it is guided into the cooling pipe 10 by gravity or a reflux pump, and enters the cooling circuit through the inlet to directly replenish the cooling medium, thereby compensating for losses caused by cooling and leakage during long-term circulation.

[0073] This design creates a partial self-circulation mechanism for the cooling medium, enabling the server to maintain a stable coolant level in the cooling circuit during prolonged high-load operation, reducing maintenance frequency and replenishment costs. Simultaneously, condensate recovery reduces wastewater discharge, improving resource utilization and environmental value.

[0074] Furthermore, a drain port can be formed at the bottom of the condensate collector 9. The drain port can be designed in various ways according to different application environments and needs. For example, the condensate can be discharged directly to the outside of the chassis by gravity drainage. In the case of limited space, the drain port can be connected to a capillary liquid guiding device to guide the condensate from the collector to the drainage area at the far end by using capillary action.

[0075] Through the extended design of the above-mentioned drainage methods, the condensate collector 9 can not only adapt to different server chassis structures and working environments, but also improve the reliability and flexibility of the drainage process. Whether relying on gravity, capillary conduction, or negative pressure pumping, it can ensure that the collected condensate can be discharged in a timely and effective manner, avoiding localized corrosion, leakage, or short circuits caused by liquid retention. This allows the server to maintain good electrical safety and long-term stability even under high-power operation.

[0076] In one embodiment, a comparison is made between the dew point temperature of the cooler surface and a preset dew point temperature threshold to obtain a comparison result, including:

[0077] The temperature information is linearized and filtered to obtain the compensated temperature value, and the humidity information is compensated by temperature compensation and digital filtering to obtain the compensated relative humidity value.

[0078] The controller calculates the saturated vapor pressure under the temperature condition based on the compensated temperature value, and converts it into the current vapor pressure by combining the relative humidity value. It then calls the reverse calculation algorithm based on the Magnus model to deduce the dew point temperature of the cooler surface.

[0079] The dew point temperature is compared with a preset dew point temperature threshold to obtain a comparison result, which includes: above the preset threshold and below the preset threshold.

[0080] The Magnus model is an empirical approximation model used to describe the relationship between water vapor pressure and temperature in the air. Its basic idea is that the saturated water vapor pressure that air can hold at different temperatures can be represented by an exponential relationship between a set of empirical coefficients and temperature. Saturated water vapor pressure refers to the pressure when water vapor in the air reaches saturation at a certain temperature. It is a key intermediate quantity for calculating relative humidity and dew point temperature. Current water vapor pressure refers to the actual pressure of water vapor in the air calculated based on relative humidity and saturated water vapor pressure under actual conditions. Dew point temperature refers to the critical temperature at which air begins to condense while keeping the air water vapor content constant. The preset dew point temperature is a threshold set by the system according to the requirements of safe operation. It can be a single value or a combination of upper and lower thresholds.

[0081] Specifically, by uniformly compensating and filtering temperature and humidity information, errors caused by sensor noise and temperature drift are significantly reduced, ensuring that the data entering the calculation stage is stable and reliable. By using saturated water vapor pressure as an intermediate quantity to connect the relationship between temperature and humidity, the dew point temperature can be derived, making the process of obtaining the dew point universal and portable. By comparing the dew point temperature with a preset threshold with hysteresis, and supplementing it with a time jitter reduction strategy, the output comparison results are guaranteed not to frequently flip due to fluctuations in a single sampling, thus providing a stable and accurate judgment basis for subsequent control modules.

[0082] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented using electronic hardware extension function modules, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware extension function modules and software, the composition and steps of the various examples have been generally described above in terms of functionality. Whether these functions are implemented using hardware extension function modules or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0083] The above provides a detailed description of a server heat dissipation method provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only intended to help understand the method and core ideas of this application. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A server, characterized in that, include: The server housing contains a graphics processing module (1), a processor (2), and an expansion function module (3) arranged sequentially along the airflow direction. A first cooler (4) is located in front of the graphics processing module (1) to cool the airflow entering the space containing the graphics processing module (1). A second cooler (5) is located on the processor (2) to cool its surface. A third cooler (6) is located between the processor (2) and the expansion function module. Both ends of the third cooler (6) extend to the corresponding inner walls of the server housing, and the centerline of the third cooler (6) is aligned with the airflow direction. The third cooler (6) is arranged in an intersecting manner to dissipate heat from the space containing the processor (2) and the extended function module (3), and to reduce the transfer of heat from the space containing the processor (2) to the space containing the extended function module (3). The third cooler (6) includes multiple guide fins (7) arranged at intervals, forming airflow channels between adjacent guide fins (7). The multiple guide fins (7) are interconnected, and the cross-sectional width of the guide fins (7) gradually decreases from bottom to top, forming an approximately triangular geometric structure. The arrangement of the guide fins (7) in the vertical space of the third cooler (6) includes one or more of the following: equidistant parallel arrangement, staggered arrangement, etc. Layered and inclined angled structures are used to create a horizontal misalignment and a vertical height difference, or to accelerate the condensate flow process through an inclined arrangement; the centerline of the third cooler (6) is set perpendicular to the airflow direction, and multiple guide fins (7) are interconnected. The third cooler (6) also includes a recovery tank, which is located below the guide fins (7) and is inclined relative to the processor (2). The centerline of the third cooler is set perpendicular to the airflow direction; the second cooler (5) includes a guide trough (8) and a condensate collector (9). The guide trough (8) is located on the side of the processor (2) facing the first cooler (4). A condensate collector (9) is located between the guide channel (8) and the processor (2), attached to the processor (2) and connected to the condensate collector (9); the bottom of the condensate collector (9) is provided with a return port, which is connected to the cooling pipe (10) through a conduit and is used in conjunction with a one-way valve or a return pump to realize the directional recovery of condensate; the condensate collector (9) is provided with a filter device or a micro purification module to purify the impurities in the collected condensate; the second cooler (5) is against the top of the processor (2); the distance from the second cooler (5) to the third cooler (6) is less than the distance from the third cooler (6) to the extended function module (3);The server housing also includes multiple cooling pipes (10) for transporting cooling medium. The first cooler (4), the second cooler (5), and the third cooler (6) are connected by the cooling pipes (10). The cooling pipes (10) contain cooling medium. The cross-sectional dimension of the cooling pipe (10) corresponding to the first cooler (4) is larger than that of the cooling pipe (10) corresponding to the second cooler (5). The cross-sectional dimension of the cooling pipe (10) corresponding to the second cooler (5) is larger than that of the cooling pipe (10) corresponding to the third cooler (6). The cooling pipes (10) are provided with spray sections corresponding to the second cooler (5). The spray sections face the surface of the processor (2) and are used to spray the cooling medium.

2. A server according to claim 1, characterized in that, The inner side of the guide channel (8) is provided with a protective layer to reduce the adhesion of condensate.

3. A server according to claim 1, characterized in that, include: A fan module (11) is provided between the graphics processing module (1) and the processor (2), and the fan module (11) is used to generate airflow in the server housing.

4. A method for controlling heat dissipation of the server according to any one of claims 1-3, characterized in that, The method includes: Obtain temperature and humidity information for each cooler; Based on the temperature and humidity information, the dew point temperature corresponding to the surface of the cooler is determined, and the condensation risk index is determined in combination with the surface temperature of the cooler. The condensation risk index is rolled out for prediction and the prediction results are compared with a preset dew point temperature threshold. The comparison results are obtained by introducing hysteresis judgment and time holding conditions. The operating state of the cooler is controlled based on the comparison results.

Images