Air exhaust cooling structure of GPU high-power-density AI data center
By designing a movable chassis, condensation structure and air-cooling structure, the problem of limited cooling range and poor effect caused by the fixed exhaust cooling mode in the prior art is solved, and flexible mobile and efficient cooling is achieved, which is suitable for high power of GPUs. Density AI data center.
Patent Information
- Application Number
- CN202420507232.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-15
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2034-03-15
AI Technical Summary
The exhaust air cooling in the prior art adopts a fixed mode, resulting in limited cooling range and poor cooling effect, slow cooling and inflexible and practical.
A exhaust cooling structure of GPU high-power density AI data center is designed, including the chassis, condensation structure and air-cooled structure. The air-cooled structure is adjusted by roller movement, screw adjustment, the condensation structure increases the refrigeration area and turbofan acceleration. Cold air flows to achieve flexible cooling and efficient heat dissipation.
It realizes more flexible cooling area movement, quickly adjusts cooling height, improves cooling speed and efficiency, and has a more significant cooling effect. It is suitable for high-power density AI data centers.
Smart Images

Figure CN221979425U_ABST
Abstract
Description
Technical Field
[0001] The utility model relates to the technical field of exhaust cooling, and in particular to an exhaust cooling structure for a GPU high power density AI data center. Background Art
[0002] In the specification of a door cover for reducing the noise of a centrifuge in the prior art (publication number CN201346499Y), it is mentioned that "one end of the sandwich air duct is connected to the air guide device, and the other end is connected to the air outlet. The motor drives the rotor to rotate at a high speed, and the air in the centrifugal chamber is driven to rotate through the air guide device and enters the sandwich air duct of the door cover and is then discharged from the air outlet, thereby forming a negative pressure inside the centrifugal chamber to inhale fresh air to cool the motor and the rotor. Preferably, two air guide devices are provided on the inner side of the door cover, the air outlet is provided on the left side or the right side of the door cover, and the sandwich air duct is a labyrinth air duct". However, the exhaust cooling in the prior art adopts a fixed mode, which not only limits the cooling range, but also has a poor cooling effect, resulting in slow heat dissipation and cooling, which is not flexible and practical. Utility Model Content
[0003] In order to overcome the defects of the prior art, an exhaust cooling structure for a GPU high power density AI data center is provided to solve the problem that the exhaust cooling in the prior art adopts a fixed mode, which not only limits the cooling range but also has a poor cooling effect, resulting in slow heat dissipation and cooling, and is not flexible and practical.
[0004] To achieve the above-mentioned purpose, an exhaust cooling structure for a GPU high-power density AI data center is provided, including a base frame, a condensation structure and an air-cooling structure, side fixing plates are fixed to the left and right parts of the upper end surface of the base frame, an upper set plate is fixed to the upper end of the inner side surface of the side fixing plate, and a lower set plate is fixed to the lower end of the inner side surface of the side fixing plate, a screw rod is sleeved between the upper set plate and the lower set plate, and the air-cooling structure is movably connected to the screw rod, a condensation structure is arranged on the rear side of the base frame, and two sets of side connecting plates are fixed between the left and right side surfaces of the condensation structure and the rear end of the base frame.
[0005] Furthermore, rollers are installed on both the left and right parts of the lower end surface of the base frame, and the rollers are solid rollers made of PE material.
[0006] Furthermore, a sprocket wheel is installed on the lower rod body of the screw rod, and a chain is transmission-connected to the sprocket wheel, and a turntable is installed on the lower end of a group of screw rods on the right front side of the chassis.
[0007] Furthermore, a central tube is arranged at the center of the condensation structure, a liquid inlet is arranged at the upper end surface of the central tube, a liquid outlet is arranged at the lower end surface of the central tube, and multiple groups of oblique flow pipes are connected to the left and right sides of the central tube.
[0008] Furthermore, the air-cooling structure is provided with a mounting frame, an upper turbofan is provided on the upper portion of the mounting frame, and a lower turbofan is provided on the lower portion of the mounting frame.
[0009] Furthermore, movable sleeves are provided on the left and right sides of the mounting frame, and each set of movable sleeves is movably connected to the screw rod on the same side, and a slider is provided on the outer side of the movable sleeve, and the slider is slidably connected to the linear slide rail on the same side, and the linear slide rail is arranged in the middle of the inner side surface of the side fixed plate.
[0010] The beneficial effects of the utility model are:
[0011] 1. The utility model is convenient for rolling movement through rollers, so that it can be easily moved indoors, and the cooling area of the GPU high power density AI data center can be changed, which is more flexible and convenient.
[0012] 2. The utility model can drive the air cooling structure to move upward by rotating the screw rod clockwise, and can drive the air cooling structure to move downward by rotating the screw rod counterclockwise, so that the cooling height position of the air cooling structure can be adjusted conveniently, and the operation is quick and convenient.
[0013] 3. The condensation structure of the utility model is convenient for inputting the condensing agent into the central tube through the liquid inlet pipe so as to refrigerate the coolant in the central tube. The central tube is connected with the oblique flow pipes on both sides so as to increase the refrigeration area, which is helpful for low-temperature cooling and faster cooling speed.
[0014] 4. The air-cooling structure in the utility model is movably connected to the screw rod through a movable sleeve, and is slidably connected to the linear slide rail through a slider, which is convenient for vertical lifting and lowering movement. The upper turbo fan and the lower turbo fan are used for exhaust cooling, which accelerates the cold air passing through the condensation structure at the rear, and blows the cold air flow to the GPU high-power density AI data center equipment, making cooling and heat dissipation faster and more convenient. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 It is a front view schematic diagram of an embodiment of the utility model;
[0016] Figure 2 It is a rear view schematic diagram of an embodiment of the utility model;
[0017] Figure 3 It is a top view schematic diagram of an embodiment of the utility model;
[0018] Figure 4 This is a rendering of the air-cooling structure of an embodiment of the utility model.
[0019] In the figure: 1. base frame; 10. roller; 11. lower plate; 12. side fixing plate; 13. upper plate; 14. linear guide rail; 15. screw rod; 16. side connecting plate; 17. chain wheel; 18. turntable; 19. chain; 2. condensation structure; 20. liquid inlet; 21. oblique flow duct; 22. center pipe; 23. liquid outlet; 3. air cooling structure; 30. mounting frame; 31. movable sleeve; 32. slider; 33. upper turbofan; 34. lower turbofan. DETAILED DESCRIPTION
[0020] In order to make the technical problems, technical solutions and beneficial effects to be solved by the utility model clearer, the utility model is further described in detail in combination with the accompanying drawings and embodiments. The specific embodiments described herein are only used to explain the utility model, and are not used to limit the utility model. Specific details such as specific system structures and technologies are proposed in order to more thoroughly understand the embodiments of the utility model. The described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. However, it should be clear to those skilled in the art that the utility model can also be implemented in other embodiments without these specific details. Based on the embodiments in the present disclosure, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present disclosure.
[0021] The specific implementation of the utility model is described in detail below with reference to the accompanying drawings.
[0022] Figure 1 It is a front view schematic diagram of an embodiment of the utility model, Figure 2 It is a rear view schematic diagram of an embodiment of the utility model, Figure 3 A schematic top view of an embodiment of the utility model and Figure 4 This is a rendering of the air-cooling structure of an embodiment of the utility model.
[0023] Reference Figures 1 to 4 As shown, the utility model provides an exhaust cooling structure for a GPU high power density AI data center, including a base frame 1, a condensing structure 2 and an air cooling structure 3, side fixing plates 12 are fixed to the left and right parts of the upper end surface of the base frame 1, and an upper sleeve plate 13 is fixed to the upper end of the inner side surface of the side fixing plate 12, and a lower sleeve plate 11 is fixed to the lower end of the inner side surface of the side fixing plate 12, a screw rod 15 is sleeved between the upper sleeve plate 13 and the lower sleeve plate 11, and the air cooling structure 3 is movably connected to the screw rod 15, a condensing structure 2 is arranged on the rear side of the base frame 1, and two sets of side connecting plates 16 are fixed between the left and right side surfaces of the condensing structure 2 and the rear end of the base frame 1.
[0024] In this embodiment, rollers 10 are installed on both the left and right parts of the lower end surface of the base frame 1, and the rollers 10 are solid rollers made of PE material.
[0025] As a preferred implementation, the utility model facilitates rolling movement through the roller 10, so as to be easily moved indoors, and the cooling area of the GPU high power density AI data center can be changed, which is more flexible and convenient.
[0026] In this embodiment, a sprocket wheel 17 is installed on the lower rod body of the screw rod 15, and a chain 19 is transmission-connected to the sprocket wheel 17, and a turntable 18 is installed on the lower end of a group of screw rods 15 on the right front side of the base frame 1.
[0027] As a preferred embodiment, the utility model can drive the air cooling structure 3 to move upward by rotating the screw rod 15 clockwise, and can drive the air cooling structure 3 to move downward by rotating the screw rod 15 counterclockwise, so as to facilitate the adjustment of the cooling height position of the air cooling structure 3 and the operation is quick and convenient.
[0028] In this embodiment, a central tube 22 is arranged at the center of the condensation structure 2, a liquid inlet 20 is arranged on the upper end surface of the central tube 22, a liquid outlet 23 is arranged on the lower end surface of the central tube 22, and a plurality of groups of oblique flow pipes 21 are connected to the left and right sides of the central tube 22.
[0029] As a preferred embodiment, the setting of the condensation structure 2 in the utility model facilitates the input of the refrigerant into the central tube 22 through the liquid inlet pipe 20 so as to refrigerate the coolant in the central tube 22. The central tube 22 is connected to the oblique flow pipes 21 on both sides so as to increase the refrigeration area, which is conducive to low-temperature cooling and faster cooling speed.
[0030] In this embodiment, the air cooling structure 3 is provided with a mounting frame 30, and an upper turbofan 33 is provided on the upper part of the mounting frame 30, and a lower turbofan 34 is provided on the lower part of the mounting frame 30; movable sleeves 31 are provided on the left and right sides of the mounting frame 30, and each set of movable sleeves 31 is movably connected to the screw rod 15 on the same side, and a slider 32 is provided on the outer side of the movable sleeve 31, and the slider 32 is slidably connected to the linear slide rail 14 on the same side, and the linear slide rail 14 is arranged in the middle of the inner side surface of the side fixed plate 12.
[0031] As a preferred embodiment, the air-cooled structure 3 in the utility model is movably connected to the screw rod 15 through a movable sleeve 31, and is slidably connected to the linear slide rail 14 through a slider 32, so as to facilitate vertical lifting and lowering movement, and utilizes an upper turbofan 33 and a lower turbofan 34 for exhaust cooling, thereby accelerating the cold air passing through the condensation structure 2 at the rear, and blowing the cold air flow toward the GPU high-power density AI data center equipment, so as to cool down and dissipate heat faster and more conveniently.
[0032] The utility model can effectively solve the problem that the exhaust cooling in the prior art adopts a fixed mode, which not only limits the cooling range but also has a poor cooling effect, resulting in slow heat dissipation and cooling, and is not flexible and practical. The utility model can flexibly move the exhaust cooling position, and use the refrigeration structure to cool the air flow first, and then use the turbo fan to blow the cooled cold air flow to the GPU high power density AI data center equipment. The cooling area is large, and the exhaust cooling is more efficient and time-saving.
[0033] The above embodiments are used to explain the present invention rather than to limit the present invention. Any modifications and changes made to the present invention within the spirit of the present invention and the scope of the claims for protection shall be included in the protection scope of the present invention.
Claims
1. An exhaust cooling structure for a GPU high power density AI data center, characterized by: The invention comprises a base frame (1), a condensation structure (2) and an air cooling structure (3); side fixing plates (12) are fixed to the left and right parts of the upper end surface of the base frame (1); an upper sleeve plate (13) is fixed to the upper end of the inner side surface of the side fixing plate (12); and a lower sleeve plate (11) is fixed to the lower end of the inner side surface of the side fixing plate (12); a screw rod (15) is sleeved between the upper sleeve plate (13) and the lower sleeve plate (11); and the air cooling structure (3) is movably connected to the screw rod (15); a condensation structure (2) is arranged on the rear side of the base frame (1); and two sets of side connecting plates (16) are fixed between the left and right sides of the condensation structure (2) and the rear end of the base frame (1).
2. According to claim 1, the exhaust cooling structure of a GPU high power density AI data center is characterized in that: Rollers (10) are installed on both the left and right parts of the lower end surface of the base frame (1), and the rollers (10) are solid rollers made of PE material.
3. According to claim 1, the exhaust cooling structure of a GPU high power density AI data center is characterized in that: A sprocket wheel (17) is installed on the lower rod body of the screw rod (15), and a chain (19) is connected to the sprocket wheel (17) for transmission, and a rotating disk (18) is installed on the lower end of a group of screw rods (15) on the right front side of the bottom frame (1).
4. According to claim 1, the exhaust cooling structure of a GPU high power density AI data center is characterized in that: A central tube (22) is arranged at the center of the condensation structure (2), a liquid inlet (20) is arranged at the upper end surface of the central tube (22), and a liquid outlet (23) is arranged at the lower end surface of the central tube (22), and a plurality of groups of oblique flow pipes (21) are connected to the left and right sides of the central tube (22).
5. The exhaust cooling structure of a GPU high power density AI data center according to claim 1, characterized in that: The air cooling structure (3) is provided with a mounting frame (30), an upper turbofan (33) is provided on the upper part of the mounting frame (30), and a lower turbofan (34) is provided on the lower part of the mounting frame (30).
6. The exhaust cooling structure of a GPU high power density AI data center according to claim 5, characterized in that: The left and right sides of the mounting frame (30) are both provided with movable sleeves (31), and each set of movable sleeves (31) is movably connected to the screw rod (15) on the same side, and the movable sleeves (31) are A slider (32) is provided on the outer side of the slide block, and the slider (32) is slidably connected to the linear slide rail (14) on the same side. The linear slide rail (14) is arranged at the middle part of the inner side surface of the side fixing plate (12).
