A scalable modular computing power center architectural design method

Abstract

The application belongs to the technical field of computing center building design, and specifically discloses a scalable modular computing center building design method, which comprises the following steps: determining the overall structure of the computing center building, introducing a building space adaptability coefficient, and evaluating the adaptability of the space to the computing equipment and the liquid cooling system; performing module division and assembly deployment based on a module expansion potential coefficient, optimizing building arrangement and infrastructure by using a network transmission loss coefficient; constructing a composite heat dissipation system, adopting a multi-stage waste heat recovery structure to improve energy utilization efficiency; deploying dynamic environment safety monitoring nodes, and realizing safety hazard detection and dynamic evaluation by combining a risk quantification algorithm; building a centralized control platform to adaptively adjust the operating state of each module; realizing rapid expansion through the assembled module, and evaluating the expansion scheme by using an expansion cost and cycle model. The application realizes the quantitative optimization of the computing center in multiple dimensions such as time, cost and efficiency, and is suitable for the construction and upgrading of various artificial intelligence computing centers.

Description

Technical Field

[0001] This invention relates to the field of computing center building design technology, and in particular to a scalable modular computing center building design method. Background Technology

[0002] With the rapid development of the digital economy, the demand for computing power has exploded. As the core carrier of digital infrastructure, computing centers are constantly upgrading in terms of construction scale and functional requirements. Existing computing center architectural designs mostly adopt traditional fixed layout models, which have the following prominent problems: First, the spatial adaptability is poor. The building structure does not match the installation and operation requirements of core facilities such as computing equipment and liquid cooling systems, which can easily lead to problems such as crowded equipment layout and chaotic pipeline laying, affecting the stability of equipment operation. Second, the modularity is low. The integration of various functional areas is insufficient, and there is a lack of standardized interfaces, making it difficult to achieve rapid expansion and functional upgrades, and unable to flexibly respond to dynamic changes in computing power demand. Third, the energy utilization efficiency is low. The heat dissipation system mostly adopts a single cooling method, with low waste heat recovery and utilization rate, resulting in a high power usage efficiency (PUE) of the computing center, which does not meet the requirements of green and low-carbon development. Fourth, the security and control system is imperfect. There is a lack of scientific methods for quantitative assessment of security risks, and the deployment of environmental monitoring nodes is unreasonable, making it difficult to achieve accurate detection and dynamic control of security hazards. Fifth, the expansion cost is high and the cycle is long. Traditional expansion methods require large-scale modification of the building structure, which is complex and affects the normal operation of existing computing equipment.

[0003] Therefore, there is an urgent need for a building design method for computing centers that can solve the above problems, achieve spatial adaptability, modular integration, high efficiency and energy saving, safety and controllability, and rapid expansion, so as to meet the needs of large-scale, intensive and green development of computing centers. Summary of the Invention

[0004] The purpose of this invention is to provide a scalable and modular computing center building design method, which solves the problems of poor spatial adaptability, low modularity, low energy efficiency, insufficient security and control, and inconvenience in expansion in existing computing center building designs, thereby reducing construction and operation costs and improving the overall benefits of computing centers.

[0005] To achieve the above objectives, the present invention provides a scalable modular computing center building design method, comprising the following steps: S1. Determine the overall structure of the computing center building and introduce a building space adaptability coefficient to assess whether the computing center building space meets the installation and operation requirements of computing equipment and liquid cooling system. S2. Introduce a module expansion potential coefficient for module division and assembly deployment; S3. Introduce network transmission loss coefficient for building layout and infrastructure optimization. S4. Construct a composite heat dissipation system for heat dissipation management and layout; S5. Evaluate and optimize energy utilization efficiency by adopting a multi-stage waste heat recovery structure; S6. Deploy several environmental safety monitoring nodes and combine them with safety risk quantification assessment algorithms to detect and dynamically assess safety hazards; S7. Construct a centralized control platform to adaptively adjust the operating status of each module; S8. Rapidly expand the computing center by adding prefabricated installation blocks and replacing functional modules, and introduce expansion cost calculation and expansion cycle calculation models to evaluate expansion schemes.

[0006] Preferably, in S1, the building space of the computing center includes the height under the beams. Space reserved for liquid cooling water pipes under the floor and reserved space at the top of the rack By introducing a building space adaptability coefficient The formula for assessing whether the building space of a computing center meets the installation and operation requirements of computing equipment and liquid cooling systems is as follows: ; in, This represents the minimum space requirements for computing equipment and liquid cooling systems.

[0007] Preferably, in S2, the computing center is divided into the following sections based on the building space layout of the computing center in S1: The computing equipment module includes storage cabinets, network cabinets, air-cooled communication cabinets, GPU servers, liquid-cooled cabinets, CDUs, column-head power distribution cabinets, and column-head accessory cabinets; The heat dissipation module includes a fluorinated phase change material heat dissipation unit, an integrated spray-type heat dissipation cabinet container, an air conditioning indoor unit, an air-cooled fluorine pump air conditioner, and a natural ventilation shaft; The energy module includes a dry-type transformer, a high-voltage well, a low-voltage well, distributed solar photovoltaic panels, and an incoming line room; The control module includes a data acquisition terminal, a remote monitoring terminal, a low-voltage electrical well, a user work area, and an office module; The safety module includes a central control center safety control room, a fire control center, laser scattering and infrared sensor detectors, a heptafluoropropane gas fire extinguishing system, safety monitoring nodes, emergency response equipment, anti-static facilities, moisture-proof facilities, safety railings, and lightning protection grounding devices; The waste heat recovery module includes a multi-stage waste heat recovery structure, a water source heat pump, and waste heat exchange components; Prefabricated assembly blocks are used to produce the functions of each module, and the functions of each module are assembled through standardized interfaces to complete the deployment. The rack layout of the computing modules follows a minimum single rack area of ​​2m². 2The maximum length of a row of server racks is 15 meters, and the width of the aisle is not less than 1 meter; The integration rationality and expansion capability of each functional module are quantitatively evaluated by introducing a module expansion potential coefficient. The quantitative formula for the module expansion potential coefficient is as follows: ; in, Increase the module's potential coefficient; To standardize the number of interfaces; This represents the total number of interfaces in the module. Reserve splicing area for prefabricated assembly blocks; This represents the total area of ​​the mounting blocks.

[0008] Preferably, S3 is as follows: A modular, flat layout is adopted, with computing equipment modules centrally arranged. A network transmission loss coefficient is introduced to optimize building layout and infrastructure, as shown in the following formula: in, This is the network transmission loss coefficient. For the corresponding network protocol, the maximum transmission distance This represents the actual network transmission distance.

[0009] Preferably, in S4, constructing a composite heat dissipation system includes the following steps: S41. The heat dissipation module adopts an integrated spray-type heat dissipation cabinet container structure, with multiple port interfaces. The spray system adjusts the spray intensity and frequency in real time according to the operating temperature of the computing equipment. S42. The heat dissipation container is equipped with a self-generating heat dissipation unit constructed with fluorinated phase change material. S43. The heat dissipation system adopts a zoned independent control and global monitoring mode. The cooling source system supports mixed deployment of liquid cooling and air cooling, with a mainstream air-liquid ratio of 5:5 and the liquid side is planned for 100% full load.

[0010] Preferably, in S4, a heat dissipation control algorithm and a temperature quantization formula are introduced to dynamically optimize the heat dissipation strategy, including: The spray intensity is dynamically controlled by calculating the real-time spray intensity of the spray system based on the real-time heat generation of the computing equipment and the ambient temperature. The formula is: ; in, For real-time spray intensity, Based on the basic spray intensity, To generate heat in real time for the equipment, For the rated heat output of the equipment, For real-time ambient temperature, Standard ambient temperature; Quantifying chip surface temperature: Real-time chip surface temperature under the synergistic effect of phase change material and spray liquid cooling. The calculation formula is: ; in, This represents the real-time heat dissipation of the spray liquid cooling system. This represents the real-time heat dissipation of the phase change material. For the specific heat capacity of the chip, For the core quality of the chip; The formula for calculating the real-time heat dissipation of phase change materials is as follows: ; in, The thermal conductivity of the phase change material is... This refers to the contact area between the phase change material and the heat dissipation surface. The phase transition temperature of a phase change material. To determine the thickness of the phase change material; The formula for quantifying heat dissipation efficiency is as follows: ; in, The total heat dissipation efficiency of the composite heat dissipation system. This represents the total heat dissipation. To assist in heat dissipation for air cooling, This represents the total heat output of the equipment.

[0011] Preferably, in S5, the multi-stage waste heat recovery structure includes: Primary recycling is used for internal heating and equipment preheating in the computing center; Secondary heat recovery is used for heating surrounding buildings. Waste heat recovery and heat dissipation are linked. Low-grade waste heat is extracted through a water source heat pump, improving the energy efficiency of the computing center. Below 1.05, introduce The calculation formula and waste heat recovery efficiency quantification model are used to evaluate and optimize energy utilization efficiency. Among them, the energy efficiency of computing centers The calculation formula is as follows: ; in, To improve the energy efficiency of computing centers, This represents the total power consumption of the computing center. The core power consumption of IT equipment For the power consumption of the cooling system, For the power consumption of the power supply and distribution system, Power consumption for other auxiliary systems; After introducing waste heat recovery, the actual effective energy utilization efficiency of the computing center The calculation formula is: ; in, The equivalent electrical energy from waste heat recovery is calculated from the recovered heat: ; In the formula, This represents the total amount of waste heat recovered. The energy efficiency ratio of a water source heat pump; The waste heat recovery efficiency quantification model is shown in the following formula: ; in, The total waste heat recovery efficiency. This is for primary waste heat recovery. This is a secondary waste heat recovery process. The total waste heat discharged by the cooling system; The formula for improving the efficiency of a water source heat pump is as follows: ; in, For the energy efficiency ratio of a water source heat pump, To enable the heat pump to output high-grade heat, Input electrical energy into the heat pump. This refers to low-grade waste heat extracted by a heat pump.

[0012] Preferably, in S6, an intelligent fire alarm and heptafluoropropane gas fire extinguishing linkage mode is adopted, laser scattering and infrared sensing are fused for detection, several dynamic and environmental safety monitoring nodes are deployed, and a safety risk quantitative assessment algorithm is used to detect and dynamically assess safety hazards. The safety risk quantitative assessment is as follows: ; in, For safety risk values, To monitor the number of nodes, For the first The probability of a potential problem occurring at each node. For the first The impact range coefficient of potential hazards at each node. For the first The risk loss coefficient of each node.

[0013] Preferably, in S7, with A multi-objective optimization model is constructed with the objective function of maximizing the overall benefit coefficient: ; in, , , , For the weighting coefficients, satisfying ; The total heat dissipation efficiency of the composite heat dissipation system. The total waste heat recovery efficiency. For real-time operating costs, For rated operating costs, This represents a safety risk value.

[0014] Preferably, in S8, the extended cost calculation model is expressed as follows: ; in, For total expansion costs, Price per module To increase the number of modules, Installation cost per unit area For splicing and installation area, Save costs by enabling modular expansion; The extended cycle calculation model is expressed as follows: ; in, For the total expansion period, For the prefabrication and transportation cycle of individual modules, To increase the number of modules, For on-site assembly and installation time, This refers to the module integration and testing cycle.

[0015] Therefore, the present invention adopts the above-mentioned scalable modular computing center building design method, and the beneficial effects are as follows: (1) This invention achieves precise design through module expansion potential coefficient, expansion cost and cycle model, which can quickly adapt to the dynamic growth demand of computing power; adopting prefabricated assembly and standardized deployment, it significantly shortens the construction cycle and reduces operation and maintenance costs, and the total cost of ownership can achieve a better performance than traditional solutions in the short term.

[0016] (2) By using the method of the present invention, the efficiency of the composite heat dissipation system can be greatly improved, the chip temperature can be stabilized and controlled, and the frequency reduction due to high temperature can be effectively avoided; through quantitative optimization of PUE and waste heat recovery, energy consumption can be significantly reduced and waste heat recovery efficiency can be improved, so as to achieve high efficiency, energy saving and green low carbon operation.

[0017] (3) This invention relies on safety risk quantitative assessment and whole-domain dynamic environment monitoring to greatly improve fire alarm and response capabilities and reduce equipment failure rate; flat layout and centralized monitoring simplify operation and maintenance process and reduce manpower input, and with adaptive optimization algorithm, realize full-process quantitative and intelligent dynamic control.

[0018] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 This is an overall flowchart of an embodiment of the scalable modular computing center building design method of the present invention; Figure 2 This is a schematic diagram of the storage device layout area according to an embodiment of the scalable modular computing center building design method of the present invention; Figure 3 This is a schematic diagram of the network equipment layout area according to an embodiment of the scalable modular computing center building design method of the present invention; Figure 4 This is a partial architectural layout diagram of the intermediate layer of an embodiment of the scalable modular computing center building design method of the present invention; Figure 5 This is an embodiment of the scalable modular computing center building design method of the present invention. Figure 4 Enlarged view of point A in the middle. Detailed Implementation

[0020] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0021] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0022] like Figure 1 As shown, a scalable modular computing center building design method includes the following steps: S1. Determine the overall structure of the computing center building and introduce a building space adaptability coefficient to assess whether the computing center building space meets the installation and operation requirements of computing equipment and liquid cooling system.

[0023] The building's overall structure adopts a three-story layout with a flat work environment design. The height under the beams is controlled at 5.6-6 meters, with 0.5-1 meter of space reserved under the floor for liquid cooling water pipes and at least 0.5 meters of space above the server racks to accommodate the liquid cooling system and cable routing requirements. Simultaneously, the building's exterior walls use integrated fire-resistant and thermally insulated decorative panels instead of traditional exterior walls. The specific structure and zoning of each floor are as follows: The ground floor is divided into multiple functional areas, including an entrance hall and the first pipe shaft area located in front of the entrance hall, as well as a spare parts warehouse for storing computing equipment, accessories and consumables, equipped with shelves and moisture-proof facilities.

[0024] One side of the first pipeline well area is equipped with an unpacking area and a testing area. The unpacking area is paved with scratch-resistant and wear-resistant floor tiles for unpacking and inspecting equipment upon arrival. The testing area is equipped with a testing bench and anti-static facilities for debugging new equipment and repairing old equipment. The other side of the pipeline well module is equipped with a maintenance module and an office module.

[0025] The entrance hall has symmetrically arranged storage device areas and network device areas on both sides, such as... Figure 2 As shown, the storage device layout area includes, in sequence, storage cabinets, indoor air conditioning units, storage cabinets, indoor air conditioning units, storage cabinets, indoor air conditioning units, air-cooled communication cabinets, and column-head power distribution cabinets; as shown... Figure 3 As shown, the network equipment deployment area is simultaneously arranged with network cabinets, air conditioning indoor units, network cabinets, network cabinets, air conditioning indoor units, network cabinets, air-cooled communication cabinets, and column-head power distribution cabinets; among them, the air conditioning indoor units are distributed to assist in overall heat dissipation.

[0026] Each side of the network equipment deployment area has an incoming line room equipped with lightning protection and grounding devices for the introduction of power and communication cables. Next to one incoming line room is a user workroom for maintenance personnel to work and guard. Next to the other incoming line room is a fire control center equipped with monitoring terminals and emergency response equipment for centralized management of the entire center's fire protection system. The sanitary area includes changing rooms, washrooms, and toilets. The exterior walls are made of perforated aluminum panels to meet the requirements of fire resistance, heat insulation, and ventilation.

[0027] like Figure 4 As shown, the middle layer is the core computing power deployment area, including the central control center security room and the second pipeline shaft area located in front of the central control center security room. Four integrated spray liquid cooling chambers are symmetrically arranged on both sides of the central control center security room and the second pipeline shaft area (maintaining indoor humidity at 40%-60%, adapting to the operating requirements of the computing power equipment); as shown... Figure 5 As shown, the integrated spray liquid cooling chamber is equipped with CDUs, eight liquid cooling cabinets, three indoor air conditioning units, air-cooled communication cabinets, and a row-head power distribution cabinet arranged in sequence. The first equipment platform is set up on one side of the central control center security control room and on one side of the second pipeline shaft area. Several GPU servers are set up on the first equipment platform. Safety railings (no less than 1.2 meters high to ensure the safety of maintenance personnel) are set up at the edge of the middle floor. Similarly, the exterior wall of the middle floor is made of perforated aluminum panel curtain wall.

[0028] The third layer is the energy supply and auxiliary support area, including the second equipment platform and the third pipeline well area located in front of the second equipment platform; two dry-type transformers and air-cooled refrigerant pump air conditioners are symmetrically arranged on both sides of the second equipment platform and the third pipeline well area. The dry-type transformers are equipped with noise reduction and heat dissipation facilities to provide stable power; the air-cooled refrigerant pump air conditioners are adapted to the heat dissipation system to centrally control the temperature of each layer.

[0029] Natural ventilation shafts are provided in the first, third, and fourth pipe shaft areas to assist indoor air circulation and reduce heat dissipation energy consumption. High-voltage electrical shafts, heating shafts, weak current shafts, and low-voltage shafts are symmetrically arranged on both sides of the natural ventilation shafts (aligned with the corresponding positions on the first floor to facilitate vertical laying of pipelines).

[0030] The rooftop is powered by solar energy, with distributed solar photovoltaic panels installed to supplement the power consumption of the central auxiliary system, reduce reliance on grid power, and meet the needs of green and low-carbon development. The area of ​​photovoltaic panels is precisely calculated based on the central auxiliary power demand to ensure power supply stability.

[0031] The computing center building space of this invention includes the height under the beams. Space reserved for liquid cooling water pipes under the floor and reserved space at the top of the rack By introducing a building space adaptability coefficient Assess whether the building space of the computing center (height under beams, reserved space for pipelines, etc.) meets the installation and operation requirements of computing equipment and liquid cooling systems to ensure spatial adaptability. This provides a quantitative basis for the overall structural design of the building, avoids insufficient or wasted space, ensures the convenience of subsequent equipment deployment, pipeline laying and operation and maintenance, and meets the standardized and refined design requirements of the modular computing center.

[0032] Building space adaptability coefficient The calculation formula is as follows: ; in, This represents the minimum space requirements for computing equipment and liquid cooling systems.

[0033] S2. Introduce a module expansion potential coefficient for module division and assembly deployment; Based on the architectural layout of the S1 computing center, the computing center is divided into the following six functional modules: (1) The computing power equipment module includes storage cabinets, network cabinets, air-cooled communication cabinets, GPU servers, liquid-cooled cabinets, CDUs, column-head power distribution cabinets and column-head accessory cabinets.

[0034] (2) The heat dissipation module includes a fluorinated phase change material heat dissipation unit, an integrated spray-type heat dissipation cabinet container, an air-cooled indoor unit, an air-cooled fluorine pump air conditioner, and a natural ventilation shaft.

[0035] (3) The energy module includes a dry-type transformer, a high-voltage well, a low-voltage well, distributed solar photovoltaic panels and an incoming line room.

[0036] (4) The control module includes a data acquisition terminal, a remote monitoring terminal, a low-voltage well, a user studio, and an office module.

[0037] (5) The safety module includes the central control center safety control room, fire control center, laser scattering and infrared sensor detector, heptafluoropropane gas fire extinguishing system, safety monitoring nodes, emergency response equipment, anti-static facilities, moisture-proof facilities, safety railings and lightning protection grounding devices.

[0038] (6) The waste heat recovery module includes a multi-stage waste heat recovery structure, a water source heat pump and a waste heat exchange component.

[0039] Prefabricated modular assembly blocks are used to produce each module, and the internal equipment of the modules is integrated in advance. On-site deployment of each module can only be completed by splicing together through standardized interfaces. Among them, the rack layout of the computing power module follows the minimum area of ​​2m² per rack. 2 (Meets the installation and heat dissipation space requirements of core computing power equipment), with a maximum length of 15 meters for a row of cabinets (to avoid uneven heat dissipation and inconvenience of operation and maintenance caused by excessively long layouts), and a passage width of no less than 1 meter (to ensure the passage of operation and maintenance personnel, equipment maintenance and emergency response space), laying the foundation for subsequent modular expansion and standardized operation and maintenance.

[0040] This invention introduces a module expansion potential coefficient to quantitatively evaluate the integration rationality and expansion capabilities of each functional module. This provides a precise quantitative basis for modular deployment and expansion scheme design, avoiding expansion difficulties and increased costs caused by non-standard module division and inconsistent interfaces. The formula for quantifying the module expansion potential coefficient is as follows: ; in, The module expansion potential coefficient indicates that the higher the module standardization level, the more space is reserved for expansion, and the stronger the expansion capability. To standardize the number of interfaces, the more standardized the interfaces, the more convenient it is to assemble and replace modules, avoiding expansion obstacles caused by interface incompatibility; This represents the total number of interfaces in the module, including standardized interfaces and dedicated interfaces. The higher the proportion of standardized interfaces, the greater the flexibility for expansion. The more space is reserved for splicing prefabricated assembly blocks, the more modules can be added and the greater the expansion space. This represents the total area of ​​the mounting blocks.

[0041] S3. Introduce network transmission loss coefficients for building layout and infrastructure optimization, specifically: A modular, flat layout is adopted, with computing equipment modules centrally arranged. A dedicated coolant storage area and exhaust system are provided. There are at least two power and communication cable entry points with a spacing of ≥15 meters. A network transmission loss coefficient is incorporated into the building layout and infrastructure optimization, as shown in the following formula: in, This is the network transmission loss coefficient. For the corresponding network protocol, the maximum transmission distance This represents the actual network transmission distance.

[0042] S4. Construct a composite heat dissipation system for heat dissipation management and layout.

[0043] A composite heat dissipation system is constructed by integrating integrated spray-type heat dissipation, phase change material self-generated heat dissipation, and liquid cooling technology, including the following steps: S41. The heat dissipation module adopts an integrated spray-type heat dissipation cabinet container structure, with multiple port interfaces. The spray system adjusts the spray intensity and frequency in real time according to the operating temperature of the computing equipment.

[0044] S42. The heat dissipation container is equipped with a self-generating heat dissipation unit constructed with fluorinated phase change material, which controls the temperature fluctuation of the chip surface within ±0.5℃ and keeps it stable below 60℃ for a long time.

[0045] S43. The heat dissipation system adopts a zoned independent control and global monitoring mode. The cooling source system supports mixed deployment of liquid cooling and air cooling, with a mainstream air-liquid ratio of 5:5 and the liquid side is planned for 100% full load.

[0046] This invention dynamically optimizes heat dissipation strategies by introducing heat dissipation control algorithms and temperature quantization formulas, including: (1) Dynamic control of spray intensity: The real-time spray intensity of the spray system is dynamically calculated based on the real-time heat generation of the computing equipment and the ambient temperature. The formula is: ; in, For real-time spray intensity, The basic spraying intensity is the equipment's rated operating condition. To generate heat in real time for the equipment, For the rated heat output of the equipment, For real-time ambient temperature, This refers to the standard ambient temperature.

[0047] (2) Quantify the chip surface temperature. Under the synergistic effect of phase change material and spray liquid cooling, the real-time temperature of the chip surface is measured. The calculation formula is: ; in, This represents the real-time heat dissipation of the spray liquid cooling system. This represents the real-time heat dissipation of the phase change material. For the specific heat capacity of the chip, For the core quality of the chip.

[0048] The formula for calculating the real-time heat dissipation of phase change materials is as follows: ; in, The thermal conductivity of the phase change material is... This refers to the contact area between the phase change material and the heat dissipation surface. The phase transition temperature of a phase change material. To determine the thickness of the phase change material; (3) Quantitative evaluation of heat dissipation efficiency, the formula is as follows: ; in, The overall heat dissipation efficiency of the composite heat dissipation system must be no less than 95%. This represents the total heat dissipation. To assist in heat dissipation for air cooling, This represents the total heat output of the equipment.

[0049] S5. An evaluation and optimization of energy utilization efficiency is conducted using a multi-stage waste heat recovery structure, which includes: Primary recycling is used for internal heating and equipment preheating in the computing center.

[0050] Secondary heat recovery is used for heating surrounding buildings. Waste heat recovery and heat dissipation are linked. Low-grade waste heat is extracted through a water source heat pump, improving the energy efficiency of the computing center. Below 1.05, introduce The calculation formula and waste heat recovery efficiency quantification model are used to evaluate and optimize energy utilization efficiency.

[0051] Among them, the energy efficiency of computing centers The calculation formula is as follows: ; in, To improve the energy efficiency of computing centers, This represents the total power consumption of the computing center. The core power consumption of IT equipment For the power consumption of the cooling system, For the power consumption of the power supply and distribution system, The power consumption of other auxiliary systems is in kW.

[0052] After introducing waste heat recovery, the energy utilization efficiency of the actual effective computing center is improved. The calculation formula is: ; in, The equivalent electrical energy from waste heat recovery is calculated from the recovered heat: ; In the formula, This represents the total amount of waste heat recovered. This refers to the energy efficiency ratio of a water source heat pump.

[0053] The waste heat recovery efficiency quantification model is shown in the following formula: ; in, The total waste heat recovery efficiency. This is for primary waste heat recovery. This is a secondary waste heat recovery process. This refers to the total waste heat discharged by the cooling system.

[0054] The formula for improving the efficiency of a water source heat pump is as follows: ; in, For water source heat pumps, the energy efficiency ratio must be ≥4. To enable the heat pump to output high-grade heat, Input electrical energy into the heat pump. This refers to low-grade waste heat extracted by a heat pump.

[0055] S6. Equipped with a precise, efficient, and low-pollution fire protection system, employing an intelligent fire alarm and heptafluoropropane gas extinguishing linkage mode, and using laser scattering and infrared sensor fusion detection, achieving an alarm accuracy rate of ≥99% and a fire extinguishing response time of 5-10 seconds; deploying several environmental safety monitoring nodes, combined with a safety risk quantitative assessment algorithm, to detect and dynamically assess safety hazards. The safety risk quantitative assessment is as follows: ; in, The safety risk value must be ≤0.1. To monitor the number of nodes, For the first The probability of a potential problem occurring at each node. For the first The impact range coefficient of potential hazards at each node. For the first The risk loss coefficient of each node.

[0056] S7. Construct a centralized control platform to adaptively adjust the operating status of each module, realize unified monitoring and independent control of each functional module. The centralized control platform has an adaptive adjustment algorithm that can automatically optimize the operating status of each module according to parameters such as computing load, ambient temperature, and energy consumption, and achieve the optimal balance of time, cost, efficiency, and energy.

[0057] This invention is based on A multi-objective optimization model is constructed with the objective function of maximizing the overall benefit coefficient: ; in, , , , For the weighting coefficients, satisfying ; The total heat dissipation efficiency of the composite heat dissipation system. The total waste heat recovery efficiency. Real-time operating cost, unit: yuan / hour For rated operating costs, Safety risk value, unit: yuan / hour.

[0058] S8. Rapidly expand the computing center by adding prefabricated installation blocks and replacing functional modules, and introduce expansion cost calculation and expansion cycle calculation models to evaluate expansion schemes.

[0059] The extended cost calculation model is expressed as follows: ; in, Total expansion cost, unit: yuan; Price is the unit price for a single module, in yuan / unit; To increase the number of modules, Installation cost per unit area, unit: yuan / m 3 , For the splicing installation area (m²) 2 , Cost savings through modular expansion, unit: yuan.

[0060] The extended cycle calculation model is expressed as follows: ; in, For the total expansion period, For the prefabrication and transportation cycle of individual modules, To increase the number of modules, For on-site assembly and installation time, The unit for module integration testing is days.

[0061] Example 1: Step 1: First, conduct the overall structural design of the computing center building: adopt a three-story flat layout, and determine the height under the beams to be 5.8 meters, the space under the floor for liquid cooling water pipes to be 0.8 meters, and the space above the server racks to be 0.6 meters by calculating the building space adaptability coefficient formula.

[0062] Assuming minimum space requirements for computing equipment and liquid cooling system =4.4m, substituting into the formula, we get =(5.8−0.8−0.6) / 4.4=1.2≥1, which meets the spatial adaptability requirements; the exterior walls of the building are made of fireproof and heat-insulating integrated decorative panels, and the bottom, middle and third floors are arranged according to the above detailed areas. The roof is covered with distributed solar photovoltaic panels, which shortens the construction period by 30%.

[0063] Step Two: Modular Division and Prefabricated Deployment. The system is divided into six functional modules, using precast concrete assembly blocks. The module expansion potential coefficient is calculated using a formula. It offers excellent scalability; the computing module has a single rack area of ​​2.5m². 2 Each row of cabinets is 12 meters long, with 1.2-meter-wide passages at both ends, and each module is connected via a standardized flange interface.

[0064] For example, a module has a total of 100 interfaces, of which 90 are standardized interfaces, and the total area of ​​the mounting block is 100m². 2 30m² reserved for splicing 2 ,but =(90 / 100)×(30 / 100)=0.27, indicating weak expansion potential; while this invention increases the number of standardized interfaces to 92 by increasing the number of optimized interfaces, while reserving sufficient splicing space, taking the reserved splicing area ratio as 1 (that is, all mounting blocks have reserved space for splicing, with no space redundancy or waste, achieving...) =0.92, which can quickly respond to the demand for computing power growth and can be expanded by adding new modules without modifying the original building and module structure.

[0065] Step 3: Building layout and infrastructure optimization. The computing equipment modules are centrally located in the middle of the building's ground floor. Based on the network transmission loss coefficient formula, the actual transmission distance is 45 meters. =90%≤100%, no transmission exceeding limits; two power / communication inlet rooms are set up, spaced 18 meters apart, and the coolant storage area is equipped with an explosion-proof exhaust system.

[0066] Step 4: The heat dissipation module adopts an integrated spray-type heat dissipation cabinet container structure, with 4 spray ports. The basic spray intensity of 5L / min is determined through a dynamic control algorithm, and is adjusted in real time according to the heat generated by the equipment and the ambient temperature. The container is filled with fluorinated phase change material, and the cabinet temperature is stabilized at 40-55℃ through a chip surface temperature quantification model. The air-liquid ratio of the cooling source system is 5:5, and the liquid side is planned for 100% full load. =98.5% heat dissipation efficiency calculated using the formula.

[0067] Step 5: Employing a two-stage preheating and recovery structure, the waste heat recovery efficiency is calculated using a quantitative formula. =82%≥80%; calculated using the precise PUE measurement formula. =1050kW, =1000kW, foundation =1.05, after introducing waste heat recovery =1.02, achieving efficient energy utilization; primary recovery is used for computer room heating, and secondary recovery raises the waste heat to 50°C for heating of surrounding office buildings.

[0068] Step 6: Equip with an intelligent fire alarm and heptafluoropropane gas extinguishing system, using laser scattering and infrared sensor fusion detection. =0.08≤0.1 The safety risk quantification assessment formula is used to calculate that the alarm accuracy rate is 99.2%, and the extinguishing agent is released within 8 seconds after the fire is confirmed; 50 dynamic and environmental safety monitoring nodes are deployed, and the fire protection system adopts a dual power supply redundancy design.

[0069] Step 7: Construct a centralized control platform, embed an adaptive adjustment multi-objective optimization algorithm, and set weight coefficients. =0.3、 =0.3、 =0.25、 =0.15, real-time data collection of computing load, temperature and humidity, energy consumption, etc., and remote monitoring via computer client / mobile APP, the platform automatically optimizes operating parameters. The overall benefit coefficient remains stable at over 0.95.

[0070] Step 8: When computing power demand increases by 50%, the cost and cycle calculation model determines the need for 2 additional computing power modules and 1 cooling module. The total expansion cost is reduced by 42% compared to the traditional method, and the total expansion cycle is 12 days. =12≤15 days); modules can be quickly assembled through standardized interfaces without modifying the original building structure, and each module can be upgraded and maintained independently.

[0071] In this embodiment, the quantitative formula and algorithm model of the present invention achieve full-process design optimization, shortening the construction cycle of the computing center by 40% compared with the traditional method, reducing operation and maintenance costs by 35%, improving energy utilization efficiency by 25%, improving heat dissipation efficiency by 45%, and reducing equipment failure rate by 60%. This realizes the quantitative design, precise control and dynamic optimization of the computing center, adapting to the long-term development needs of artificial intelligence computing centers.

[0072] Therefore, the present invention adopts the above-mentioned scalable modular computing center building design method, which effectively solves various technical pain points in the existing computing center building design. It achieves full-process optimization through a multi-dimensional quantitative model, improves efficiency and reduces costs by relying on prefabricated and standardized design, takes into account energy saving, low carbon emissions and safety and stability, adapts to the development needs of artificial intelligence computing centers, and has high practical value and promotion prospects.

[0073] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A scalable modular computing center building design method, characterized in that, Includes the following steps: S1. Determine the overall structure of the computing center building and introduce a building space adaptability coefficient to assess whether the computing center building space meets the installation and operation requirements of computing equipment and liquid cooling system. S2. Introduce a module expansion potential coefficient for module division and assembly deployment; S3. Introduce network transmission loss coefficient for building layout and infrastructure optimization. S4. Construct a composite heat dissipation system for heat dissipation management and layout; S5. Evaluate and optimize energy utilization efficiency by adopting a multi-stage waste heat recovery structure; S6. Deploy several environmental safety monitoring nodes and combine them with safety risk quantification assessment algorithms to detect and dynamically assess safety hazards; S7. Construct a centralized control platform to adaptively adjust the operating status of each module; S8. Rapidly expand the computing center by adding prefabricated installation blocks and replacing functional modules, and introduce expansion cost calculation and expansion cycle calculation models to evaluate expansion schemes.

2. The scalable modular computing center building design method according to claim 1, characterized in that, In S1, the computing center building space includes the height under the beams. Space reserved for liquid cooling water pipes under the floor and reserved space at the top of the rack ; By introducing a building space adaptability coefficient The formula for assessing whether the building space of a computing center meets the installation and operation requirements of computing equipment and liquid cooling systems is as follows: ； in, This represents the minimum space requirements for computing equipment and liquid cooling systems.

3. The scalable modular computing center building design method according to claim 2, characterized in that, In S2, the computing center is divided into the following sections based on the building layout of the computing center in S1: The computing equipment module includes storage cabinets, network cabinets, air-cooled communication cabinets, GPU servers, liquid-cooled cabinets, CDUs, column-head power distribution cabinets, and column-head accessory cabinets; The heat dissipation module includes a fluorinated phase change material heat dissipation unit, an integrated spray-type heat dissipation cabinet container, an air conditioning indoor unit, an air-cooled fluorine pump air conditioner, and a natural ventilation shaft; The energy module includes a dry-type transformer, a high-voltage well, a low-voltage well, distributed solar photovoltaic panels, and an incoming line room; The control module includes a data acquisition terminal, a remote monitoring terminal, a low-voltage electrical well, a user work area, and an office module; The safety module includes a central control center safety control room, a fire control center, laser scattering and infrared sensor detectors, a heptafluoropropane gas fire extinguishing system, safety monitoring nodes, emergency response equipment, anti-static facilities, moisture-proof facilities, safety railings, and lightning protection grounding devices; The waste heat recovery module includes a multi-stage waste heat recovery structure, a water source heat pump, and waste heat exchange components; Prefabricated assembly blocks are used to produce the functions of each module, and the functions of each module are assembled through standardized interfaces to complete the deployment. The rack layout of the computing modules follows a minimum single rack area of ​​2m². 2 The maximum length of a row of server racks is 15 meters, and the width of the aisle is not less than 1 meter; The integration rationality and expansion capability of each functional module are quantitatively evaluated by introducing a module expansion potential coefficient. The quantitative formula for the module expansion potential coefficient is as follows: ； in, Increase the module's potential coefficient; To standardize the number of interfaces; This represents the total number of interfaces in the module. Reserve splicing area for prefabricated assembly blocks; This represents the total area of ​​the mounting blocks.

4. The scalable modular computing center building design method according to claim 3, characterized in that, S3 specifically refers to: A modular, flat layout is adopted, with computing equipment modules centrally arranged. A network transmission loss coefficient is introduced to optimize building layout and infrastructure, as shown in the following formula: in, This is the network transmission loss coefficient. For the corresponding network protocol, the maximum transmission distance This represents the actual network transmission distance.

5. The scalable modular computing center building design method according to claim 4, characterized in that, In S4, constructing a composite heat dissipation system includes the following steps: S41. The heat dissipation module adopts an integrated spray-type heat dissipation cabinet container structure, with multiple port interfaces. The spray system adjusts the spray intensity and frequency in real time according to the operating temperature of the computing equipment. S42. The heat dissipation container is equipped with a self-generating heat dissipation unit constructed with fluorinated phase change material. S43. The heat dissipation system adopts a zoned independent control and global monitoring mode. The cooling source system supports mixed deployment of liquid cooling and air cooling, with a mainstream air-liquid ratio of 5:5 and the liquid side is planned for 100% full load.

6. The scalable modular computing center building design method according to claim 5, characterized in that, In S4, a heat dissipation control algorithm and a temperature quantization formula are introduced to dynamically optimize the heat dissipation strategy, including: The spray intensity is dynamically controlled by calculating the real-time spray intensity of the spray system based on the real-time heat generation of the computing equipment and the ambient temperature. The formula is: ； in, For real-time spray intensity, Based on the basic spray intensity, To generate heat in real time for the equipment, For the rated heat output of the equipment, For real-time ambient temperature, Standard ambient temperature; Quantifying chip surface temperature: Real-time chip surface temperature under the synergistic effect of phase change material and spray liquid cooling. The calculation formula is: ； in, This represents the real-time heat dissipation of the spray liquid cooling system. This represents the real-time heat dissipation of the phase change material. For the specific heat capacity of the chip, For the core quality of the chip; The formula for calculating the real-time heat dissipation of phase change materials is as follows: ； in, The thermal conductivity of the phase change material is... This refers to the contact area between the phase change material and the heat dissipation surface. The phase transition temperature of a phase change material. To determine the thickness of the phase change material; The formula for quantifying heat dissipation efficiency is as follows: ； in, The total heat dissipation efficiency of the composite heat dissipation system. This represents the total heat dissipation. To assist in heat dissipation for air cooling, This represents the total heat output of the equipment.

7. The scalable modular computing center building design method according to claim 6, characterized in that, In S5, the multi-stage waste heat recovery structure includes: Primary recycling is used for internal heating and equipment preheating in the computing center; Secondary heat recovery is used for heating surrounding buildings. Waste heat recovery and heat dissipation are linked. Low-grade waste heat is extracted through a water source heat pump, improving the energy efficiency of the computing center. Below 1.05, introduce The calculation formula and waste heat recovery efficiency quantification model are used to evaluate and optimize energy utilization efficiency. Among them, the energy efficiency of computing centers The calculation formula is as follows: ； in, To improve the energy efficiency of computing centers, This represents the total power consumption of the computing center. The core power consumption of IT equipment For the power consumption of the cooling system, For the power consumption of the power supply and distribution system, Power consumption for other auxiliary systems; After introducing waste heat recovery, the actual effective energy utilization efficiency of the computing center The calculation formula is: ； in, The equivalent electrical energy from waste heat recovery is calculated from the recovered heat: ； In the formula, This represents the total amount of waste heat recovered. The energy efficiency ratio of a water source heat pump; The waste heat recovery efficiency quantification model is shown in the following formula: ； in, The total waste heat recovery efficiency. This is for primary waste heat recovery. This is a secondary waste heat recovery process. The total waste heat discharged by the cooling system; The formula for improving the efficiency of a water source heat pump is as follows: ； in, For the energy efficiency ratio of a water source heat pump, To enable the heat pump to output high-grade heat, Input electrical energy into the heat pump. This refers to low-grade waste heat extracted by a heat pump.

8. The scalable modular computing center building design method according to claim 7, characterized in that, In S6, an intelligent fire alarm and heptafluoropropane gas extinguishing linkage mode are adopted, using laser scattering and infrared sensing fusion detection. Several dynamic and environmental safety monitoring nodes are deployed, and combined with a safety risk quantitative assessment algorithm, safety hazard detection and dynamic assessment are carried out. The safety risk quantitative assessment is as follows: ； in, For safety risk values, To monitor the number of nodes, For the first The probability of a potential problem occurring at each node. For the first The impact range coefficient of potential hazards at each node. For the first The risk loss coefficient of each node.

9. The scalable modular computing center building design method according to claim 8, characterized in that, In S7, with A multi-objective optimization model is constructed with the objective function of maximizing the overall benefit coefficient: ； in, , , , For the weighting coefficients, satisfying ; The total heat dissipation efficiency of the composite heat dissipation system. The total waste heat recovery efficiency. For real-time operating costs, For rated operating costs, This represents a safety risk value.

10. The scalable modular computing center building design method according to claim 9, characterized in that, In S8, the extended cost calculation model is expressed as follows: ； in, For total expansion costs, Price per module To increase the number of modules, Installation cost per unit area For splicing and installation area, Save costs by enabling modular expansion; The extended cycle calculation model is expressed as follows: ； in, For the total expansion period, For the prefabrication and transportation cycle of individual modules, To increase the number of modules, For on-site assembly and installation time, This refers to the module integration and testing cycle.

Images