Flow control method for cabinet-level pump-driven two-phase flow cooling system

Abstract

The invention relates to a flow control method for a cabinet-level pump-driven two-phase flow cooling system, and belongs to the technical field of cooling systems. Comprising a pump set, a shell and tube condenser, 2-16 cabinet cold plate branches, a condenser inlet temperature and pressure sensor, a condenser outlet temperature and pressure sensor, a flow meter and an electric control valve, wherein the pump set is applied to a cabinet-level pump-driven two-phase flow cooling system and comprises two pumps which are connected in parallel; the method further comprises the following steps: S1, collecting multiple parameters; s2, dynamic target flow calculation; s3, multi-dimensional closed-loop flow regulation is carried out; s4, pump set-flow cooperative control is carried out; and S5, blocking and cavitation early warning and self-repairing are carried out. According to the flow control method of the cabinet-level pump-driven two-phase flow cooling system, through multi-parameter collection and dynamic target flow calculation and in combination with PID closed-loop adjustment, real-time matching of the flow and the thermal load is achieved, and the heat dissipation efficiency is improved; by adopting a flow-temperature-pressure three-parameter decoupling PID control strategy, the response speed of the system can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of cooling system technology, specifically to a flow control method for a cabinet-level pump-driven two-phase flow cooling system. Background Technology

[0002] In existing technologies, rack-level server cooling often employs air cooling or a single liquid cooling solution, which suffers from low cooling efficiency, high noise, and high energy consumption. As server power density continues to increase, traditional cooling methods are insufficient to meet the cooling demands of high-density racks. While pump-driven two-phase flow cooling technology can provide higher cooling efficiency, it still faces the following technical bottlenecks in practical applications:

[0003] 1. Insufficient flow control precision: Existing solutions mostly use fixed flow or simple PID control, which cannot adapt to dynamically changing server heat load, resulting in energy waste at low load and insufficient heat dissipation at high load;

[0004] 2. System response lag: When the thermal load changes abruptly, traditional control strategies respond slowly, which can easily lead to local overheating and affect server stability;

[0005] 3. Uneven flow distribution across multiple branches: The heat load of each branch in a multi-node cabinet varies greatly, and traditional systems are prone to "flow competition", resulting in thermal imbalance.

[0006] 4. Lack of intelligent early warning and self-repair mechanism: The system lacks real-time judgment and adaptive adjustment capabilities for abnormal operating conditions such as pipeline blockage and pump cavitation, resulting in low system reliability;

[0007] 5. Poor environmental adaptability: Existing solutions exhibit significant performance degradation under extreme high and low temperature environments and lack adaptive adjustment mechanisms.

[0008] Therefore, there is an urgent need for a cabinet-level pump-driven two-phase flow cooling system flow control method that can achieve high precision, fast response, intelligent early warning, multi-branch coordination, and environmental adaptability. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides a flow control method for a rack-level pump-driven two-phase flow cooling system. This method offers advantages such as dynamic and precise flow control, rapid response to changes in heat load, multi-branch coordinated heat dissipation, intelligent early warning and self-repair, strong environmental adaptability, and significant energy-saving effects. It solves the problems of slow response, high energy consumption, uneven heat distribution, poor reliability, and weak environmental adaptability in high-density rack cooling using traditional fixed flow or simple PID control methods.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A flow control method for a rack-level pump-driven two-phase flow cooling system includes a pump group with two pumps in parallel, a shell-and-tube condenser, 2-16 integrated and fitted simulated heat sources for the rack-level pump-driven two-phase flow cooling system, a condenser inlet temperature and pressure sensor, a condenser outlet temperature and pressure sensor, a flow meter, and an electric regulating valve.

[0012] in;

[0013] The pump set with the two pumps in parallel is installed in the medium output pipeline of the condenser to provide circulation power for the two-phase flow working fluid, and the power of the pump set is dynamically adapted to the temperature of the cold plate.

[0014] The inlet pipe of the shell-and-tube condenser is equipped with an inlet temperature and pressure sensor to collect inlet pressure data. and inlet temperature The outlet pipeline is equipped with an outlet temperature and pressure sensor to collect outlet pressure data. and outlet temperature ,and The pressure difference is used for pipeline identification;

[0015] The flow meter is installed at the front end of the electric regulating valve to collect the real-time flow of the system, and the flow data and temperature and pressure data form a multi-dimensional feedback.

[0016] Each cabinet cold plate branch is connected in parallel to the output pipeline of the pump group. Each branch is equipped with an independent temperature sensor to collect the surface temperature of the cold plate. The attached simulated heat source is attached to the heat absorption surface of the cold plate with thermal grease. The bonding gap is no more than 0.1mm and the thermal resistance is no more than 0.05K / W.

[0017] The flow control method includes the following steps:

[0018] S1. Multi-parameter acquisition: Data is acquired via condenser inlet / outlet temperature and pressure sensors. The surface temperature of the cold plate is collected by temperature sensors in each cold plate branch, and the highest temperature is selected. By collecting the real-time operating power of the simulated heat sources attached to each cold plate, the heat load of the corresponding cold plate branch can be obtained. to Real-time flow rate is collected by the flow meter system. ;

[0019] S2. Dynamic target flow rate calculation: based on the average operating pressure of the cooling medium in the system. Average isobaric specific heat capacity Using the formula

[0020]

[0021] Calculate target flow ;in This is a dynamic thermal margin correction factor with an initial value of 1.2, which is adjusted in real time according to the phase change degree of the two-phase flow and the pipeline resistance.

[0022] S3, Multi-dimensional Closed-Loop Flow Regulation: Based on " and "The deviation" is a proportional term, " The rate of change is the differential term. The deviation is taken as the integral term, and a PID closed-loop control strategy is used to output the adjustment amount of the electric regulating valve opening, so that the actual flow rate matches the target flow rate. ;

[0023] S4, Pump Set-Flow Coordinated Control: Based on The pump power is adjusted in stages based on the comparison results with the preset threshold.

[0024] S5. Blockage and Cavitation Early Warning and Self-Healing: Through... The pressure difference threshold is used to determine the risk of pipeline blockage. and "deviation rate" combined The risk of cavitation is determined by comparing the saturated vapor pressure of the cooling medium. After triggering an early warning, the valve opening and pump unit operating status are automatically adjusted to achieve self-repair.

[0025] Furthermore, in step S2, the dynamic thermal margin correction coefficient The adjustment logic is as follows:

[0026] when hour, Adaptively adjusted to 1.0-1.1;

[0027] when hour, The adaptive adjustment is set to 1.2-1.3;

[0028] When the system is in steady state, that is and hour, Keep it at 1.2;

[0029] The coefficient “1.2” is determined based on experimental data of the latent heat of phase change of the cooling medium under the system working pressure and the friction loss along the pipeline, and is a phase change heat fluctuation compensation coefficient obtained by orthogonal experimental fitting.

[0030] Furthermore, in step S3, the PID closed-loop control employs a "parameter adaptive optimization mechanism":

[0031] The adjustment cycle is synchronized with the flow meter sampling cycle, and the value ranges from 0.5s to 2s.

[0032] when When the preset heat exchange efficiency range of 2K-8K is exceeded, the PID proportional coefficient is temporarily increased to 1.2-1.5 times the original value, the derivative coefficient remains unchanged, and the integral coefficient is reduced to 0.8-0.9 times the original value.

[0033] when When the flow rate returns to the preset range and the flow deviation is ≤±3%, the PID parameters will automatically recover to their initial values.

[0034] Furthermore, in step S4, the specific threshold and execution logic of the pump group power grading adjustment strategy are as follows:

[0035] High-temperature emergency zone: when At that time, the pump set immediately switches to full power operation, and the two pumps start synchronously and output in coordination;

[0036] Normal working range:

[0037] when At that time, the pump set operating power and It exhibits linear positive correlation adjustment, and the adjustment formula is: ,in 70% of the rated power This is the power regulation coefficient, with a value ranging from 0.02 to 0.05 kW / K;

[0038] Low temperature energy saving range: when When the pump set is switched to 40%-60% of its rated power, only one pump works while the other is on standby.

[0039] Furthermore, in step S5, the criteria for judging blockage and cavitation and the self-healing logic are as follows:

[0040] Pipeline blockage risk: When If the blockage persists for more than 3 seconds, it is considered a risk of pipeline blockage. The system will automatically increase the opening of the electric regulating valve by 10%-20% while reducing the pump power by 10%-15%. If the blockage is resolved within 30 seconds after adjustment... If the device fails to return to the safe zone, an audible and visual alarm will be triggered.

[0041] Pump cavitation risk: When and When the pressure is lower than the saturated vapor pressure of the cooling medium, it is considered a risk of cavitation. The system will automatically reduce the opening of the electric regulating valve by 5%-10% and start the standby pump to operate in coordination. If the flow deviation is not reduced within 10 seconds after adjustment, the pump set will automatically reduce the frequency to 50% of the rated power.

[0042] Furthermore, in step S2, the average isobaric specific heat capacity of the cooling medium The "dynamic lookup table method" is used to obtain the data based on the system's real-time average pressure. and the present The corresponding medium state is dynamically matched from the pre-stored thermodynamic parameter table of the cooling medium. value;

[0043] The two-phase flow medium of the cooling system is R134a. The dynamic adjustment range is 1.2-1.8 kJ / (kg・K).

[0044] Furthermore, the adaptability design and traffic distribution logic of the cabinet cold plate branch are as follows:

[0045] The number of branch circuits can be flexibly configured from 2 to 16, and it is compatible with 4U-42U server racks. Each cold plate branch circuit is used to cool 1-2 server nodes.

[0046] Heat load of each branch to It can be directly acquired by collecting the real-time operating power signal of the simulated heat source, with a collection response time of ≤0.1s, without the need for additional configuration of a heat flux density sensor;

[0047] When adding or removing cold plate branches, the control system automatically identifies the change in the number of branches and redistributes the target flow based on the real-time heat load ratio of each branch, avoiding the problem of multiple branches "competing for flow" or thermal imbalance.

[0048] Furthermore, in step S3, the PID closed-loop control adopts a three-parameter decoupling logic of "flow rate-temperature-pressure":

[0049] The core control parameter is the "deviation between actual traffic and target traffic," with a weighting of 60%.

[0050] by" The "rate of change of temperature difference" is used as an auxiliary adjustment parameter, with a weighting of 25%.

[0051] by" "Pressure difference deviation" is used as an interference compensation parameter, with a weighting of 15%.

[0052] The control system calculates the weighting coefficients of each parameter in real time and eliminates multi-parameter coupling interference through a decoupling algorithm, achieving a temperature control accuracy of ±0.4℃.

[0053] Furthermore, the pump set with dual pumps in parallel is equipped with "master / standby coordinated switching logic":

[0054] During steady-state operation, only the main pump operates, while the standby pump remains in standby mode, and the main pump power is determined according to... Dynamic adjustment;

[0055] when When cavitation risk is triggered, the standby pump will start automatically, and the two pumps will work together, increasing the total power of the pump set to 1.8-2.0 times the rated power of a single pump;

[0056] when Once the system returns to normal operating range and the cavitation risk is eliminated, the standby pump will shut down after a 10-second delay.

[0057] When the main pump fails, i.e. the flow rate drops by ≥50% or the pump body temperature is ≥85℃, the standby pump will automatically switch to the main pump operation within 0.5s to ensure continuous cooling of the system.

[0058] Compared with the prior art, the present invention provides a flow control method for a rack-level pump-driven two-phase flow cooling system, which has the following beneficial effects:

[0059] 1. The flow control method of this rack-level pump-driven two-phase flow cooling system achieves real-time matching of flow and heat load by acquiring multiple parameters (temperature, pressure, flow, and heat load) and calculating dynamic target flow, combined with PID closed-loop regulation, thereby improving heat dissipation efficiency. By adopting a "flow-temperature-pressure" three-parameter decoupled PID control strategy, the response time is shortened by more than 40%, adapting to sudden changes in heat load and effectively improving the system response speed.

[0060] 2. The flow control method of this cabinet-level pump-driven two-phase flow cooling system automatically reduces power at low loads by more than 18% through graded control of pump power and adjustment of dynamic thermal margin coefficient; and avoids the phenomenon of "flow competition" by real-time acquisition of branch heat load and intelligent flow distribution, thereby improving the temperature uniformity of cold plate by 25%.

[0061] 3. The flow control method of this cabinet-level pump-driven two-phase flow cooling system can judge the risk of pipeline blockage and pump cavitation in real time by pressure difference and flow deviation, and automatically adjust the valve opening and pump status, which can effectively improve system reliability and has intelligent early warning and self-repair capabilities.

[0062] 4. The flow control method of this cabinet-level pump-driven two-phase flow cooling system ensures that the cooling system can maintain stable operation in extreme high and low temperature environments without the risk of overheating or solidification. It adapts to complex deployment environments and can effectively enhance environmental adaptability.

[0063] 5. The flow control method of this rack-level pump-driven two-phase flow cooling system allows for flexible configuration of the number of branches from 2 to 16, making it compatible with various server rack specifications and providing strong system scalability. Attached Figure Description

[0064] Figure 1 This is a flow control structure diagram of a flow control method for a cabinet-level pump-driven two-phase flow cooling system according to the present invention.

[0065] Figure 2 This is a flow control flowchart of a flow control method for a cabinet-level pump-driven two-phase flow cooling system according to the present invention.

[0066] Figure 3 The diagram shows the pump power regulation characteristic curve of the flow control method for a rack-level pump-driven two-phase flow cooling system according to the present invention.

[0067] Figure 4 This is a block diagram of the PID controller structure for a flow control method of a cabinet-level pump-driven two-phase flow cooling system according to the present invention.

[0068] Figure 5 This is a schematic diagram of the arrangement of electric valves and temperature sensors in the flow control method of a cabinet-level pump-driven two-phase flow cooling system according to the present invention. Detailed Implementation

[0069] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0070] Please see Figures 1 to 5 .

[0071] Example 1: Typical server rack (42U, 16 nodes)

[0072] System configuration: Coolant is R134a, heat load range is 5kW–30kW, control cycle is 1s.

[0073] Implementation steps:

[0074] S1. Multi-parameter acquisition: Data is acquired via condenser inlet / outlet temperature and pressure sensors. The surface temperature of the cold plate is collected by temperature sensors in each cold plate branch, and the highest temperature is selected. By collecting the real-time operating power of the simulated heat sources attached to each cold plate, the heat load of the corresponding cold plate branch can be obtained. to Real-time flow rate is collected by the flow meter system. ;

[0075] S2. Dynamic target flow rate calculation: based on the average operating pressure of the cooling medium in the system. Average isobaric specific heat capacity Using the formula

[0076]

[0077] Calculate target flow ;in This is a dynamic thermal margin correction factor with an initial value of 1.2, which is adjusted in real time according to the phase change degree of the two-phase flow and the pipeline resistance.

[0078] S3, Multi-dimensional Closed-Loop Flow Regulation: Based on " and "The deviation" is a proportional term, " The rate of change is the differential term. The deviation is taken as the integral term, and a PID closed-loop control strategy is used to output the adjustment amount of the electric regulating valve opening, so that the actual flow rate matches the target flow rate. ;

[0079] S4, Pump Set-Flow Coordinated Control: Based on The pump power is adjusted in stages based on the comparison results with the preset threshold.

[0080] S5. Blockage and Cavitation Early Warning and Self-Healing: Through... The pressure difference threshold is used to determine the risk of pipeline blockage. and "deviation rate" combined The risk of cavitation is determined by comparing the saturated vapor pressure of the cooling medium. After triggering an early warning, the valve opening and pump unit operating status are automatically adjusted to achieve self-repair.

[0081] Experimental results: Compared with fixed flow control, the system response time is shortened by 40%, energy consumption is reduced by 18%, and temperature uniformity is improved by 25%.

[0082] Example 2: Sudden Heat Load Test

[0083] Test system configuration:

[0084] Test subject: 42U standard server rack, configured with 16 rack cold plate branches, each branch corresponding to 1 server node;

[0085] Cooling medium: R134a;

[0086] System parameters:

[0087] Pump set: Two pumps in parallel, with a rated power of 1.5kW per pump;

[0088] Control cycle: 1 second;

[0089] Initial heat load: uniform load at all nodes, total heat load 20kW;

[0090] Initial cold plate temperature: 38℃±1.5K;

[0091] Sudden change condition: Simulate a sudden high load at node 8 (branch), and the power of the corresponding simulated heat source increases from 1.25kW to 1.875kW (an increase of 50%) within 1 second, and the total heat load of the system increases from 20kW to 20.625kW accordingly.

[0092] Testing process and data collection:

[0093] A1. The system runs stably for 5 minutes under a uniform load of 20kW, and the baseline data of each sensor is recorded.

[0094] A2. Trigger the power surge command at node 8 at t=0s;

[0095] A3. Collect the following data at high speed at 100ms intervals for 10 seconds:

[0096] Temperature of each cold plate branch (focus on monitoring node 8);

[0097] System real-time traffic ( );

[0098] Condenser inlet and outlet temperatures ( ) and pressure ( );

[0099] Electric regulating valve opening;

[0100] Pump unit operating status and power.

[0101] The response process of the control method of this invention:

[0102] B1. Real-time acquisition of multiple parameters and identification of heat load (t=0-0.2s):

[0103] The system collected the heat load at node 8 within 0.1 seconds. The sudden increase in signals;

[0104] Meanwhile, the temperature sensor on the cold plate at this node detected a temperature rise trend.

[0105] B2. Dynamic target flow recalculation and PID control (t=0.2-1.5s):

[0106] The control system is based on the new total heat load. Combined with real time , Using the formula

[0107]

[0108] Recalculate the target flow G;

[0109] The PID controller is based on "flow deviation" and combines it with " "rate of change" and " "Deviation", adjusting the opening of the electric regulating valve from 52% to 68% within 1 second;

[0110] actual system traffic The flow rate was increased from 22.4 L / min to 28.1 L / min to meet new heat dissipation requirements.

[0111] B3. Pump group power coordinated adjustment (t=1.5-3.0s):

[0112] The system detected the temperature of the cold plate at node 8. The temperature rises to near the preset threshold (rated temperature * 1.02).

[0113] According to the linear adjustment strategy of the "normal working range" in claim 4, the total power of the pump unit is gradually increased from 2.1kW to 2.7kW, thereby enhancing the circulation power.

[0114] B4. System stability and establishment of a new equilibrium (t=3.0-10.0s):

[0115] After adjusting the flow rate and pump power, the temperature of the cold plate at node 8 reached a peak of 41.8℃ at t=3s, and then began to drop.

[0116] At t=6s, the temperature at this node stabilized at 40.2℃, and the maximum temperature difference between nodes recovered to within 3K;

[0117] The system has entered a new thermal equilibrium state, and no overheating alarm has been triggered.

[0118] The test results and key data are shown in Table 1.

[0119] Table 1

[0120] index Test Results Remark System response time ≤3.0s The total time from the sudden increase in heat load to the temperature starting to drop Maximum temperature fluctuation <2K Difference between peak and initial temperature of the cold plate at node 8 Flow regulation overshoot ≤5% The maximum deviation of G from the new target flow G Pump power regulation delay 1.5s Time from recognizing changes in T_max to starting power adjustment Does this trigger an alarm? no No overheating, blockage, or cavitation alarms were detected throughout the entire process. Time to return to steady state 6s Time from the onset of the mutation to the temperature of all nodes stabilizing

[0121] This embodiment demonstrates that the flow control method of the present invention can respond quickly, accurately, and stably to sudden high-load events of server nodes. Through the control logic of "real-time thermal load identification → dynamic recalculation of target flow → PID multi-parameter closed-loop adjustment → pump power coordination", the system can complete the heat dissipation capacity adaptation within 3 seconds, effectively suppressing temperature fluctuations, avoiding local overheating, and significantly improving the heat dissipation reliability and safety of high-density cabinets under dynamic loads.

[0122] Example 3: High and Low Temperature Environmental Adaptability Test (Supplementary Scenario)

[0123] Test conditions:

[0124] Test object: The same 42U server rack (16 nodes) as in Example 1, with the same cooling system and the same working fluid, R134a;

[0125] Environmental conditions: Simulates extreme high and low temperature environments, with low temperature conditions at -10℃, normal temperature conditions at 25℃, and high temperature conditions at 45℃;

[0126] Heat load: Fixed at medium load 15kW (covering the daily operating conditions of most servers);

[0127] Control parameters: The control cycle remains 1s, and the PID coefficients are (Kp=5.2, Ki=0.8, Kd=1.5, adapted to the thermodynamic characteristics of R134a).

[0128] Experimental steps:

[0129] C1. Start the system at -10℃, 25℃, and 45℃ respectively, and run it stably for 30 minutes;

[0130] C2. Data is collected every 1 second: temperature of each cold plate, system flow rate, pump power, and inlet and outlet temperature and pressure of the condenser;

[0131] C3. Record the system's temperature fluctuation range, energy consumption, and response stability (duration without alarm) under different environments.

[0132] 3. Experimental data are shown in Table 2.

[0133] Table 2

[0134] Ambient temperature Cold plate temperature fluctuation range System energy consumption (kW·h / 30min) Stable running time without alarms Key conclusions -10℃ (low temperature) ＜1.5K 2.8 30 minutes (stable throughout) The low-power control of the pump unit was effective at low temperatures, and no risk of solidification of R134a was observed. 25℃(normal temperature) ＜1K 2.2 30 minutes (stable throughout) The system achieves optimal efficiency and temperature uniformity at room temperature. 45℃ (high temperature) ＜2K 3.5 30 minutes (stable throughout) The dual-pump coordinated control works effectively at high temperatures, with no overheating alarms.

[0135] Comparative Example 1: Fixed Flow Control Scheme (Compared with the Invention)

[0136] Comparison criteria:

[0137] Test subject: exactly the same as in Example 1 (42U rack, 16 nodes, R134a working fluid).

[0138] Control scheme: The existing technology of "fixed flow control" is adopted (no PID dynamic adjustment is performed, the flow rate is fixed at 15L / min, and the pump set always operates at the rated power of a single pump).

[0139] Test environment: same as in Example 1, ambient temperature (25℃), heat load range 5kW-30kW, test duration 30min, control cycle 1s (consistent with the present invention, to ensure fair comparison).

[0140] Experimental steps:

[0141] D1. Start the fixed flow control system, set the flow rate to 15L / min, and operate the pump set at its rated power.

[0142] D2. The heat load is gradually increased from 5kW to 30kW (5kW increase every 5min), and data is collected every 1s: temperature of each cold plate, system energy consumption, and response time;

[0143] D3. Record the system performance under different heat loads and compare it with the dynamic PID control scheme in Example 1.

[0144] 3. The experimental data and comparison results are shown in Table 3.

[0145] Table 3

[0146] Test metrics Comparative Example 1 (Constant Flow Rate) Example 1 (PID Dynamic Control) Advantages (relative improvements) of this invention System response time (when heat load changes from 5kW to 30kW) 8.5s 5.1s Shortened by 40% (echoing data from Example 1) Total system energy consumption (kW·h) over 30 minutes 12.5 10.25 A reduction of 18% (echoing data from Example 1). Temperature uniformity of cold plate (maximum temperature difference) 6.8K 5.1K A 25% improvement (echoing data from Example 1) Maximum temperature of cold plate under heat load of 30kW 52.3℃ (close to the overheat alarm threshold) 45.8℃ (far below the alarm threshold) More stable temperature control, no risk of overheating Energy consumption at low load (5kW) (kW·h / 30min) 10.8 6.3 Energy consumption reduced by 41.7% (compensating for the low-load advantage)

[0147] Comparison of proportions:

[0148] This comparative example uses the existing mainstream fixed flow control scheme and compares it with the dynamic PID flow control scheme of the present invention under the same test conditions. The results show that the present invention, through "target flow calculation + PID closed-loop regulation + pump group staged control", is significantly better than fixed flow control in terms of response speed, energy consumption, and temperature uniformity. Especially when the heat load fluctuates greatly (such as 5kW→30kW), it can quickly adjust the flow to match the heat dissipation demand and avoid the defects of "energy waste at low load and insufficient heat dissipation at high load" of the fixed flow scheme, highlighting the inventiveness and practicality of the present invention.

[0149] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0150] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A flow control method for a rack-level pump-driven two-phase flow cooling system, characterized in that, This includes a pump set with two pumps in parallel for use in rack-level pump-driven two-phase flow cooling systems, a shell-and-tube condenser, rack cold plate branches with 2-16 integrated and fitted simulated heat sources, condenser inlet temperature and pressure sensors, condenser outlet temperature and pressure sensors, flow meters, and electric regulating valves. in; The pump set with the two pumps in parallel is installed in the medium output pipeline of the condenser to provide circulation power for the two-phase flow working fluid, and the power of the pump set is dynamically adapted to the temperature of the cold plate. The inlet pipe of the shell-and-tube condenser is equipped with an inlet temperature and pressure sensor to collect inlet pressure data. and inlet temperature ; The outlet pipeline is equipped with an outlet temperature and pressure sensor to collect outlet pressure data. and outlet temperature ,and The pressure difference is used for pipeline identification; The flow meter is installed at the front end of the electric regulating valve to collect the real-time flow of the system, and the flow data and temperature and pressure data form a multi-dimensional feedback. Each cabinet cold plate branch is connected in parallel to the output pipeline of the pump group. Each branch is equipped with an independent temperature sensor to collect the surface temperature of the cold plate. The attached simulated heat source is attached to the heat absorption surface of the cold plate with thermal grease. The bonding gap is no more than 0.1mm and the thermal resistance is no more than 0.05K / W. The flow control method includes the following steps: S1. Multi-parameter acquisition: Data is acquired via condenser inlet / outlet temperature and pressure sensors. The surface temperature of the cold plate is collected by temperature sensors in each cold plate branch, and the highest temperature is selected. By collecting the real-time operating power of the simulated heat sources attached to each cold plate, the heat load of the corresponding cold plate branch can be obtained. to ; Real-time flow rate is collected by the flow meter system. ; S2. Dynamic target flow rate calculation: based on the average operating pressure of the cooling medium in the system. Average isobaric specific heat capacity Using the formula Calculate target flow ;in This is a dynamic thermal margin correction factor with an initial value of 1.2, which is adjusted in real time according to the phase change degree of the two-phase flow and the pipeline resistance. S3, Multi-dimensional Closed-Loop Flow Regulation: Based on " and "The deviation" is a proportional term, "rate of change" is the differential term, The deviation is taken as the integral term, and a PID closed-loop control strategy is used to output the adjustment amount of the electric regulating valve opening, so that the actual flow rate matches the target flow rate. ; S4, Pump Set-Flow Coordinated Control: Based on The pump power is adjusted in stages based on the comparison results with the preset threshold. S5. Blockage and Cavitation Early Warning and Self-Healing: Through... The pressure difference threshold is used to determine the risk of pipeline blockage. and "deviation rate" combined The risk of cavitation is determined by comparing the saturated vapor pressure of the cooling medium. After triggering an early warning, the valve opening and pump unit operating status are automatically adjusted to achieve self-repair.

2. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, In step S2, the dynamic thermal margin correction coefficient The adjustment logic is as follows: when hour, Adaptively adjusted to 1.0-1.1; when hour, The adaptive adjustment is set to 1.2-1.

3. When the system is in steady state, that is and hour, Keep it at 1.2; The coefficient "1.2" is determined based on experimental data of the latent heat of phase change of the cooling medium under the system working pressure and the friction loss along the pipeline, and is obtained by orthogonal experimental fitting as a phase change heat fluctuation compensation coefficient.

3. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, In step S3, the PID closed-loop control employs a "parameter adaptive optimization mechanism": The adjustment cycle is synchronized with the flow meter sampling cycle, and the value ranges from 0.5s to 2s. when When the preset heat exchange efficiency range of 2K-8K is exceeded, the PID proportional coefficient is temporarily increased to 1.2-1.5 times the original value, the derivative coefficient remains unchanged, and the integral coefficient is reduced to 0.8-0.9 times the original value. when When the flow rate returns to the preset range and the flow deviation is ≤±3%, the PID parameters will automatically recover to their initial values.

4. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, In step S4, the specific threshold and execution logic of the pump group power grading adjustment strategy are as follows: High-temperature emergency zone: when At that time, the pump set immediately switches to full power operation, and the two pumps start synchronously and output in coordination; Normal working range: when At that time, the pump set operating power and It exhibits linear positive correlation adjustment, and the adjustment formula is: ,in 70% of the rated power This is the power regulation coefficient, with a value ranging from 0.02 to 0.05 kW / K; Low temperature energy saving range: when When the pump set is switched to 40%-60% of its rated power, only one pump works while the other is on standby.

5. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, In step S5, the criteria for judging blockage and cavitation and the self-healing logic are as follows: Pipeline blockage risk: When If the blockage persists for more than 3 seconds, it is considered a risk of pipeline blockage. The system will automatically increase the opening of the electric regulating valve by 10%-20% while reducing the pump power by 10%-15%. If the blockage is resolved within 30 seconds after adjustment... If the device fails to return to the safe zone, an audible and visual alarm will be triggered. Pump unit cavitation risk: When and When the pressure is lower than the saturated vapor pressure of the cooling medium, it is considered a risk of cavitation. The system will automatically reduce the opening of the electric regulating valve by 5%-10% and start the standby pump to operate in coordination. If the flow deviation is not reduced within 10 seconds after adjustment, the pump set will automatically reduce the frequency to 50% of the rated power.

6. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, In step S2, the average isobaric specific heat capacity of the cooling medium The "dynamic lookup table method" is used to obtain the data based on the system's real-time average pressure. and the present The corresponding medium state is dynamically matched from the pre-stored thermodynamic parameter table of the cooling medium. value; The two-phase flow medium of the cooling system is R134a. The dynamic adjustment range is 1.2-1.8 kJ / (kg・K).

7. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, The adaptability design and traffic distribution logic of the cold plate branch of the cabinet are as follows: The number of branch circuits can be flexibly configured from 2 to 16, and it is compatible with 4U-42U server racks. Each cold plate branch circuit is used to cool 1-2 server nodes. Heat load of each branch to It can be directly acquired by collecting the real-time operating power signal of the simulated heat source, with a collection response time of ≤0.1s, without the need for additional configuration of a heat flux density sensor; When adding or removing cold plate branches, the control system automatically identifies the change in the number of branches and redistributes the target flow based on the real-time heat load ratio of each branch, avoiding the problem of multiple branches "competing for flow" or thermal imbalance.

8. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, In step S3, the PID closed-loop control adopts a three-parameter decoupling logic of "flow rate-temperature-pressure": The core control parameter is the "deviation between actual traffic and target traffic," with a weighting of 60%. by" The "rate of change of temperature difference" is used as an auxiliary adjustment parameter, with a weighting of 25%. by" "Pressure difference deviation" is used as an interference compensation parameter, with a weighting of 15%. The control system calculates the weighting coefficients of each parameter in real time and eliminates multi-parameter coupling interference through a decoupling algorithm, achieving a temperature control accuracy of ±0.4℃.

9. The flow control method for a rack-level pump-driven two-phase flow cooling system according to claim 1, characterized in that, The pump set with dual pumps in parallel is equipped with "master / standby coordinated switching logic": During steady-state operation, only the main pump operates, while the standby pump remains in standby mode, and the main pump power is determined according to... Dynamic adjustment; when When cavitation risk is triggered, the standby pump will start automatically, and the two pumps will work together, increasing the total power of the pump set to 1.8-2.0 times the rated power of a single pump; when Once the system returns to normal operating range and the cavitation risk is eliminated, the standby pump will shut down after a 10-second delay. When the main pump fails, i.e. the flow rate drops by ≥50% or the pump body temperature is ≥85℃, the standby pump will automatically switch to the main pump operation within 0.5s to ensure continuous cooling of the system.

Images