An electrothermal coupling liquid cooling response evaluation method, system, device and storage medium

Abstract

This invention discloses an electrothermal coupling liquid cooling response assessment method, system, device, and storage medium, relating to the fields of information technology and green energy management. It includes establishing an electrothermal coupling real-time operation model reflecting the correlation between power and temperature variables in server power and liquid cooling heat exchange, based on the power consumption and heat generation behavior within a liquid-cooled data center; based on the electrothermal coupling real-time operation model, establishing a liquid-cooled data center with chilled water circulation, cooling water circulation, and server room air circulation, constructing a multi-dimensional response index system, and generating a response collaborative control strategy under service quality constraints and thermal safety threshold constraints; under the multi-dimensional response index system, performing sensitivity analysis on power change behavior based on the response collaborative control strategy, constructing a multi-dimensional index set, and quantitatively calculating the demand response potential; the method of this invention improves the accuracy of demand response in liquid-cooled data centers through precise modeling and multi-dimensional assessment.

Description

Technical Field

[0001] This invention relates to the fields of information technology and green energy management technology, specifically to an electrothermal coupled liquid cooling response evaluation method, system, device, and storage medium. Background Technology

[0002] With the continuous development of information technologies such as artificial intelligence, big data, and cloud computing, data centers are expanding in scale, with significantly increased computing density and operating load levels. Data centers are gradually becoming high-energy-consuming and high-load centralized power-consuming units. During long-term continuous operation, data centers generate stable and concentrated load demands on the power system, which can easily exacerbate the pressure on the power grid during peak hours. Against the backdrop of energy structure transformation and increasing power system regulation needs, exploring the load regulation potential of data centers during operation and enabling them to participate in demand response has become an important research direction in the current intersection of information technology and energy management.

[0003] In the overall energy consumption of data centers, server power consumption and cooling system power consumption account for the majority. As the power density of server racks continues to increase, traditional air cooling methods are gradually becoming insufficient to meet heat dissipation demands. Liquid cooling technology, due to its higher heat exchange efficiency and more stable temperature control capabilities, is widely used in high-density data centers. The high specific heat characteristics of liquid cooling media create significant thermal inertia during data center operation, providing a physical basis for the coordinated operation between power regulation and heat release. Most existing research on demand response in data centers still focuses on air-cooled data centers, with the analysis emphasizing computing load regulation and task scheduling, while relatively insufficient consideration is given to the operating characteristics of liquid-cooled data centers.

[0004] Existing methods for assessing demand response potential often employ simplified power models or static energy consumption indicators, which fail to adequately describe the relationship between server power changes and the heat exchange process of the cooling system. In liquid-cooled data centers, server power changes directly affect heat generation levels, and these heat generation changes are transmitted through the liquid cooling loop and act on chilled water, cooling water, and the air environment of the computer room. There is a clear electrothermal coupling relationship between these operational links. The lack of characterization of these coupling relationships can easily lead to discrepancies between the demand response capability assessment results and the actual operating conditions.

[0005] Meanwhile, the demand response process is affected by both service quality constraints and temperature safety constraints. Some studies only introduce service quality constraints on the computing side without uniformly modeling the temperature evolution process. Although some studies consider the operating boundary of the cooling system, they do not conduct correlation analysis between power change behavior and thermal safety conditions. This fragmented approach is difficult to accurately reflect the real adjustment space of data centers under multiple constraints. Existing evaluation methods mostly use single or a few indicators to describe demand response capabilities, making it difficult to reflect the impact of power change behavior on the overall operating status from multiple dimensions. Summary of the Invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by this invention is that existing liquid-cooled data center demand response assessment methods have problems such as insufficient consideration of electrothermal coupling, single assessment dimensions, and ineffective integration of service quality and thermal safety constraints. As a result, the assessment results cannot truly reflect the demand response capability, flexibility, and safety of liquid-cooled data centers in actual operation. Therefore, how to establish a liquid-cooled data center demand response assessment framework based on electrothermal coupling and realize multi-dimensional quantitative calculation of potential has become an urgent problem to be solved.

[0008] To address the aforementioned technical problems, this invention provides the following technical solution: an electrothermal coupling liquid cooling response evaluation method, comprising: establishing an electrothermal coupling real-time operation model reflecting the correlation between power and temperature variables between server power and liquid cooling heat exchange based on the power consumption and heat generation behavior within a liquid-cooled data center; based on the electrothermal coupling real-time operation model, establishing a liquid-cooled data center involving chilled water circulation, cooling water circulation, and server room air circulation, constructing a multi-dimensional response index system, and generating a response collaborative control strategy under service quality constraints and thermal safety threshold constraints; under the multi-dimensional response index system, performing sensitivity analysis on power change behavior based on the response collaborative control strategy, constructing a multi-dimensional index set, and quantitatively calculating the demand response potential; the response collaborative control strategy includes a collaborative response mode where server and liquid-cooled data center power decrease simultaneously; and a response mode where server power remains constant while liquid-cooled data center power decreases.

[0009] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation method of the present invention, the electrothermal coupling real-time operation model includes: obtaining the corresponding heat generation based on the real-time power of the server; describing the mass flow rate of the cooling medium and the inlet and outlet temperature difference in the liquid cooling loop through the law of conservation of energy; and obtaining the current heat exchange. In the electrothermal coupling real-time operation model, a first-order inertial element is introduced to describe the electromechanical characteristics and time lag relationship between the server power change and the liquid cooling heat exchange response.

[0010] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation method described in this invention, the liquid-cooled data center includes a natural water cooling design using a cooling tower as the core economizer. This design is primarily based on the WSE configuration of the cooling tower and determines the parameter settings for the chiller, water pumps (cooling water pumps and condensate pumps), cooling tower, and plate heat exchanger. Based on three operating modes, the power utilization efficiency is optimized using an outdoor natural cold source. The three operating modes include full natural cooling, partial natural cooling, and full main unit cooling. Full natural cooling includes cooling from the cooling tower to handle all loads. Partial natural cooling includes using the cooling tower to pre-cool the return water. Full main unit cooling includes cooling from a traditional chiller.

[0011] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation method of the present invention, the service quality constraints and thermal safety threshold constraints include: the service quality constraints include limiting the server power adjustment range; the thermal safety threshold constraints include limiting the server outlet temperature to not exceed a preset upper temperature limit.

[0012] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation method described in this invention, the multi-dimensional response index system includes temperature, power, time, and comprehensive evaluation dimensions. The temperature dimension includes monitoring equipment safety using the server's maximum temperature and temperature rise indicators. The power dimension includes describing load shedding characteristics using adjustable power, rebound power, and rebound rate. The time dimension includes defining adjustment time to evaluate response speed and defining response time to evaluate the time to recover to normal operation. The comprehensive evaluation dimension includes introducing total response capacity, average response capacity, demand response quality indicators, and a comprehensive demand response index. The comprehensive demand response index includes balancing response capability, duration, and the negative impact of power rebound to generate a priority reference for power dispatch. A larger total responsive capacity and duration indicate a stronger response capability and higher priority, while a higher rebound rate poses a greater threat to the power grid, indicating a greater harm to the response and lower priority. The comprehensive demand response index is constructed and expressed as: , in, This represents the comprehensive demand response index. Using Dynamic Voltage / Frequency Scaling (DVFS) technology, the energy consumption of the server cluster is adjusted. The server's power consumption can be expressed as a function expression related only to the CPU frequency. Taylor expansion and Laplace transform of this function expression generate a complex frequency domain operating point power regulation control expression, expressed as: , in, This represents the change in server power in the complex frequency domain. The slope representing the local linearization of the server's power consumption at a certain operating point. This represents the frequency adjustment amount in the complex frequency domain of the server. Given that the compressor's dynamic response is limited by its electromechanical characteristics and exhibits a significant inertial time lag compared to the millisecond-level adjustment of the server CPU, a first-order inertial element is introduced into the power control to construct a complex frequency domain dynamic response model, expressed as: , in, This represents the change in power of the liquid cooling system. This indicates the change in the frequency of the internal compressor. This represents the proportionality coefficient of the first-order inertial element in the cooling system. This represents the time constant of the first-order inertial element of the cooling system.

[0013] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation method described in this invention, the response collaborative control strategy includes: establishing a functional relationship between the server's average response time and power using a queuing theory model; allocating response tasks to the server and the liquid-cooled data center using a server power reduction coefficient; regulating power on the server side using dynamic voltage and frequency adjustment technology; coordinating with the liquid-cooled data center by adjusting the compressor's operating frequency; and real-time judgment of the relationship between the server's outlet temperature and a safety threshold. Once the limit is reached or the safety threshold is exceeded, a feedback mechanism will execute forced cooling. By introducing a server power reduction coefficient, the remaining power adjustment (1-α) is undertaken by the liquid cooling system. This process can be expressed as: , , in, This represents the response power allocated to the server. This represents the response power allocated to the liquid cooling system. Indicates a demand response signal. This represents the server power reduction factor, the maximum value of which is determined by both the demand response signal and the power constraint, ranging from 0 to this maximum. Within the specified range, DC system operators can select appropriate values ​​based on operational needs.

[0014] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation method of the present invention, the multi-dimensional index set includes: recording the response power curves under different server power reduction coefficients or different temperature safety thresholds within the time interval corresponding to the start and end times of power change; performing integral calculation on the response power curves; and obtaining the total response capacity, average response capacity, and rebound rate by adjusting the server power reduction coefficient and the server maximum temperature safety limit parameter, thereby generating a sensitivity spectrum.

[0015] Another objective of this invention is to provide an electrothermal coupling liquid cooling response evaluation system, which quantitatively calculates demand response potential based on an electrothermal coupling real-time operation model and a multi-dimensional response index system, thus solving the problem that current demand response evaluation methods cannot fully consider electrothermal coupling relationships, power change paths, and thermal safety constraints.

[0016] As a preferred embodiment of the electrothermal coupling liquid cooling response evaluation system of the present invention, it includes: an electrothermal coupling modeling module, a response coordination control module, and a demand response evaluation module; the electrothermal coupling modeling module is used to establish a dynamic coupling relationship between server power and liquid cooling heat exchange process based on an electrothermal coupling real-time operation model reflecting the correlation between power and temperature variables in liquid cooling data centers, and obtains the current heat exchange and power changes through the law of conservation of energy; the response coordination control module is used to generate three cooling modes, including full natural cooling, partial natural cooling, and full host cooling, based on the electrothermal coupling real-time operation model, and establish a multi-dimensional response index system using service quality constraints and thermal safety threshold constraints, generating a coordinated response mode between server power and liquid cooling data centers; the demand response evaluation module is used to evaluate power change behavior through queuing theory models and sensitivity analysis, and quantitatively calculate response depth, response duration, and power rebound characteristics based on the server power reduction coefficient and the response capability of the liquid cooling system, and accurately evaluate the demand response potential of the liquid cooling data center.

[0017] Another object of the present invention is to provide an electrothermal coupling liquid cooling response evaluation device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the electrothermal coupling liquid cooling response evaluation method.

[0018] Another object of the present invention is to provide an electrothermal coupling liquid cooling response evaluation storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the electrothermal coupling liquid cooling response evaluation method.

[0019] The beneficial effects of this invention are as follows: The electrothermal coupling liquid cooling response evaluation method provided by this invention establishes a real-time operating model of electrothermal coupling between server power and liquid cooling heat exchange, avoiding the simplistic handling of cooling system and power changes in existing methods; by designing multiple response modes, it enables data centers to flexibly adapt to the requirements of power grid demand response under different operating conditions; it evaluates the demand response potential through multiple dimensions such as temperature, power, and time; and it achieves better results in terms of accurate modeling, flexible response, and comprehensive evaluation, making the demand response evaluation of liquid-cooled data centers not only more accurate but also adaptable to complex working environments and safety requirements. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 The above is an overall flowchart of an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0022] Figure 2 This is a schematic diagram of the natural water cooling principle of an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0023] Figure 3 This is a diagram of the real-time operation model of an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0024] Figure 4 This is a quantitative diagram of the external characteristics of an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0025] Figure 5 This is a demand response collaborative control diagram for an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0026] Figure 6 Temperature and power timing diagrams for a mode 1 parameter sensitivity analysis scenario of 850kW response power for an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0027] Figure 7 Temperature and power timing diagrams for a mode 1 parameter sensitivity analysis scenario of 880kW response power for an electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention.

[0028] Figure 8 The temperature and power timing diagram of the mode 2 parameter sensitivity analysis scenario of the electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention is shown in the maximum temperature limit of 30°C for the server.

[0029] Figure 9 The temperature and power timing diagram of the mode 2 parameter sensitivity analysis scenario of the electrothermal coupling liquid cooling response evaluation method provided in Embodiment 1 of the present invention is shown in the maximum temperature limit of 32°C for the server. Detailed Implementation

[0030] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0031] Example 1, referring to Figure 1-5 As an embodiment of the present invention, an electrothermal coupling liquid cooling response evaluation method is provided, comprising:

[0032] S1: Based on the power consumption and heat generation behavior in liquid-cooled data centers, establish a real-time electrothermal coupling operation model that reflects the relationship between power and temperature variables between server power and liquid cooling heat exchange.

[0033] It should be noted that the electrothermal coupling real-time operation model includes obtaining the corresponding heat generation based on the server's real-time power, describing the mass flow rate of the cooling medium and the inlet and outlet temperature difference in the liquid cooling loop through the law of conservation of energy, and obtaining the current heat exchange. In the electrothermal coupling real-time operation model, a first-order inertial element is introduced to describe the electromechanical characteristics and time lag relationship between the server power change and the liquid cooling heat exchange response.

[0034] From the perspective of energy producers and consumers, servers are considered as the core heat source. The heat they generate is transferred through liquid-cooled data centers and computer room environments. Based on an electrothermal coupling real-time operation model, the energy consumption characteristics of information technology equipment and supporting cooling infrastructure are characterized. The focus is on the chiller compressor, which accounts for the highest proportion of energy consumption in the cooling system. Based on the abstract modeling method of the law of conservation of energy, the refrigeration efficiency ratio and heat exchange process equations are used to accurately describe the power consumption of the chiller and the heat exchange capacity of the evaporator and condenser. The electromechanical characteristics and time delay of the refrigeration system during power regulation are characterized by a first-order inertial element.

[0035] The chiller unit compressor serves as the refrigeration cycle, consuming electricity to transfer heat from the evaporator to the condenser via refrigerant. Based on the law of conservation of energy, as referenced... Figure 2 As shown, an abstract model of the chiller unit is constructed using a reasonable COP (coefficient of performance) and heat exchange process, namely, the abstract model X1 of the chiller unit: , in, This indicates the energy consumption of the chiller unit. This indicates the total heat transferred by the evaporator. This indicates the total heat transferred by the condenser.

[0036] The energy consumption of the chiller unit is expressed as follows: , The total heat transferred by the evaporator is expressed as: , The total heat transferred by the condenser is expressed as: , in, This indicates the power output of the chiller unit under normal operating conditions. Indicates the time difference in operation of the chiller unit. This indicates the heat transfer power of the evaporator. This indicates the heat transfer power of the condenser.

[0037] The heat transfer power of the evaporator is expressed as: , , The heat transfer power of the condenser is expressed as: , in, This indicates the energy efficiency ratio of the chiller unit under operating conditions. This indicates the rated energy efficiency ratio of the chiller unit. This represents the ratio of the chiller unit's energy efficiency ratio during normal operation to its rated energy efficiency ratio. This indicates the rated power of the chiller unit. This represents the physical signal of the controller.

[0038] It should also be noted that by setting a reasonable electrothermal efficiency ratio and using the "Controlled Heat Flow Source" module in Simulink to maintain a constant heat flow rate, a computer room model describing the heat exchange subsystem between the internal equipment and the external space of the data center is constructed. The heat loss generated by the heater and the temperature change over time are obtained and expressed as follows: , , in, This indicates the specific heat capacity of air under constant pressure. Indicates the air quality inside the house. This indicates the current indoor air temperature. This indicates the current temperature of the hot air entering the room. This indicates the temperature of the hot air entering the house. This represents the equivalent thermal resistance of the building. This represents the heat loss rate, which is the temperature change caused by heat loss per unit time. This represents the rate of change of heat loss, that is, the change in heat loss within a room per unit time. This indicates the rate of change of room temperature, that is, the rate at which the temperature changes per unit of time.

[0039] In each simulation step, the corresponding heat generation is first calculated based on the real-time power of the server and input into the plate heat exchanger R1. The heat exchanger calculates the server outlet temperature by combining the inlet temperature and flow rate of the chilled water loop. Then, the server outlet temperature is used as the load input of the chiller unit LA to determine the power change of the chiller unit. The chiller unit outputs the updated chilled water supply temperature through a first-order inertial link and feeds it back to the inlet of the plate heat exchanger. This realizes the continuous coupling mapping of server power change, heat generation, chilled water loop, condensate loop and computer room air loop, forming a complete real-time operation model of electrothermal coupling of liquid-cooled data center.

[0040] S2: Based on the electrothermal coupling real-time operation model, establish a liquid-cooled data center between chilled water circulation, cooling water circulation and computer room air circulation, construct a multi-dimensional response index system, and generate a response collaborative control strategy under service quality constraints and thermal safety threshold constraints.

[0041] It should be noted that, as referred to Figure 2 As shown, the liquid-cooled data center includes a natural water-cooled design with a cooling tower (TA) as the core economizer. This design is primarily based on the WSE configuration of the cooling tower TA and determines the parameter settings for the chiller unit (LA), water pumps (cooling water pump L1, condensate pump L2), cooling tower TA, and plate heat exchanger (R1). This reflects the actual power consumption ratio of each component in the liquid-cooled system in real time. Based on three operating modes, it optimizes energy utilization efficiency using an outdoor natural cooling source. The three operating modes include full natural cooling, partial natural cooling, and full host cooling.

[0042] For reference Figure 3 As shown, the liquid-cooled data center consists of two independent chilled water and condensate liquid loops and one server room air network loop. The chilled water is an ethylene glycol aqueous solution, and the condensate is ordinary water. During operation, the chilled water loop absorbs heat from the servers and transfers it to the condensate loop. During the transfer, the server room air loop simulates the heat exchange between the internal equipment of the liquid-cooled data center and the outside environment through the air. The condensate loop uses a cooling tower (TA) to discharge heat into the environment. The heat transfer between the chilled water loop and the condensate water loop occurs directly through the liquid-liquid heat exchanger and indirectly through the chiller unit (LA). When the cooling capacity of the heat exchanger is insufficient, it is supplemented by the chiller unit (LA).

[0043] The server temperature is the measured temperature at the contact surface between the liquid cooling system and the server, not the CPU's internal junction temperature. When the liquid-cooled data center receives the start signal (SS) and triggers a demand response, the server temperature rises as the operating status adjusts, reaching its peak when the response end signal (CS) is received. This is represented as: , in, This indicates the maximum temperature during the entire server response process. This indicates the temperature when the server is running normally. This represents the temperature increment during the response process.

[0044] During demand response in a liquid-cooled data center, power is reduced and maintained for a certain period. After the response ends, the power overshoot phenomenon, i.e., power bounce, occurs during the recovery phase due to the controller accelerating the system's return to steady-state logic. This is represented as: , , in, This indicates the power of the liquid-cooled data center during response. This indicates the power consumption of a liquid-cooled data center during normal operation. This represents the power adjustment amount during the response process. Indicates the rebound power. This represents the rebound rate, which is a number greater than 1.

[0045] The service quality constraints and thermal safety threshold constraints include limiting the server power adjustment range; the thermal safety threshold constraints include limiting the server outlet temperature to not exceed the preset upper temperature limit.

[0046] The multi-dimensional response indicator system includes temperature, power, time, and comprehensive evaluation dimensions. The temperature dimension monitors equipment safety using server maximum temperature and temperature rise indicators. The power dimension describes load shedding characteristics using adjustable power, rebound power, and rebound rate. The time dimension defines adjustment time to assess response speed and response time to assess the time to restore normal operation. The comprehensive evaluation dimension introduces total response capacity, average response capacity, demand response quality indicators, and a comprehensive demand response index. The comprehensive demand response index balances response capability, duration, and the negative impact of power rebound to generate a priority reference for power dispatch.

[0047] The temperature rise index assesses equipment safety risks by monitoring the maximum temperature rise during the response process. When the temperature rise threshold is set and exceeds the threshold, the thermal safety risk level is determined to be upgraded, and the feedback mechanism is triggered.

[0048] The response coordination control strategy includes: using a queuing theory model to establish a functional relationship between the average server response time and power; allocating response tasks to the server and liquid-cooled data center through a server power reduction coefficient; using dynamic voltage and frequency adjustment technology for power regulation on the server side; coordinating with the liquid-cooled data center by adjusting the compressor operating frequency; and judging the relationship between the server outlet temperature and the safety threshold in real time. Once the limit is reached or the safety threshold is exceeded, the feedback mechanism will execute forced cooling.

[0049] It should also be noted that the response time is defined as the total time from receiving the response signal to the system fully recovering to normal operating status, expressed as: , , in, Indicates the adjustment time. This indicates the time point at which the power reaches the target value. Indicates the time point at which the response signal was received. Indicates response time. This indicates the time point at which the response end signal was received.

[0050] Given that the power grid dispatch level does not usually directly perceive the operational constraints within demand response resources, for liquid-cooled data centers, the core operational constraint is server temperature. The data center controller must integrate a closed-loop control mechanism based on temperature feedback to automatically trigger the cooling system's adjustment intervention when the server temperature reaches the upper limit of the safety threshold, and introduce corresponding quantitative indicators for evaluation.

[0051] First, the total integral of the power decrease during the response process is defined as the total response capacity, expressed as: , in, Indicates the total response capacity, as shown in the reference. Figure 4-5 As shown, the system performs relevant control processes at each step until the target is achieved or the server power margin and temperature limit are exceeded, ultimately outputting the total response capacity throughout the entire response process. express and The difference in operating power at any given time.

[0052] Such as runtime external characteristic quantization Figure 4As shown, (a) is a quantitative diagram of the external temperature characteristics of a liquid-cooled data center server, which is used to show that after the liquid-cooled data center receives the start signal (SS), the server temperature gradually rises from the normal operating temperature, and reaches the maximum temperature of the server during the entire response process when the close signal (CS) is received. The temperature increment during the response process is denoted as ΔT; (b) is a quantitative diagram of the external power and time characteristics of a liquid-cooled data center, which is used to show the power reduction process during the demand response, the power rebound process after the response ends, and the definitions of adjustment time AT and response time ResT.

[0053] based on The average response capacity is defined as the sum of the total power decrease during the response process and... The ratio is expressed as: , in, To represent the average response capacity, two comprehensive indicators are introduced to evaluate the performance of liquid-cooled data centers in the demand response process. and The quality of the response is evaluated by the ratio between the two values, ranging from 0 to 1, with values ​​closer to 1 representing higher quality. The smaller the value, and the less the server temperature is limited during the response process, or the more limited it is in the short term.

[0054] A larger total responsive capacity and duration indicate a stronger response capability and higher priority. Conversely, a higher rebound rate poses a greater threat to the power grid, indicating a more severe response and lower priority. A comprehensive demand response index is constructed and expressed as follows: , in, This represents the overall demand response index.

[0055] Dynamic Voltage / Frequency Scaling (DVFS) technology is used to regulate the energy consumption of server clusters. The power consumption of the servers can be expressed as a function that is only related to the CPU frequency. Taylor expansion and Laplace transform of this function generate a complex frequency domain operating point power regulation control expression, which is expressed as: , in, This represents the change in server power in the complex frequency domain. The slope representing the local linearization of the server's power consumption at a certain operating point. This indicates the frequency adjustment amount of the multi-frequency domain server.

[0056] Given that the dynamic response of a compressor is limited by its electromechanical characteristics and exhibits a significant inertial time lag compared to the millisecond-level adjustment of a server CPU, a first-order inertial element is introduced into the power regulation to construct a complex frequency domain dynamic response model, expressed as: , in, This represents the change in power of the liquid cooling system. This indicates the change in the frequency of the internal compressor. This represents the proportionality coefficient of the first-order inertial element in the cooling system. This represents the time constant of the first-order inertial element of the cooling system.

[0057] Based on the above control method, a collaborative control strategy between the server and the liquid-cooled data center is established. The response signal receiving and power distribution module receives the demand response signal from the power grid, determines the server's response power constraint while considering Quality of Service (QoS), and then allocates the target response power to the server and the liquid cooling system. The judgment manager verifies the liquid cooling system's demand response power by comparing the lower limit power with the upper limit temperature. The processed signal is transmitted to the liquid-cooled data center through the controller in the signal control manager. The specific process is as follows: (i) During the response signal reception and power distribution phase, the liquid-cooled data center simultaneously receives demand response signals and server power constraint signals. Based on QoS constraints, it ensures that the task processing response time is below the maximum delay limit. From the perspective of the power system, a correlation between QoS and regulating power is established during the demand response process. The M / M / 1 queuing model is used to model the average server response time as a function of server power, expressed as: , in, This indicates the average server response time. The server's service rate can be viewed as a measure of its performance. A monotonically increasing function. This indicates the server's task arrival rate. This indicates the server's latency boundary.

[0058] In time At this point, after expanding the above equation into a Taylor series, the incremental relationship can be expressed as follows: , in, express The minute increment, express The minute increment, Indicates that the server is Power at any moment Indicates the server at time Service speed, Indicates the server at time The server task arrival rate, express In time The derivative at point .

[0059] Service quality can be controlled by setting the adjusted power in load regulation, using a relationship proportional to the rated power as follows: , in, This represents the server power adjustment amount considering QoS. This represents the server demand response constraint factor considering QoS. This indicates the server's rated power.

[0060] The server's power constraint signal is represented as: , in, This indicates the server's power constraint signal. This indicates the server's real-time power.

[0061] By introducing a server power reduction factor, the remaining power regulation of 1-α is handled by the liquid cooling system, and this process can be expressed as: , , in, This represents the response power allocated to the server. This represents the response power allocated to the liquid cooling system. Indicates a demand response signal. This represents the server power reduction factor, the maximum value of which is determined by both the demand response signal and the power constraint, ranging from 0 to this maximum. Within the specified range, DC system operators can select appropriate values ​​based on operational needs.

[0062] The DC system includes the power distribution network and control loop that provides DC power to server clusters, chillers, and auxiliary equipment. The DC system operator refers to the personnel who monitor and regulate the power distribution network of the liquid-cooled data center. They are responsible for allocating and adjusting the output power of the DC bus or DC power supply link according to the demand response signal and the server power reduction factor to ensure that the server and liquid cooling system perform power regulation tasks under constraints.

[0063] (ii) In the execution judgment manager, since the front-end module has fully considered the QoS and remaining power constraints of the server, the response power of all servers is equal to the allocated power. For the liquid cooling system, the response power allocated to the server will be compared with the real-time power value of the liquid cooling system. The output will be processed by a logical AND operation with another judgment module. The latter compares the real-time temperature of the server with the set upper limit value, and the initial output is 1. This mechanism aims to ensure that the server always operates within the allowable temperature range, which is expressed as: , , in, This indicates the server's actual response capability in the demand response. This indicates the actual response power of the liquid cooling system in demand response. This indicates the real-time power value of the liquid cooling system. This indicates the server's real-time temperature. This indicates the upper limit for the server temperature setting.

[0064] (III) The signal control manager will transmit the power allocation signals of the server and the liquid cooling system to the DVFS controller and the PI controller respectively. The server's task power signal is superimposed on its idle baseline. The two controlled power values ​​can be expressed as: , , in, This indicates the server's power at a previous point in time. This indicates the power of the liquid cooling system at a previous time point. This indicates the server's baseline power.

[0065] (iv) After the control signal is sent to the server and the internal compressor of the liquid cooling system, the server outlet temperature, operating power and liquid cooling system operating power will be fed back to the front-end module in real time for judgment.

[0066] S3: Under the multi-dimensional response index system, sensitivity analysis of power change behavior is carried out based on the response collaborative control strategy, a multi-dimensional index set is constructed, and the demand response potential is quantitatively calculated.

[0067] Within the time interval corresponding to the start and end times of power change, record the response power curves under different server power reduction coefficients or different temperature safety thresholds.

[0068] By integrating the response power curve and adjusting the server power reduction factor and the server's maximum temperature safety limit parameter, the total response capacity, average response capacity, and rebound rate are obtained, and a sensitivity spectrum is generated.

[0069] It should be noted that, firstly, within the established multi-dimensional indicator system framework, diverse power system demand response scenarios are set up for the liquid-cooled data center. A demand response collaborative control method coupled with the server cluster and the liquid cooling system is applied. The system synchronizes the received power grid demand response signals to the response signal receiving and power distribution module. Combined with the server's real-time power constraint signals, the dynamic load distribution logic is executed among the subsystems through dynamic voltage / frequency adjustment and variable frequency control of the chiller unit's LA compressor. Subsequently, the sensitivity analysis and feedback optimization stage of key control parameters is entered. This process involves systematically changing the core adjustment coefficients and constraint limits in the algorithm to quantitatively analyze the impact mechanism of each variable on the overall system response characteristics.

[0070] It should also be noted that the server power reduction factor was adjusted to explore the power distribution weight between computing tasks and the cooling system; and the maximum temperature limit setting range of the server was changed in stages. During the parameter change process, the system's regulation efficiency, duration and negative effects such as power rebound were quantitatively evaluated and mapped using the designed response quality index and demand response comprehensive index.

[0071] Finally, the evolution of the impact of each extracted parameter on response potential, security, and stability is used as closed-loop feedback information and fed back to the controller parameter preset library of the actual system. This provides a standardized parameter configuration guide and operational decision reference for liquid-cooled data centers to participate in regional power system demand response in different regions and scales.

[0072] Example 2, as referred to Figure 6-9 As one embodiment of the present invention, an electrothermal coupling liquid cooling response evaluation method is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0073] First, the typical response modes of liquid-cooled data centers were analyzed, and sensitivity analysis of key parameters was performed under these modes.

[0074] For a single liquid-cooled data center at a certain operating moment, its computing load has three operating modes: increase, decrease, and remain unchanged. The power of the liquid cooling system at a certain operating moment is also divided into the above three states according to the system cooling demand of different computing load changes. Thus, the two work together to have a distribution of 9 states as shown in Table 1.

[0075] Table 1 Response Status of Liquid-Cooled Data Center

[0076] Free state: 1; Unreasonable state: 2; Locked state: 3; Reasonable demand response state: 4.

[0077] We consider the state where both server and liquid cooling system power are increased to be a free state, meaning that the increased server power leads to a rise in temperature, and the liquid cooling system responds by increasing its power to increase cooling capacity. When both are in a hold-up state, this data center is in a locked state.

[0078] In addition, there are four states in which the power adjustment of the liquid cooling system follows the power change of the server inversely, so we define these states as unreasonable states.

[0079] Finally, the remaining three states are defined as reasonable demand response states, each corresponding to a real-world scenario where a liquid-cooled data center participates in demand response: In Mode 1, both server and liquid cooling system power are reduced, making it the preferred mode for current load distribution calculation response capacity and unrestricted data centers. Mode 2 keeps server power constant and uses thermal inertia to reduce liquid cooling system power, simulating a situation where critical server loads cannot be moved. Mode 3 simulates a load transfer state where server power increases while liquid cooling system power remains constant, but since demand response mainly refers to load reduction, this mode will not be analyzed in detail later.

[0080] Subsequent assessments of the potential of liquid-cooled data centers to participate in demand response will be based on Mode 1 and Mode 2.

[0081] Table 2 Model Parameter Settings

[0082] Based on the established electrothermal coupling real-time operation model and the constructed demand response potential assessment index system, the response capabilities under the two modes are analyzed quantitatively and qualitatively.

[0083] Mode 1: Both the server and the liquid cooling system respond to demand, resulting in a simultaneous decrease in power.

[0084] In this mode, when the response time reaches 10 minutes, the server outlet temperature still does not exceed 30 degrees Celsius, and the temperature change shows a small-proportion, low-power upward trend. However, there is a break in the rebound power. Before this break, the power rebound power at the action space point is small and changes slowly. After this break, it shows a surge trend. Therefore, considering the impact on the power grid after the response and under the premise of safety, the temperature rise of the server cannot be used as the only constraint on the response of the liquid-cooled data center. However, the strong inertial trend of the temperature has already strongly demonstrated the rationality of this response mode. In the future, optimizing the power rebound suppression technology will greatly improve its application possibilities.

[0085] Mode 2: Server power is maintained while the liquid cooling system responds to demand and power is reduced.

[0086] In this mode, the liquid cooling power is reduced by the same amount. When the system reaches thermal equilibrium, It has little impact on the server's temperature rise, but the continuous heat buildup in the cooling loop can lead to... Follow The increase in size also shows a surge trend, although this pattern exists within the same... The temperature rise will be slightly higher than in mode 1, but this is evident if set up properly. and ,That and Still acceptable, so Mode 2 provides an alternative contract model for the computer collaboration market that does not involve server load transfer, as it adapts to situations where server computing load cannot participate in demand response due to protection or other reasons.

[0087] Further evaluation of the responsiveness of liquid-cooled data centers under these two modes and their impact on... , Temperature limit (TL) Sensitivity analysis was performed on relevant variables.

[0088] In Mode 1, set TL to 30°C, adjust the power to 850kW and 880kW respectively, and adjust the distribution coefficient. See the reference for the temperature and power timing diagrams of each device in the liquid-cooled data center. Figure 6-7 As shown in the graph, when the server temperature reaches its upper limit, the compressor in the liquid cooling system will operate in "on-off" mode, maintaining the temperature at TL with a fluctuation range. As can be seen from 0.1 and 0.15 respectively, When the value is 0.1, the liquid cooling system reduces power more, so the temperature reaches the upper limit earlier. This results in more "bulges" in the power reduction, which will affect the responsiveness of the liquid-cooled data center.

[0089] The values ​​of each evaluation indicator are shown in Table 3. From the table, we can see that... 0.15 It is 0.1. When it is 850kW, Both demand response quality and demand response quality increased by 7.68%. Increased by 0.6%, When it is 880kW, Both demand response quality and demand response quality increased by 8.09%. It increased by 0.57%, which shows that although A smaller value is more beneficial to the quality of service for computing data, but it has a negative impact on the overall responsiveness of the liquid-cooled data center.

[0090] Furthermore, when the temperature reaches its upper limit during the response process, the same Based on the value The larger the size, the larger the proportion of the "convex domain," and the smaller the quality of the demand response. Compared to 850kW For 880kW, the demand response quality is... The values ​​were 2.88% and 2.49% higher at 0.1 and 0.15, respectively.

[0091] exist When it is 0.2, both Under these circumstances, the server temperature did not exceed TL, so the demand response quality and They are close to and all close to 1, but due to the delay in adjusting to the target power, i.e., AT is not equal to 0, this makes... It will be lower than the target setting value, which reduces the quality of demand response. In other words, the smaller the response delay, the greater the quality of demand response and the better the response effect. Ideally, the demand response quality is 1 under no delay conditions.

[0092] Table 3 Response performance indicators under different allocation coefficients and regulation power

[0093] Table 4 Response capability indicators under different temperature thresholds and response times

[0094] In Mode 2, the liquid cooling system is completely shut down, TL is set to 30°C and 32°C respectively, and the response time is adjusted. , as reference Figure 8-9 As can be seen from the figure, the temperature rise rate of mode 2 is significantly higher than that of mode 1, and the proportion of the "protruding domain" is also higher.

[0095] The values ​​of each evaluation index are shown in Table 4. From the table, it can be seen that under the same TL, since the compressor remains in "on-off" mode for the extended period of time, therefore... The larger, The smaller the demand response quality, the lower the TL (total temperature) at 30°C. Compared to 600s, 1200s and 1800s Both demand response quality and demand response quality decreased by 4.91% and 5.95% respectively, at a TL of 32°C. Compared to 600s, 1200s and 1800s Both demand response quality and demand response quality decreased by 1.53% and 2.44%, respectively, but at this time... However, the increase is due to the rebound power not being as large as expected. The increase and the small area of ​​a single peak in the "protruding region" make it possible to... Among the constituent variables The determination of [the specific factor] carries greater weight. In different response scenarios, some prioritize the average quality of the response, while others place more emphasis on the duration of the response. The aforementioned requirements for response quality and [the specific factor] are related to [the specific factor]. The different trends of change in specific scenarios demonstrate that the proposed response capability assessment system provides more comprehensive support for the scheduling of liquid-cooled data centers.

[0096] Example 3, an embodiment of the present invention, provides an electrothermal coupling liquid cooling response evaluation system, including an electrothermal coupling modeling module, a response coordination control module, and a demand response evaluation module.

[0097] The electrothermal coupling modeling module is used to establish a real-time operating model of electrothermal coupling based on the relationship between server power and liquid cooling heat exchange in a liquid-cooled data center, reflecting the correlation between power and temperature variables. It establishes the dynamic coupling relationship between server power and liquid cooling heat exchange process and obtains the current heat exchange and power changes through the law of conservation of energy.

[0098] The response coordination control module is used to generate three cooling modes based on the electrothermal coupling real-time operation model, including full natural cooling, partial natural cooling, and full host cooling. It adopts service quality constraints and thermal safety threshold constraints to establish a multi-dimensional response index system and generate a coordinated response mode between server power and liquid-cooled data center.

[0099] The demand response assessment module is used to evaluate power change behavior through queuing theory models and sensitivity analysis. Based on the server power reduction factor and the response capability of the liquid cooling system, it quantitatively calculates the response depth, response duration and power rebound characteristics, and accurately assesses the demand response potential of the liquid-cooled data center.

[0100] This embodiment also provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements an electrothermal coupling liquid cooling response evaluation method as proposed in the above embodiment.

[0101] This embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements an electrothermal coupling liquid cooling response evaluation method as proposed in the above embodiment.

[0102] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0103] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0104] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0105] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0106] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. 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 be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating the response of electrothermal coupling liquid cooling, characterized in that, include: Based on the power consumption and heat generation behavior in liquid-cooled data centers, a real-time electrothermal coupling operation model reflecting the relationship between power and temperature variables is established between server power and liquid cooling heat exchange. Based on the electrothermal coupling real-time operation model, a liquid-cooled data center is established, which integrates chilled water circulation, cooling water circulation, and computer room air circulation. A multi-dimensional response index system is constructed, and a response collaborative control strategy is generated under service quality constraints and thermal safety threshold constraints. Under the multi-dimensional response index system, sensitivity analysis of power change behavior is carried out based on the response collaborative control strategy, a multi-dimensional index set is constructed, and the demand response potential is quantitatively calculated. The response coordination control strategies include a coordinated response mode in which the power of the server and the liquid-cooled data center decreases simultaneously; and a response mode in which the server power remains constant while the power of the liquid-cooled data center decreases.

2. The electrothermal coupling liquid cooling response evaluation method as described in claim 1, characterized in that: The electrothermal coupling real-time operation model includes, The corresponding heat generation is obtained based on the real-time power of the server. The mass flow rate of the cooling medium and the inlet and outlet temperature difference in the liquid cooling loop are described by the law of conservation of energy to obtain the current heat exchange. In the electrothermal coupling real-time operation model, a first-order inertial element is introduced to describe the electromechanical characteristics and time lag relationship between server power changes and liquid cooling heat exchange response.

3. The electrothermal coupling liquid cooling response evaluation method as described in claim 1 or 2, characterized in that: The liquid-cooled data center includes, The natural water cooling design, which uses a cooling tower (TA) as the core economizer, is mainly based on the WSE configuration of the cooling tower (TA) and determines the parameter settings of the chiller (LA), water pump, cooling tower (TA) and plate heat exchanger (R1). Based on the three operating modes, the power utilization efficiency value is optimized by using an outdoor natural cold source. The three operating modes include full natural cooling, partial natural cooling, and full host cooling. The aforementioned all-natural cooling includes a cooling tower (TA) supplying cooling to handle all the load; The partial natural cooling includes pre-cooling the return water using a cooling tower (TA); The complete host cooling includes conventional chiller (LA) cooling.

4. The electrothermal coupling liquid cooling response evaluation method as described in claim 3, characterized in that: The service quality constraints and thermal safety threshold constraints include, The quality of service constraints include limiting the server power adjustment range; The thermal safety threshold constraint includes limiting the server outlet temperature to no more than a preset upper temperature limit.

5. The electrothermal coupling liquid cooling response evaluation method as described in any one of claims 1, 2, and 4, characterized in that: The multi-dimensional response indicator system includes, Temperature dimension, power dimension, time dimension, and comprehensive evaluation dimension; The temperature dimension includes monitoring device safety through the server's maximum temperature and temperature rise index; The power dimension includes using adjustable power, rebound power, and rebound rate to describe the load reduction characteristics; The time dimension includes defining the adjustment time to assess the response speed and defining the response time to assess the time to recover to normal operation. The comprehensive evaluation dimensions include total response capacity, average response capacity, demand response quality indicators, and demand response comprehensive index. The comprehensive demand response index includes a priority reference for power dispatch that balances response capability, duration, and the negative impact of power rebound. A larger total responsive capacity and duration indicate a stronger response capability and higher priority. Conversely, a higher rebound rate poses a greater threat to the power grid, indicating a more severe response and lower priority. A comprehensive demand response index is constructed and expressed as follows: ， in, This represents the overall demand response index; Dynamic Voltage / Frequency Scaling (DVFS) technology is used to regulate the energy consumption of server clusters. The power consumption of the servers can be expressed as a function that is only related to the CPU frequency. Taylor expansion and Laplace transform of this function generate a complex frequency domain operating point power regulation control expression, which is expressed as: ， in, This represents the change in server power in the complex frequency domain. The slope representing the local linearization of the server's power consumption at a certain operating point. This indicates the frequency adjustment amount of the multi-frequency domain server; Given that the dynamic response of a compressor is limited by its electromechanical characteristics and exhibits a significant inertial time lag compared to the millisecond-level adjustment of a server CPU, a first-order inertial element is introduced into the power regulation to construct a complex frequency domain dynamic response model, expressed as: ， in, This represents the change in power of the liquid cooling system. This indicates the change in the frequency of the internal compressor. This represents the proportionality coefficient of the first-order inertial element in the cooling system. This represents the time constant of the first-order inertial element of the cooling system.

6. The electrothermal coupling liquid cooling response evaluation method as described in claim 5, characterized in that: The response collaborative control strategy includes: A queuing theory model is used to establish a functional relationship between the average server response time and power. The response tasks are allocated to the server and the liquid-cooled data center through the server power reduction coefficient. The server side uses dynamic voltage and frequency adjustment technology to regulate power, and the liquid-cooled data center coordinates by adjusting the compressor operating frequency. The relationship between the server outlet temperature and the safety threshold is judged in real time. Once the limit is reached or the safety threshold is exceeded, the feedback mechanism will execute forced cooling. By introducing a server power reduction factor, the remaining power regulation of 1-α is handled by the liquid cooling system, and this process can be expressed as: ， ， in, This represents the response power allocated to the server. This represents the response power allocated to the liquid cooling system. Indicates a demand response signal. This represents the server power reduction factor, the maximum value of which is determined by both the demand response signal and the power constraint, ranging from 0 to this maximum. Within the specified range, DC system operators can select appropriate values ​​based on operational needs.

7. The electrothermal coupling liquid cooling response evaluation method as described in any one of claims 1, 2, 4, and 6, characterized in that: The multi-dimensional indicator set includes, Within the time interval corresponding to the start and end times of power change, record the response power curves under different server power reduction coefficients or different temperature safety thresholds. The response power curve is integrally calculated, and the total response capacity, average response capacity, and rebound rate are obtained by adjusting the server power reduction coefficient and the server maximum temperature safety limit parameter, thereby generating a sensitivity spectrum.

8. An electrothermal coupling liquid cooling response evaluation system, employing the electrothermal coupling liquid cooling response evaluation method as described in any one of claims 1 to 7, characterized in that: It includes an electrothermal coupling modeling module, a response coordination control module, and a demand response assessment module; The electrothermal coupling modeling module is used to establish a dynamic coupling relationship between server power and liquid cooling heat exchange process based on a real-time electrothermal coupling model that reflects the correlation between power and temperature variables in the liquid-cooled data center. The module obtains the current heat exchange and power changes through the law of conservation of energy. The response coordination control module is used to generate three cooling modes, including full natural cooling, partial natural cooling and full host cooling, based on the electrothermal coupling real-time operation model. It adopts service quality constraints and thermal safety threshold constraints to establish a multi-dimensional response index system and generate a coordinated response mode between server power and liquid-cooled data center. The demand response assessment module is used to evaluate power change behavior through queuing theory model and sensitivity analysis. Based on the server power reduction factor and the response capability of the liquid cooling system, it quantitatively calculates the response depth, response duration and power rebound characteristics, and accurately assesses the demand response potential of the liquid-cooled data center.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the electrothermal coupling liquid cooling response evaluation method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the electrothermal coupling liquid cooling response evaluation method according to any one of claims 1 to 7.

Images