The heat treatment system of the computing center and the control method of the system

Abstract

This application relates to the field of computing center technology, and particularly to a heat handling system and control method for a computing center. The system includes a liquid cooling unit, a thermal buffer unit, and a power generation unit. The liquid cooling unit transfers heat generated by the computing equipment in the computing center to a liquid cooling medium. The thermal buffer unit buffers the heat carried by the liquid cooling medium and adjusts the flow distribution ratio of the buffered liquid cooling medium based on control commands. The power generation unit converts the heat carried by the incoming buffered liquid cooling medium into electrical energy. A controller is used to generate control commands based on the status data of at least one of the computing equipment, liquid cooling unit, thermal buffer unit, and power generation unit, and controls at least one of the liquid cooling unit, thermal buffer unit, and power generation unit based on the control commands. This solves the problems of simple connection between liquid cooling and power generation systems, poor coupling effect, and poor power generation stability in related technologies.

Description

Technical Field

[0001] This application relates to the field of computing center technology, and in particular to a computing center heat treatment system and a control method for the treatment system. Background Technology

[0002] With the rapid development of artificial intelligence, big data, and high-performance computing, the power density of servers in computing centers has increased dramatically. Traditional air cooling technology can no longer meet the heat dissipation requirements of high heat flux density chips, and liquid cooling technologies such as cold plate and immersion cooling have become the mainstream choice due to their superior heat dissipation performance. Liquid cooling systems can efficiently remove the heat generated by computing equipment, but the low-grade heat energy at the coolant outlet (40℃-70℃) is usually directly discharged, resulting in a large amount of energy waste. ORC (Organic Rankine Cycle) power generation technology can effectively convert medium and low temperature heat sources into electrical energy, providing a feasible path for recovering waste heat from liquid cooling in computing centers and improving energy utilization efficiency.

[0003] Although related technologies have attempted to combine liquid cooling systems with ORC power generation systems to achieve waste heat recovery, the transient changes in the temperature and flow rate of the liquid cooling working fluid caused by fluctuations in computing load directly impact the power generation system, resulting in poor power generation operation stability and poor coupling effect between the liquid cooling and power generation systems, thus failing to achieve efficient and stable heat recovery. Summary of the Invention

[0004] This application provides a heat treatment system for a computing center and a control method for the system, in order to solve the problems of simple connection between liquid cooling and power generation systems, poor coupling effect, and poor power generation stability in related technologies.

[0005] The first aspect of this application provides a heat treatment system for a computing center, comprising: a liquid cooling unit, a thermal buffer unit, and a power generation unit. The liquid cooling unit transfers heat generated by the computing equipment in the computing center to a liquid cooling medium. The thermal buffer unit buffers the heat carried by the liquid cooling medium and adjusts the flow distribution ratio of the buffered liquid cooling medium based on control commands. The power generation unit converts the heat carried in the buffered liquid cooling medium into electrical energy. A controller is provided to generate control commands based on the status data of at least one of the computing equipment, the liquid cooling unit, the thermal buffer unit, and the power generation unit, and to control at least one of the liquid cooling unit, the thermal buffer unit, and the power generation unit based on the control commands.

[0006] Optionally, it also includes: an auxiliary heat dissipation unit, used to receive the liquid cooling medium when the power generation unit is unable to handle the heat carried by the liquid cooling medium, thereby achieving heat dissipation for the computing center.

[0007] Optionally, the thermal buffer unit includes: a thermally stabilizing buffer tank, a bypass branch, and a heat exchanger assembly, wherein the thermally stabilizing buffer tank is used to absorb heat carried by the liquid cooling working fluid or to release heat to the liquid cooling working fluid; the bypass branch is used to introduce the liquid cooling working fluid to the auxiliary heat dissipation unit; and the heat exchanger assembly is used to transfer the heat carried by the liquid cooling working fluid to the ORC evaporator or ORC working fluid of the power generation unit.

[0008] Optionally, the thermal buffer unit also includes a multi-way proportional regulating valve group for regulating the flow ratio of liquid cooling working fluid into the thermal pressure buffer tank, the power generation unit, and the bypass branch.

[0009] Optionally, the multi-way proportional control valve assembly includes a first three-way proportional valve and a second bypass valve, wherein the first three-way proportional valve is used to adjust the flow ratio of the liquid cooling working fluid entering at least one of the power generation unit and the thermal stabilizing buffer tank; and the second bypass valve is used to introduce the liquid cooling working fluid into the auxiliary heat dissipation unit.

[0010] Optionally, the control layer includes a security control layer and an energy efficiency optimization layer, with the security control layer having a higher priority than the energy efficiency optimization layer. The control commands include a first command and a second command. The security control layer generates a first command when at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold, or when the temperature rise rate of at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate. Based on the first command, the security control layer controls a first three-way proportional valve and / or a second bypass valve to perform a target action. The target action includes at least one of adjusting the flow ratio of the first three-way proportional valve, closing the first three-way proportional valve, and opening the second bypass valve. The energy efficiency optimization layer calculates target operating parameters of at least one of the liquid cooling unit, the heat buffer unit, and the power generation unit based on a target optimization function, state data, and constraints. Based on the target operating parameters, the energy efficiency optimization layer generates a second command and controls at least one of the liquid cooling unit, the heat buffer unit, and the power generation unit based on the second command.

[0011] Optionally, the security control layer is further configured to: if at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold and less than a second preset threshold, or if the temperature rise rate corresponding to at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate and less than a second preset rate, then adjust the first three-way proportional valve based on the first instruction to reduce the flow rate ratio of the liquid cooling medium into the power generation unit; if at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset threshold, or if the temperature rise rate corresponding to at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset rate, then close the first three-way proportional valve and open the second bypass valve based on the first instruction.

[0012] Optionally, the controller also includes a state machine, which determines the corresponding operating mode based on the system's operating conditions and controls the working state of the control layer based on the operating mode.

[0013] Optionally, the objective function is: max J = P_ORC - (P_pumps + P_aux) - λ(T_target - T_out) 2 ; Where J is the net revenue objective function of the computing center's heat treatment system; P_ORC is the generator output power in the power generation unit; P_pumps is the power consumption of the liquid cooling main circulation pump and ORC working fluid pump in the liquid cooling unit; and P_aux is the power consumption of the auxiliary heat dissipation equipment in the auxiliary heat dissipation unit. T_target is the penalty weighting coefficient; T_target is the desired liquid cooling fluid outlet temperature; T_out is the liquid cooling fluid outlet temperature.

[0014] A second aspect of this application provides a control method for a computing center heat treatment system, used to control the computing center heat treatment system as described in the above embodiment. The method includes the following steps: acquiring status data of at least one of the computing equipment, liquid cooling unit, heat buffer unit, and power generation unit in the computing center; generating control instructions based on the status data; and controlling at least one of the liquid cooling unit, heat buffer unit, and power generation unit based on the control instructions.

[0015] A third aspect of this application provides a computer-readable storage medium having a computer program or instructions stored thereon, which is executed by a processor to perform a control method for a computing center heat treatment system as described above.

[0016] The fourth aspect of this application provides a computer program product, including a computer program or instructions, which, when executed, implement the control method of the computing center heat treatment system as described in the above embodiments.

[0017] Therefore, this application has at least the following beneficial effects: This application provides a heat handling system for a computing center, comprising a liquid cooling unit, a thermal buffer unit, and a power generation unit. The thermal buffer unit buffers the heat from the liquid cooling medium and adjusts the flow distribution ratio of the buffered liquid cooling medium according to control commands. The controller generates control commands based on the status data of at least one of the computing equipment, liquid cooling unit, thermal buffer unit, and power generation unit, and performs coordinated control of the liquid cooling, thermal buffer, and power generation units. This achieves efficient utilization of heat from the computing center, converting heat into electrical energy, reducing energy waste, improving the coupling between liquid cooling and heat generation, and mitigating the direct impact of computing load fluctuations on the power generation unit through the buffering effect of the thermal buffer unit, thus improving the operational stability of the power generation unit. Therefore, this solves the technical problems of simple connection between liquid cooling and power generation systems, poor coupling effect, and poor power generation stability in related technologies.

[0018] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0019] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of a computing center heat treatment system provided according to an embodiment of this application; Figure 2 This is a schematic diagram of a heat buffer unit provided according to an embodiment of this application; Figure 3 This is a flowchart of the collaborative control provided according to an embodiment of this application; Figure 4 This is a detailed structural diagram of the computing center heat treatment system provided according to an embodiment of this application; Figure 5 This is a comparison diagram of the effects of the power generation unit provided according to the embodiments of this application; Figure 6 This is a flowchart of a control method for a computing center heat treatment system provided according to an embodiment of this application.

[0020] Explanation of reference numerals in the attached figures: Server cold plate / immersion unit 1, main circulation pump 2, liquid cooling manifold 3, inlet temperature sensor 4, thermal pressure stabilizing buffer tank 5, first three-way proportional valve 6, second bypass valve 7, heat exchanger / dry cooler 8, ORC evaporator 9, expander-generator set 10, ORC condenser 11, ORC working fluid pump 12, controller 13, computing center heat treatment system 100, liquid cooling unit 101, thermal buffer unit 102, power generation unit 104. Detailed Implementation

[0021] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0022] Before describing the solution of this application, let me first introduce the relevant technologies of this application to help you understand the solution of this application.

[0023] ORC (Organic Regenerative Thermal Energy) is a mature low-grade thermal energy generation technology that utilizes low-boiling-point organic working fluids as the circulation medium to effectively convert medium- and low-temperature heat sources (such as industrial waste heat, geothermal energy, and solar energy) into electrical energy. Theoretically, applying ORC technology to recover liquid-cooled waste heat from computing centers has great potential. However, directly connecting the two in a simple physical way faces significant technical challenges in practical applications, mainly in the following three aspects: (1) Dynamic Coupling Mismatch Problem: The computing load of the computing center has typical characteristics of rapid and large-amplitude fluctuations (e.g., switching between AI model training and inference tasks, scheduling and migration of large-scale jobs, etc.), which directly leads to drastic and transient changes in the heat load, coolant flow rate and outlet temperature of the liquid cooling system in a short period of time. The ORC system, especially its core components evaporator and expander, has a large thermal inertia and a relatively narrow stable operating range. Without effective buffering, temperature and flow fluctuations on the liquid cooling side will directly impact the ORC system, which can easily cause drastic fluctuations in evaporation pressure, unstable superheat or subcooling at the heat exchanger outlet, and consequently lead to runaway expander speed, a sharp drop in power generation efficiency, and even frequent protective shutdowns, seriously affecting the reliability and continuous operation capability of the system.

[0024] (2) Conflict in core control objectives: The primary task of a liquid cooling system is to ensure the safety of computing devices, that is, to quickly and unconditionally remove the heat generated by the chips and ensure that their core temperature (such as the junction temperature of the CPU / GPU) is always kept below the safe upper limit (usually 85-95℃). This requires the cooling system to have extremely fast response speed and sufficient heat dissipation capacity. In contrast, the ORC power generation system pursues the maximization of energy recovery efficiency, and its thermodynamic efficiency (such as Carnot efficiency) is positively correlated with the temperature of the heat source. Therefore, the ORC system expects to obtain the highest and most stable heat source temperature possible. These two objectives are inherently contradictory in control strategies: for safety, the liquid cooling system tends to operate with a large flow rate and low temperature difference; while for efficiency, the ORC system expects the liquid cooling system to provide a heat source with a small flow rate and high temperature difference.

[0025] (3) Insufficient system integration and scheduling strategies: Solutions in related technologies often employ simple bypass valves or constant temperature control valves for coarse regulation, lacking a deep understanding and refined management of the dynamic characteristics of the entire coupled system. These solutions fail to establish a unified and dynamic collaborative scheduling mechanism between ensuring safety and maximizing net benefits. Therefore, when facing various complex operating conditions such as high load, low load, drastic load fluctuations, equipment failure, or maintenance, the system struggles to achieve globally optimal operation, resulting in a trade-off between different aspects and failing to fully realize the potential of waste heat recovery.

[0026] To address these issues, this application provides a heat treatment system for a computing center, solving at least one of the aforementioned technical problems. The system incorporates a dynamic heat buffer routing unit that combines heat capacity buffering and multi-path routing functions. This unit, through the synergistic action of a thermally stabilized buffer tank and a multi-port proportional valve group, smooths out thermal shocks on the liquid cooling side, providing a stable supply of high-quality heat with controlled temperature fluctuations to the ORC system. Simultaneously, a hierarchical control method combining safety hard constraints and energy efficiency optimization is employed, along with a state machine to achieve smooth mode switching under different operating conditions. This application significantly improves the continuous stable operating time and net power generation of the ORC system while ensuring absolute chip temperature safety. It upgrades the loose coupling between liquid cooling and waste heat power generation in the computing center to a controllable strong coupling, achieving an optimal balance between cooling reliability and heat recovery economy, resulting in significant energy-saving benefits and engineering application value.

[0027] Specifically, Figure 1 This is a schematic flowchart of a heat treatment system for a computing center provided in an embodiment of this application.

[0028] like Figure 1 As shown, the computing center heat processing system 100 includes: a liquid cooling unit 101, a heat buffer unit 102, a power generation unit 104, and a controller 13.

[0029] The liquid cooling unit 101 transfers the heat generated by the computing equipment in the computing center to the liquid cooling medium. The heat buffer unit 102 buffers the heat carried by the liquid cooling medium and adjusts the flow distribution ratio of the buffered liquid cooling medium based on control commands. The power generation unit 104 converts the heat carried in the buffered liquid cooling medium into electrical energy. The controller 13 generates control commands based on the status data of at least one of the computing equipment, liquid cooling unit 101, heat buffer unit 102 and power generation unit 104, and controls at least one of the liquid cooling unit 101, heat buffer unit 102 and power generation unit 104 based on the control commands.

[0030] It is understood that the embodiments of this application construct a computing center heat processing system 100, including a liquid cooling unit 101, a heat buffer unit 102, and a power generation unit 104. The heat buffer unit 102 is used to buffer the heat of the liquid cooling working fluid, and at the same time adjusts the flow distribution ratio of the buffered liquid cooling working fluid according to the control command. The controller is used to generate control commands based on the status data of at least one of the computing equipment, the liquid cooling unit 101, the heat buffer unit 102, and the power generation unit 104, and to perform coordinated control of the liquid cooling, heat buffer, and power generation units, thereby realizing the efficient utilization of the heat of the computing center, converting heat into electrical energy, reducing energy waste, improving the coupling between liquid cooling and heat generation, and through the buffering effect of the heat buffer unit, the direct impact of computing load fluctuations on the power generation unit can be mitigated, and the operational stability of the power generation unit can be improved.

[0031] The liquid cooling unit 101 in this embodiment can also be called the computing power liquid cooling main circuit, which directly provides cooling for the computing equipment and transfers the heat of the equipment to the liquid cooling medium. It includes a cold plate / immersion heat exchange unit 1, a main circulation pump 2, sensors, etc. The thermal buffer unit 102 can also be called a DTBRU (Dynamic Thermal Buffer & Routing Unit), which is the core buffer module connecting the liquid cooling unit and the power generation unit. It is used to realize heat smoothing and flow distribution. The power generation unit 104 is an ORC power generation circuit, which is a module that converts thermal energy into electrical energy through ORC evaporator 9, expander-generator set 10, etc. The status data can include the chip core temperature of the algorithm equipment, the temperature / pressure / flow rate of the liquid cooling medium, the ORC evaporation pressure, etc.

[0032] Specifically, the liquid cooling unit 101 in this embodiment is directly responsible for providing cooling for computing devices (such as servers, computing nodes, etc.), including a cold plate or immersion heat exchange unit 1 that is directly or indirectly thermally coupled to the server chip, a main circulation pump 2 (preferably a variable frequency speed control pump) for driving the circulation of coolant, a filter, a degassing device, and a series of temperature, pressure, and flow sensors for monitoring the system status. Its function is to efficiently transfer the heat generated by the computing device to a centralized coolant collection pipeline.

[0033] The power generation unit 104 in this embodiment is a standard Rankine cycle system, including an ORC evaporator 9 (thermally coupled to the heat exchanger assembly of the DTBRU), an expander-generator set 10 that converts thermal energy into mechanical energy and drives a generator, an ORC condenser 11 for condensing exhaust steam into liquid, an ORC working fluid pump 12 for driving the ORC working fluid circulation, and auxiliary equipment such as a liquid receiver and a safety valve. The organic working fluid used is selected according to the heat source temperature. For example, for a heat source of 60-90°C, environmentally friendly working fluids such as R1233zd(E), R245fa, and R1234ze(E) can be selected.

[0034] The system proposed in this application can be designed as a modular skid-mounted device, which facilitates standardized deployment in newly built computing centers or upgrades of existing computing centers' liquid cooling systems.

[0035] Furthermore, in one embodiment of this application, the computing center heat treatment system 100 of this application embodiment further includes: an auxiliary heat dissipation unit.

[0036] The auxiliary heat dissipation unit is used to receive the liquid cooling medium when the power generation unit is unable to handle the heat carried by the liquid cooling medium, thereby achieving heat dissipation for the computing center.

[0037] It is understood that the computing center heat treatment system 100 in this application embodiment also includes an auxiliary heat dissipation unit, which is used to receive the liquid cooling medium when the power generation unit 104 is unable to handle the heat carried by the liquid cooling medium, thereby achieving heat dissipation for the computing center and realizing dual protection for the computing center heat dissipation. This ensures that the heat dissipation needs of the computing equipment can still be met when the power generation unit 104 fails, thus avoiding the risk of overheating of the computing equipment.

[0038] The auxiliary heat dissipation unit in this application embodiment can also be called an auxiliary heat dissipation circuit. Specifically, the auxiliary heat dissipation unit is the safety bottom line of the system, including a heat exchanger / dry cooler 8, an electric / pneumatic valve that can achieve rapid switching, and an optional plate heat exchanger. When the ORC system is shut down, under maintenance, or malfunctions, or when the extreme high heat load exceeds the ORC processing capacity, it unconditionally takes over all heat dissipation tasks and prioritizes the cooling safety of the computing equipment.

[0039] Furthermore, in one embodiment of this application, the thermal buffer unit 102 includes: a thermally stabilizing buffer tank 5, a bypass branch, and a heat exchanger assembly, wherein the thermally stabilizing buffer tank 5 is used to absorb heat carried by the liquid cooling working fluid or release heat to the liquid cooling working fluid; the bypass branch is used to introduce the liquid cooling working fluid to the auxiliary heat dissipation unit; and the heat exchanger assembly is used to transfer the heat carried by the liquid cooling working fluid to the ORC evaporator or ORC working fluid of the power generation unit.

[0040] Among them, the thermal stabilizing buffer tank 5 can be a closed pressure vessel or an atmospheric pressure insulated water tank, filled with PCM (Phase Change Material) or high specific heat material to increase its equivalent heat capacity; the bypass branch is the fluid passage in the thermal buffer unit, used to directly introduce the liquid cooling working fluid around the power generation unit 104 into the auxiliary heat dissipation unit; the heat exchanger assembly is the heat exchange interface between the liquid cooling unit and the power generation unit 104, transferring the heat of the liquid cooling working fluid to the ORC working fluid or ORC evaporator, thereby providing a heat source for the power generation unit 104.

[0041] It is understood that the thermal buffer unit 102 in this embodiment includes a thermally stabilizing buffer tank 5, a bypass branch, and a heat exchanger assembly to achieve the functions of heat buffering, path switching, and heat exchange. The thermally stabilizing buffer tank 5 can achieve peak and valley filling of heat, suppressing temperature fluctuations of the liquid cooling working fluid to provide a stable heat source for the power generation unit 104. The bypass branch can provide a fast-switching heat dissipation path for the liquid cooling working fluid to shorten the response time of the auxiliary heat dissipation unit. The heat exchanger assembly realizes efficient heat exchange between the liquid cooling working fluid and the working fluid of the power generation unit.

[0042] Specifically, the thermal stabilizing buffer tank 5 in this embodiment can be a sealed pressure vessel or an atmospheric pressure insulated water tank. To enhance its buffering capacity, a special layered flow guiding structure can be set inside the tank to promote thermal stratification and slow down the mixing of hot and cold fluids. The outside of the tank is wrapped with a high-efficiency insulation layer to reduce heat loss. In order to achieve a larger equivalent heat capacity within a limited volume, the buffer tank can be filled with PCM modules or high specific heat solid material modules. The design of its effective heat capacity (C_buf) should meet certain engineering criteria, preferably: C_buf>= Q_step / ΔT_allow, where Q_step represents the maximum one-time step thermal shock (joules) that may occur on the computing side within a preset control time window, and ΔT_allow is the maximum temperature fluctuation range (e.g., <= 1-3℃) that is allowed to be transferred to the heat source side of the ORC evaporator.

[0043] The heat exchanger assembly in this embodiment serves as the heat exchange interface between the liquid cooling loop and the ORC loop. It can be a plate heat exchanger, a shell-and-tube heat exchanger, or a microchannel heat exchanger to achieve efficient heat exchange. It can be a standalone component or integrated with the ORC evaporator as a single evaporative heat exchange assembly. Through coordinated control of this heat exchanger and the aforementioned valve group, the rate of temperature change of the heat entering the ORC evaporator (i.e., the liquid cooling working fluid) can be actively controlled at an extremely low level, preferably no higher than 1°C / minute. This control target can also be set in reverse based on the tolerance of the selected expander to the rate of change of evaporative pressure.

[0044] Furthermore, it should be noted that the thermal buffer pressure stabilizing tank 5 in this application embodiment can also be set as a double buffer tank or a layered buffer tank depending on the specific situation. Specifically, for larger-scale systems with more severe fluctuations, a single thermal pressure stabilizing buffer tank 5 can be designed as two buffer tanks connected in series, or a physical partition can be set in a single tank to form a clear high-temperature zone and a low-temperature zone. Through a more refined layered liquid extraction and replenishment strategy, the temperature fluctuation of the heat source can be further reduced, and a higher level of stability can be achieved.

[0045] Furthermore, in one embodiment of this application, the thermal buffer unit 102 further includes a multi-way proportional regulating valve group for regulating the flow ratio of liquid cooling working fluid into the thermal pressure stabilizing buffer tank, the power generation unit, and the bypass branch.

[0046] It is understood that the thermal buffer unit 102 in this application embodiment also includes a multi-way proportional regulating valve group, which actively regulates the flow ratio of liquid cooling working fluid into the thermal pressure stabilizing buffer tank 5, the power generation unit, and the bypass branch, so as to realize the fine and controllable distribution of flow within the thermal buffer unit 102.

[0047] The multi-port proportional control valve assembly of this application embodiment can realize continuous proportional control to adapt to the gradual and sudden changes in computing load and improve the smoothness of system regulation.

[0048] Furthermore, in one embodiment of this application, the multi-way proportional control valve group includes a first three-way proportional valve 6 and a second bypass valve 7, wherein the first three-way proportional valve 6 is used to adjust the flow ratio of liquid cooling working fluid entering at least one of the power generation unit 104 and the thermal stabilizing buffer tank 5; and the second bypass valve 7 is used to introduce liquid cooling working fluid into the auxiliary heat dissipation unit.

[0049] The first three-way proportional valve 6 has one inlet and two outlets. The inlet is connected to the liquid cooling unit, and the outlets are connected to the power generation unit 104 and the thermal stabilizing buffer tank 5, respectively. The second bypass valve 7 is a fast-response valve that enables the rapid introduction of liquid cooling fluid into the auxiliary heat dissipation unit.

[0050] It is understood that the multi-way proportional regulating valve group in this application embodiment includes a first three-way proportional valve 6 and a second bypass valve 7. The first three-way proportional valve 6 is used for flow ratio regulation between the power generation unit and the thermal stabilizing buffer tank 5, and the second bypass valve 7 is used for flow control of the liquid cooling working fluid to the auxiliary heat dissipation unit. By dividing the multi-way proportional regulating valve group, the control of each valve corresponds to the target, thereby improving the accuracy and response speed of flow regulation. By setting a separate second bypass valve 7 for the auxiliary heat dissipation unit, rapid path switching under emergency conditions can be achieved, thereby improving the response efficiency of safety protection.

[0051] Specifically, the multi-port proportional control valve assembly of this application embodiment is key to achieving precise flow distribution. It includes at least a first three-way proportional valve 6, which enables continuous and smooth flow distribution between its three ports (connected to the ORC heating branch, bypass branch, and buffer tank charging / discharging branch, respectively). Furthermore, it includes a second fast bypass valve, used in emergency situations (such as in safety protection mode) to completely cut off the heating branch to the ORC at the fastest speed (e.g., within 1-2 seconds) and redirect all flow to the auxiliary cooling circuit. The valve assembly can also be configured with check valves, anti-backflow valves, and electromagnetic emergency shut-off valves as needed to enhance system operational safety.

[0052] In summary, the heat buffer unit 102 of this application embodiment is connected in series between the liquid cooling manifold 3 of the liquid cooling main circuit and the ORC evaporator 9, and its structure is as follows: Figure 2 As shown, it includes at least a thermally stabilizing buffer tank 5 (a container with a specific heat capacity for temporarily storing or releasing heat); a multi-way proportional control valve assembly (a set of electric or pneumatic control valves that can continuously distribute flow proportionally between different paths); a bypass branch (a fluid passage for bypassing the ORC system in a specific mode); and a heat exchanger assembly (for transferring heat from the liquid-cooled working fluid to the ORC working fluid).

[0053] Furthermore, to achieve precise sensing and closed-loop control of the DTBRU's internal state, high-precision temperature, pressure, and flow sensors are installed at key locations such as the DTBRU's inlet and outlet, the upper, middle, and lower parts of the buffer tank, and the inlet and outlet of the ORC evaporator. Data from these sensors is sent to the controller to construct a closed-loop control system for heat source stability. Simultaneously, the controller integrates diagnostic and protection logic for various potential faults, including over-temperature, over-pressure, low flow, and valve jamming.

[0054] Further, in one embodiment of this application, the controller 13 includes a security control layer and an energy efficiency optimization layer, with the security control layer having a higher priority than the energy efficiency optimization layer. The control commands include a first command and a second command. The security control layer is used to generate a first command when at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold, or when the temperature rise rate corresponding to at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate. Based on the first command, the security control layer controls the first three-way proportional valve 6 and / or the second bypass valve 7 to perform a target action. The target action includes at least one of adjusting the flow ratio of the first three-way proportional valve 6, closing the first three-way proportional valve 6, and opening the second bypass valve 7. The energy efficiency optimization layer is used to calculate the target operating parameters of at least one of the liquid cooling unit 101, the heat buffer unit 102, and the power generation unit 104 based on the target optimization function, state data, and constraints. Based on the target operating parameters, the energy efficiency optimization layer generates a second command and controls at least one of the liquid cooling unit 101, the heat buffer unit 102, and the power generation unit 104 based on the second command.

[0055] Among them, the security control layer has the highest priority, with the sole objective of ensuring the heat dissipation safety of computing equipment, and operates under hard constraints; the energy efficiency security layer has a priority security control layer, which, under the constraints of the security control layer, aims to maximize the net power of waste heat power generation; the first instruction is a control instruction generated by the security control layer, focusing on the execution of security protection actions; the second instruction is a control instruction generated by the energy efficiency optimization layer, focusing on the optimization of system operating parameters; the constraints are defined thresholds for the safe operation of the system, such as the upper limit of chip core temperature, the upper limit of liquid coolant outlet temperature, the upper limit of ORC evaporation pressure, and the overheat range; the first preset threshold and the first preset rate are used to characterize the temperature threshold and temperature rise rate threshold for the system to enter the warning state, and the specific values ​​can be set according to the specific situation, without specific limitations.

[0056] It is understood that the controller in this application embodiment is a hierarchical control architecture, with the safety control layer having a higher priority than the energy efficiency optimization layer. This resolves the conflict between heat dissipation safety and power generation efficiency. Specifically, the safety control layer generates a first instruction when at least one of the temperature of the computing device and the outlet temperature of the liquid coolant exceeds a first preset threshold, or when the temperature rise rate of at least one of the temperature of the computing device and the outlet temperature of the liquid coolant exceeds a first preset rate. Based on the first instruction, it controls the multi-channel proportional regulating valve group to execute the target action to ensure system heat dissipation safety. The energy efficiency optimization layer can calculate the target operating parameters of each unit based on the target optimization function, system state data, and safety constraints, and generate a second instruction based on the target operating parameters. The second instruction is used to control the system and maximize power generation efficiency within the range of satisfying heat dissipation safety.

[0057] Specifically, this application divides the system's control logic into two distinct priority layers: a security control layer and an energy optimization layer. The security control layer is the highest priority layer, its sole objective being to ensure the security of the computing equipment. It operates with hard constraints that must never be violated. When any security risk is detected, it can unconditionally reject or interrupt the control commands of the lower layer, namely the energy optimization layer. The energy optimization layer is a secondary optimization layer, its objective being to maximize net power generation or minimize equivalent comprehensive energy efficiency within the safe area defined by the security control layer by optimizing the scheduling of system resources.

[0058] Furthermore, in one embodiment of this application, the security control layer is further configured to: if at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold and less than a second preset threshold, or if the temperature rise rate corresponding to at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate and less than a second preset rate, then adjust the first three-way proportional valve 6 based on a first instruction to reduce the flow rate ratio of the liquid cooling medium flowing into the power generation unit; if at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset threshold, or if the temperature rise rate corresponding to at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset rate, then close the first three-way proportional valve 6 and open the second bypass valve 7 based on a first instruction.

[0059] The second preset threshold and the second preset rate can be set according to specific circumstances, and there are no specific limitations on them. They are the emergency protection temperature threshold and the temperature rise rate of the system, respectively.

[0060] It is understood that the security control layer of this application embodiment can adjust the first three-way proportional valve 6 based on the first instruction when at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold and less than a second preset threshold, or when the temperature rise rate corresponding to the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate and less than a second preset rate, thereby reducing the flow rate ratio of the liquid cooling medium into the power generation unit 104, so as to preserve the waste heat power generation capacity as much as possible while ensuring heat dissipation safety and improving the energy recovery efficiency of the system; when at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset threshold, or when the temperature rise rate corresponding to at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset rate, it indicates an emergency state, and then closes the first three-way proportional valve 6 and opens the second bypass valve 7 based on the first instruction to quickly reduce the temperature of the computing device to a safe range.

[0061] Furthermore, in one embodiment of this application, the objective optimization function is: max J = P_ORC - (P_pumps + P_aux) - λ(T_target - T_out) 2 ; Where J is the net revenue objective function of the computing center heat treatment system, representing the net power generation revenue under the heat dissipation safety constraints; P_ORC is the generator output power in the power generation unit; P_pumps is the power consumption of the liquid cooling main circulation pump and ORC working fluid pump in the liquid cooling unit; and P_aux is the power consumption of the auxiliary heat dissipation equipment in the auxiliary heat dissipation unit. T_target is the penalty weighting coefficient; T_target is the desired liquid cooling fluid outlet temperature; T_out is the liquid cooling fluid outlet temperature.

[0062] In summary, the controlled object, variables, constraints, and hierarchical control of this application are as follows, and the hierarchical control flow is as follows: Figure 3 As shown.

[0063] Measured variables include, but are not limited to: chip core temperature T_j (which can be read directly from the chip’s built-in sensor or estimated indirectly through a power and heat dissipation model), liquid cooling main circuit outlet temperature T_out, stratification temperature inside the DTBRU buffer tank T_buf, ORC evaporation pressure p_evap, ORC exhaust steam superheat ΔT_sh, flow rate m_dot of each key branch, and ambient dry / wet bulb temperature T_amb, etc.

[0064] The quantities to be executed include, but are not limited to: the rotational speed n_main of the main circulation pump in the liquid-cooled main circuit, the rotational speed n_wf of the ORC working fluid pump, the opening degree u_v1 of the three-way proportional valve in the DTBRU, the opening degree u_v2 of the quick bypass valve, the opening degree of the expander guide vane or the opening degree u_exp of the bypass regulating valve, etc.

[0065] The hard safety constraints include: the chip core temperature T_j must always be lower than its specification limit T_j,max (e.g., 85℃ or 95℃); the liquid coolant outlet temperature T_out must be lower than a certain set value T_out,max (e.g., 75℃); the evaporation pressure, superheat, etc. of the ORC system must be maintained within the stable operating range allowed by the equipment manufacturer; at the same time, in order to prevent system oscillation and excessive equipment wear, the rate of change of all actuators (pumps, valves) |Δu| / Δt must also be limited to r_max.

[0066] The security control layer strategy operates with an extremely short control cycle (preferably 0.5-2s) to ensure instantaneous response to security risks. Its core strategies include: S1: Real-time monitoring and early warning, continuously collecting key safety indicators such as T_j, T_out, and m_dot, and calculating their temperature rise rate dT / dt; S2: Preventive intervention. When T_j or T_out, or the corresponding rate of temperature rise, approaches but has not yet reached the limit value, i.e., entering the warning zone, the safety layer will immediately take preventive measures. These measures take priority over energy efficiency optimization. For example, the main circulation pump speed n_main is increased to enhance cooling capacity, while the three-way proportional valve u_v1 is adjusted to reduce the flow supplied to ORC and direct more heat into the buffer tank or directly bypass to the auxiliary heat dissipation circuit. S3 protective switching: When T_j or T_out, or the corresponding temperature rise rate, reaches the protection threshold, or an emergency fault such as main circuit flow loss, valve jamming, or serious abnormality of ORC system is detected, the safety layer will execute the strongest protection action, which includes: immediately forcibly closing the heating branch to ORC through command, while fully opening the bypass valve to the auxiliary heat dissipation circuit, and triggering ORC system to execute emergency load reduction or shutdown procedure. S4: Recovery and handover. When the system state returns to normal, for example, when T_j falls back to the safe recovery threshold and remains stable for a period of time (e.g., t_hold=30 seconds), the safety layer will release the forced intervention, allowing the system to gradually and smoothly resume heating to ORC, and return control to the energy efficiency optimization layer.

[0067] The energy efficiency optimization layer strategy aims to maximize net power generation. This layer operates within the limits allowed by the safety control layer, with a relatively long control cycle (preferably 2-10 seconds), and can employ model-based predictive control strategies (prediction time domain preferably 30-300 seconds). Advanced control algorithms that can be used include model predictive control, constraint projection algorithms based on reinforcement learning, or simpler methods such as lookup tables plus online correction.

[0068] The objective function J for optimization can be defined as: max J = P_ORC - (P_pump_main + P_pump_wf + P_aux) - λ(T_target, ref - T_out) 2Where P_ORC is the generator output power of the ORC system; P_pump_main and P_pump_wf are the power consumption of the liquid-cooled main circulation pump and the ORC working fluid pump, respectively, and P_pumps = P_pump_main + P_pump_wf; P_aux is the power consumption of auxiliary heat dissipation equipment such as the dry cooler fan; λ is the penalty weighting coefficient; and T_target is the reference value of the liquid-cooled outlet temperature that is expected to be maintained under the premise of meeting the safety margin. During optimization, the current state vector is first established based on the sensor-collected T_j, T_out, p_evap, ΔT_sh, flow rate m_dot of each branch, and ambient temperature T_amb. Then, P_pump_main, P_pump_wf, and P_aux are calculated based on the pump similarity law and equipment performance curves, and P_ORC is calculated based on the ORC evaporator heat exchange rate Q_evap = m_dot×c_p×(T_in - T_out) and the expander efficiency model or performance mapping. Subsequently, under the constraints of T_j≤T_j,max, T_out≤T_out,max, p_evap being within the allowable range, ΔT_sh meeting the stable evaporation requirements, and the limited change rate of each actuator, rolling optimization is performed on n_main, u_v1, u_v2, n_wf, and u_exp. Finally, the target operating parameters that maximize the objective function J are output, and the corresponding pump valves are driven by the second instruction to maximize the net power generation benefit under the premise of heat dissipation safety.

[0069] Specifically, P_pump_main and P_pump_wf can be obtained from the power model P≈Δp·Q / η or a pre-calibrated power curve based on the relationship between the corresponding pump speed and flow rate; P_aux can be obtained from the fan power model based on the number of start / stop units, speed, and ambient temperature of the dry cooler fans; P_ORC can be calculated from the ORC working fluid evaporation heat absorption, the expander isentropic efficiency, and the generator efficiency. By mapping the above power and power consumption to the objective function J, the controller can quantify the relationship between increasing power generation revenue and increasing auxiliary energy consumption, thereby selecting the operating point with the optimal net revenue.

[0070] Furthermore, in one embodiment of this application, the controller 13 further includes a state machine, which determines the corresponding operating mode based on the operating conditions of the system and controls the working state of the control layer based on the operating mode.

[0071] Among them, the state machine is the top-level operation mode management module of the controller 13; the operating condition is the operating state determined by the system's state data, such as stable high load, rapid fluctuation of computing load, extreme security risk, shutdown maintenance, etc.; the operating mode can include stable power generation mode, rapid fluctuation suppression mode, safety protection bypass mode, etc.

[0072] It is understood that the embodiments of this application can determine the corresponding operating mode based on the operating conditions of the system, control the working state of the control layer based on the operating mode, realize the smooth switching of the system operating mode, avoid the control logic confusion caused by changes in operating conditions, and improve the adaptability of the system under complex operating conditions.

[0073] Specifically, to ensure smoother and more orderly switching between different operating modes and to avoid equipment wear and system oscillations caused by frequent switching, this application preferably introduces a top-level state machine to manage the overall macroscopic operating state of the system. This state machine includes at least the following modes: In M0 shutdown maintenance mode, the ORC system is completely shut down, and all liquid cooling waste heat is discharged through the auxiliary heat dissipation circuit. In the M1 preheating grid-connected mode, during the initial startup of the ORC system, the DTBRU prioritizes charging itself and the ORC evaporator. Once the grid-connected conditions are met, the expander starts up and connects to the grid. In the M2 stable power generation mode, the system is in normal operation, controlled by the energy efficiency optimization layer, while the safety layer monitors it in the background. In the M3 fluctuation suppression mode, the system enters this mode when a drastic fluctuation in computing load is detected. DTBRU plays its core buffering role, actively absorbing thermal shocks and strictly limiting the rate of change of ORC evaporation pressure, striving for external fluctuations and internal stability. The M4 safety protection mode, triggered by the safety layer, forcibly switches out of the ORC system to fully ensure heat dissipation safety. The switching between modes is determined by a set of explicit logical conditions, which integrate key state variables such as T_j, T_out, and p_evap, the output results of the fault diagnosis module, and the minimum mode dwell time set to prevent system jitter.

[0074] In summary, the overall structure of the computing center heat treatment system in this application embodiment is as follows: Figure 4 As shown, it includes the computing power liquid cooling main circuit (i.e., liquid cooling unit), ORC circuit (i.e. power generation unit), and DTBRU (i.e., thermal buffer unit).

[0075] Among them, the computing power liquid cooling main circuit is used to transfer the heat generated by the computing power equipment to the liquid cooling working fluid; The ORC power generation circuit includes an evaporator, an expander-generator set, a condenser, and a working fluid pump connected in sequence, which are used to generate electricity using thermal energy; The DTBRU is located between the evaporator of the computing power liquid cooling main circuit and the ORC power generation circuit. It is used to buffer and route heat from the computing power liquid cooling main circuit. The DTBRU includes: a thermal pressure stabilizing buffer tank, which is connected to the computing power liquid cooling main circuit and is used to absorb or release heat to smooth the temperature fluctuation of the liquid cooling fluid; and a multi-port proportional regulating valve group, which is located between the computing power liquid cooling main circuit and the evaporator, the thermal pressure stabilizing buffer tank and the bypass branch, and is used to continuously adjust the flow ratio of the liquid cooling fluid flowing into each branch according to the control command.

[0076] The system of this application also includes: an auxiliary heat dissipation circuit, and a multi-way proportional regulating valve group capable of controlling the introduction of all or part of the liquid cooling working fluid into the auxiliary heat dissipation circuit.

[0077] The multi-port proportional control valve group of this application includes a first three-port proportional valve, whose inlet is connected to the computing power liquid cooling main circuit, and whose two outlets are respectively connected to the evaporator and the thermal stabilizing buffer tank of the ORC power generation circuit; and a second bypass valve, which is used to directly guide the liquid cooling working fluid from the computing power liquid cooling main circuit to the auxiliary heat dissipation circuit in emergency or specific working conditions.

[0078] The system of this application also includes: a central controller, connected to multiple sensors installed on the computing power liquid cooling main circuit, the dynamic thermal buffer routing unit, and the ORC power generation circuit, and used to execute a dual-objective cooperative control method. The flow of the dual-objective system control method includes: Security control layer (hard constraint): Real-time monitoring of the core temperature of the computing equipment (T) not exceeding the temperature of the first control cycle. j ) and / or the outlet temperature of the liquid cooling medium (T out When the monitored value exceeds the preset safety threshold, a protective action is immediately executed. This action includes, but is not limited to, reducing or cutting off the flow to the ORC generator circuit evaporator through the multi-way proportional regulating valve group, and increasing the flow to the auxiliary heat dissipation circuit. Energy efficiency optimization layer (soft constraint): Under the premise that the safety control layer constraints are met, the optimization algorithm runs at a rate not higher than the second control cycle, with the goal of maximizing net power generation, and coordinates the adjustment of the circulation pump speed of the computing power liquid cooling main circuit, the opening of the multi-way proportional regulating valve group, and the operating parameters of the ORC power generation circuit.

[0079] This application also includes a state machine for switching between at least three modes based on the system's operating state: M1: ORC stable power generation mode, which is controlled primarily by energy efficiency optimization. M2: Rapid fluctuation suppression mode, which absorbs thermal shock by a dynamic thermal buffer unit to limit the rate of temperature change of the heat source entering the ORC evaporator; M3: Safety protection bypass mode, completely cuts off the heat supply to ORC, and uses an auxiliary heat dissipation circuit to cool the computing equipment.

[0080] Furthermore, it should be noted that to meet the extreme reliability requirements of the computing center, critical components can be configured redundantly. For example, two identical ORC power generation circuits and DTBRUs can be configured in parallel, so that if one fails or is under maintenance, the other can seamlessly take over. Critical sensors (such as T_j and T_out) can be configured with a 2 out of 3 (2oo3) redundancy. All critical valves are selected as fail-safe, meaning that in the event of power failure or signal loss, they can automatically switch to a position that ensures cooling safety (e.g., automatically return to the fully bypass position).

[0081] The heat treatment system for the computing center in this application can produce the following effects: 1. Extremely high operational stability and reliability: Through the core DTBRU design, this application can effectively transform the rapid and disordered thermal shocks on the computing side into a smooth, controllable, and high-quality heat source supplied to the ORC system. Simulation and experimental data show that this structure can reduce the temperature fluctuation of the heat source entering the ORC evaporator by 80%-95%, and control the temperature change rate to within 1℃ / minute. This greatly reduces the adjustment frequency and start-up / shutdown times of the ORC expander, significantly improves the continuous stable operation time of the system, and increases the availability from less than 70% in the traditional direct coupling method to over 95%. 2. Absolute Thermal Safety: Safety is the fundamental design principle of this application. Through a layered control architecture and the implementation of hard constraints, this application ensures that the thermal safety of the computing device has the highest priority under any operating condition. In the event of extreme load shocks or system failures, the safety protection mechanism can make decisions in milliseconds and complete the switching action in seconds, switching the system to a fully bypassed safe mode, ensuring that the chip temperature is always below the safe red line, thus achieving the design goal of interruptible power generation and uninterrupted cooling.

[0082] 3. Significant Energy Efficiency Improvement and Economic Benefits: Under the premise of ensuring safety and stability, the energy efficiency optimization layer, through refined dynamic scheduling, maximizes the temperature of heat sources available for power generation while minimizing the parasitic power consumption of auxiliary equipment such as pumps and fans, thereby maximizing the system's net power generation. Compared with traditional direct emission schemes, this application can recover more than 50% of liquid cooling waste heat for power generation. In a typical 10MW computing center application scenario, the annual power generation is expected to reach millions of kWh, which can not only significantly reduce the equivalent power utilization efficiency of the data center, but also bring considerable direct economic benefits, and the investment payback period can typically be shortened to 3-5 years.

[0083] 4. Excellent engineering adaptability and scalability: The structure proposed in this application has a high degree of modularity and standardization, and can be designed as an independent skid-mounted device. This makes it suitable not only for newly built high-density data centers, but also for easy energy-saving retrofitting of existing data centers with deployed liquid cooling systems, demonstrating strong engineering flexibility and universality.

[0084] The computing center heat handling system proposed in this application includes a liquid cooling unit, a thermal buffer unit, and a power generation unit. The thermal buffer unit buffers the heat of the liquid cooling medium and adjusts the flow distribution ratio of the buffered liquid cooling medium according to control commands. The controller generates control commands based on the status data of at least one of the computing equipment, the liquid cooling unit, the thermal buffer unit, and the power generation unit, and performs coordinated control of the liquid cooling, thermal buffer, and power generation units. This achieves efficient utilization of the heat of the computing center, converts heat into electrical energy, reduces energy waste, improves the coupling between liquid cooling and heat generation, and mitigates the direct impact of computing load fluctuations on the power generation unit through the buffering effect of the thermal buffer unit, thereby improving the operational stability of the power generation unit.

[0085] The following describes the heat treatment system of the computing center according to specific embodiments of this application.

[0086] Example 1: Stable high-load continuous power generation condition.

[0087] In a 10MW computing center where the system of this application is deployed, the computing load rate is consistently maintained at a high level of 80%-95% during weekdays. The target value T_out,ref of the liquid cooling system outlet temperature is set to 68℃ by the energy efficiency optimization layer. This temperature has a sufficient safety margin of more than 20℃ within the safety specifications given by the chip supplier (e.g., T_j,max = 90℃).

[0088] Under this operating condition, the central controller 13 enters M2: stable power generation mode. Controller 13 instructs the opening degree u_v1 of the three-way proportional valve 6 to maintain a high heating ratio (e.g., 95%), allowing most of the heat-carrying coolant (approximately 68°C) to flow through the heat exchanger assembly of the DTBRU, transferring heat to the evaporator 9 in the ORC loop. A small portion of the flow (e.g., 5%) flows into the thermal pressure stabilization buffer tank 5 to suppress temperature disturbances caused by minor fluctuations in computing power. The ORC system uses R1233zd(E) as the working fluid, with an evaporation temperature of approximately 60°C and an evaporation pressure stably maintained at around 0.2 MPa. The condenser side is cooled by a dry cooler 8 at an ambient temperature of 25°C, with a condensation pressure of approximately 0.08 MPa. Under this stable operating condition, the output power of the ORC expander-generator set 10 is close to its rated design value; the total power generation P_ORC of the system is measured to be 250 kW. After deducting the power consumption of the liquid-cooled main circulation pump 2 (approximately 40kW), the ORC working fluid pump 12 (approximately 10kW), and the dry cooler fan (approximately 50kW), the system achieves a net power generation of approximately 150kW. This power can be directly fed into the data center's internal power grid to offset some of the power consumption of IT equipment or auxiliary facilities, significantly reducing the data center's equivalent overall energy consumption.

[0089] Example 2: Stable heating and power maintenance under rapid fluctuation conditions.

[0090] As the scene shifts to nighttime, the computing center runs a hybrid AI training and inference task, causing its total power to periodically jump from 3MW (30% load) to 9MW (90% load) within minutes.

[0091] When the load suddenly jumps from 30% to 90%, the liquid cooling main circuit outlet temperature T_out rapidly rises from 50°C to nearly 70°C within a short period (e.g., within 30 seconds). At this time, the central controller 13 automatically switches to M3: fluctuation suppression mode. The thermal stabilizing buffer tank 5 of the DTBRU (filled with PCM, with a phase change point of 60°C) begins to absorb a large amount of thermal shock, and the PCM inside undergoes a solid-liquid phase change, effectively clamping the temperature of the coolant flowing out of the DTBRU. At the same time, the controller 13 fine-tunes the three-way proportional valve 6, temporarily reducing the flow directly supplied to the evaporator 9 and increasing the flow into the buffer tank 5. Through this active buffering and routing adjustment, although the DTBRU inlet temperature jumps by 20°C within 30 seconds, the temperature change rate of its outlet (i.e., the heat source supplied to the ORC evaporator) is successfully controlled to below 1°C / minute, and the temperature fluctuation range is suppressed within + / -1.5°C.

[0092] Thanks to this, the evaporation pressure p_evap of the ORC system only experiences small, smooth fluctuations, and the expander-generator set 10 does not require drastic adjustments or trigger protection. Comparative simulations show that... Figure 5 As shown, Figure 5 This paper demonstrates a comparison of the heat source temperature response and ORC system output power under typical step load fluctuation conditions, with and without the DTBRU of this application. Without the DTBRU, such a severe thermal shock would cause the ORC system to shut down protectively due to excessive evaporation pressure change rate. However, with the application of this application, the ORC system output power only drops briefly by about 10% at the initial stage of load switching, and then recovers to a stable high power level matching the 9MW load within 2-3 minutes, achieving continuous and uninterrupted power generation under severe fluctuation conditions.

[0093] Example 3: Safety protection bypass operation (extreme load or fault).

[0094] Suppose that after a system maintenance, a valve in the liquid cooling main circuit fails to fully open, causing the main circuit flow rate m_dot to be less than 50% of the normal value. At the same time, a high-priority computing task suddenly comes online, causing the computing load to instantly reach 100% (10MW).

[0095] At this time, the safety control layer detected that the chip core temperature T_j rapidly increased within a few seconds and reached the preset maximum protection threshold T_j, trip = 85℃. The central controller 13 immediately triggered M4: safety protection bypass mode. This mode has the highest system authority and its actions include: (1) immediately (<1 second) issuing commands to the three-way proportional valve 6 and the fast bypass valve 7 to forcibly close the passage to the ORC evaporator 9 completely, while switching 100% of the coolant flow to the auxiliary heat dissipation circuit; (2) at the same time, sending an emergency shutdown command to the ORC system to safely shut it down; (3) at the same time, increasing the speed of the liquid-cooled main circulation pump 2 to 100% of the rated speed and starting all the fans of the dry cooler 8 in the auxiliary heat dissipation circuit to operate at maximum power.

[0096] Under this series of forced interventions, although ORC power generation was completely interrupted, the entire system's heat dissipation capacity was used to protect the computing equipment. Actual test data shows that within 15 seconds of switching to M4 mode, the chip core temperature T_j rapidly dropped from a peak of 85℃ to a safe range below 75℃, effectively preventing hardware damage or system crashes due to overheating, fully demonstrating the design principle of uninterrupted cooling.

[0097] Figure 6 This is a flowchart illustrating the control method of the heat treatment system of the computing center according to an embodiment of this application.

[0098] like Figure 6 As shown, the control method for the heat treatment system of the computing center, used to control the aforementioned heat treatment system, includes the following steps: In step S101, the status data of at least one of the computing power equipment, liquid cooling unit, thermal buffer unit and power generation unit in the computing power center are obtained.

[0099] In step S102, control commands are generated based on the status data, and at least one of the liquid cooling unit, thermal buffer unit, and power generation unit is controlled based on the control commands.

[0100] It should be noted that the foregoing explanation of the embodiment of the computing center heat treatment system also applies to the control method of the computing center heat treatment system in this embodiment, and will not be repeated here.

[0101] This application also provides a computer-readable storage medium storing a computer program or instructions thereon, which, when executed by a processor, implements the control method of the above-described computing center heat treatment system.

[0102] This application also provides a computer program product, including a computer program or instructions, which, when executed, implement the control method of the computing center heat treatment system described above.

[0103] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0104] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0105] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0106] It should be understood that the various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented using 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 more 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 (FPGAs), field-programmable gate arrays (FPGAs), etc.

[0107] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

Claims

1. A heat processing system for a computing center, characterized in that, include: The system comprises a liquid cooling unit, a thermal buffer unit, and a power generation unit. The liquid cooling unit transfers the heat generated by the computing equipment in the computing center to the liquid cooling medium. The thermal buffer unit buffers the heat carried by the liquid cooling medium and adjusts the flow distribution ratio of the buffered liquid cooling medium based on control commands. The power generation unit converts the heat carried by the buffered liquid cooling medium into electrical energy. The controller is configured to generate the control command based on the status data of at least one of the computing device, the liquid cooling unit, the thermal buffer unit, and the power generation unit, and to control at least one of the liquid cooling unit, the thermal buffer unit, and the power generation unit based on the control command.

2. The computing center heat processing system according to claim 1, characterized in that, Also includes: An auxiliary heat dissipation unit is used to receive the liquid cooling medium when the power generation unit is unable to handle the heat carried by the liquid cooling medium, thereby achieving heat dissipation for the computing center.

3. The computing center heat processing system according to claim 2, characterized in that, The thermal buffer unit includes: a thermally stabilizing buffer tank, a bypass branch, and a heat exchanger assembly, wherein... The thermally stabilizing buffer tank is used to absorb the heat carried by the liquid cooling working fluid or to release heat to the liquid cooling working fluid. The bypass branch is used to introduce the liquid cooling medium into the auxiliary heat dissipation unit; The heat exchanger assembly is used to transfer the heat carried by the liquid cooling working fluid to the ORC evaporator or ORC working fluid of the power generation unit.

4. The computing center heat processing system according to claim 3, characterized in that, The thermal buffer unit also includes a multi-port proportional regulating valve group for regulating the flow ratio of the liquid cooling working fluid into the thermal pressure stabilizing buffer tank, the power generation unit, and the bypass branch.

5. The computing center heat treatment system according to claim 4, characterized in that, The multi-way proportional control valve assembly includes a first three-way proportional valve and a second bypass valve, wherein... The first three-way proportional valve is used to adjust the flow rate ratio of the liquid cooling working fluid entering at least one of the power generation unit and the thermal pressure stabilizing buffer tank; The second bypass valve is used to introduce the liquid cooling medium into the auxiliary heat dissipation unit.

6. The computing center heat processing system according to claim 5, characterized in that, The control layer includes a safety control layer and an energy efficiency optimization layer, with the safety control layer having a higher priority than the energy efficiency optimization layer. The control commands include a first command and a second command, wherein... The security control layer is configured to generate a first instruction when at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold, or when the temperature rise rate of at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate, and control the first three-way proportional valve and / or the second bypass valve to perform a target action based on the first instruction, wherein the target action includes at least one of adjusting the flow ratio of the first three-way proportional valve, closing the first three-way proportional valve, and opening the second bypass valve; The energy efficiency optimization layer is used to calculate the target operating parameters of at least one of the liquid cooling unit, the thermal buffer unit, and the power generation unit based on the target optimization function, the state data, and the constraints, generate the second instruction based on the target operating parameters, and control at least one of the liquid cooling unit, the thermal buffer unit, and the power generation unit based on the second instruction.

7. The computing center heat treatment system according to claim 6, characterized in that, The security control layer is further used for: If at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset threshold and less than a second preset threshold, or if the temperature rise rate of at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than a first preset rate and less than a second preset rate, then the first three-way proportional valve is adjusted based on the first instruction to reduce the flow rate of the liquid cooling medium into the power generation unit. If at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset threshold, or if the temperature rise rate of at least one of the temperature of the computing device and the outlet temperature of the liquid cooling medium is greater than or equal to the second preset rate, then the first three-way proportional valve is closed and the second bypass valve is opened based on the first instruction.

8. The computing center heat processing system according to claim 1, characterized in that, The controller also includes a state machine, which determines the corresponding operating mode based on the operating conditions of the system, and controls the working state of the control layer based on the operating mode.

9. The computing center heat treatment system according to claim 6, characterized in that, The objective optimization function is: max J = P_ORC - (P_pumps + P_aux) - λ(T_target - T_out) 2 ; Where J is the net revenue objective function of the computing center heat treatment system, P_ORC is the generator output power in the power generation unit; P_pumps is the power consumption of the liquid cooling main circulation pump and working fluid pump in the liquid cooling unit; and P_aux is the power consumption of the auxiliary heat dissipation equipment in the auxiliary heat dissipation unit. T_target is the penalty weighting coefficient; T_target is the desired liquid cooling fluid outlet temperature; T_out is the liquid cooling fluid outlet temperature.

10. A control method for a heat treatment system in a computing center, characterized in that, The method is used to control the heat treatment system of the computing center as described in any one of claims 1-9, wherein the method includes the following steps: Acquire status data of at least one of the computing power equipment, liquid cooling unit, thermal buffer unit, and power generation unit in the computing power center; Control commands are generated based on the status data, and at least one of the liquid cooling unit, the thermal buffer unit, and the power generation unit is controlled based on the control commands.

Images