A pressure self-adaptive immersion two-phase liquid cooling system and a boiling regulation method thereof

Abstract

The application discloses a pressure self-adaptive immersion two-phase liquid cooling system and a regulation method thereof, relates to the technical field of server heat dissipation, and is used for data center heat dissipation. The system comprises a sealed cavity, a pressure adjusting mechanism, an ultrasonic wave generating device, a temperature monitoring module, a power consumption obtaining module, a pressure monitoring module and a controller. The controller calculates a target boiling temperature according to chip power consumption, solves a target pressure according to a saturated steam pressure-temperature mapping, and makes the pressure approach the target value through closed-loop adjustment. Meanwhile, the ultrasonic wave power is adjusted according to the comparison result of the chip temperature and the safety threshold value to promote bubble detachment. The method comprises data acquisition, target boiling point calculation, pressure solving, closed-loop pressure regulation and ultrasonic wave control. The application realizes that the boiling point is self-adaptive to power consumption, avoids waste of low-load energy consumption and heat dissipation shortage of high-load, strengthens heat exchange by breaking the gas film layer through ultrasonic waves, suppresses overshoot oscillation based on the gas-liquid two-phase volume model feedforward correction, and is suitable for efficient and stable operation under all working conditions.

Description

Technical Field

[0001] This invention relates to the field of server heat dissipation technology, and in particular to an immersion liquid cooling technology for high heat flux density chips in data centers, specifically a pressure-adaptive immersion two-phase liquid cooling system and its boiling control method. Background Technology

[0002] With the rapid development of fields such as artificial intelligence and high-performance computing, the power consumption and heat flux density of data center chips continue to rise. Traditional air cooling and single-phase liquid cooling technologies are gradually approaching their physical limits in terms of heat dissipation efficiency and pump power consumption. Immersion two-phase liquid cooling technology, as a revolutionary heat dissipation solution, directly immerses servers in a low-boiling-point, high-dielectric-strength coolant (such as electronic fluorinated liquid). It utilizes the large amount of latent heat of vaporization absorbed by the coolant during phase change (boiling) on ​​the surface of heat-generating components, achieving extremely high heat dissipation efficiency and extremely low overheating. It is considered one of the key paths to solving the heat dissipation challenges of next-generation hyperscale data centers.

[0003] However, in practical engineering applications, existing immersion two-phase liquid cooling systems still face the following significant technical defects and implementation difficulties: Firstly, the boiling temperature and chip power consumption cannot be dynamically matched, resulting in low operating efficiency. The saturation temperature (boiling point) of the coolant is positively correlated with the system pressure. Existing technologies typically employ a fixed-pressure constant-value operation mode, maintaining the cavity pressure at a constant target value throughout the entire operating cycle. However, the load of data center servers exhibits significant tidal and fluctuating characteristics. When the chip is under low load, the heat generation decreases. If high pressure is maintained to keep the coolant at a high boiling point, the boiling intensity will be extremely weak, or even unable to sustain nucleus boiling. In this case, the system mainly relies on single-phase convective heat transfer, resulting in a significant decrease in heat dissipation efficiency and unnecessary energy waste in the coolant circulation pump and condenser unit. Conversely, when the chip experiences a sudden high load peak, the heat flux density rises sharply. Under fixed pressure, the coolant boiling point is constant, and there is an upper limit to the heat that can be carried away by boiling per unit time. This easily leads to "insufficient boiling" or "premature arrival of the critical heat flux density (CHF)," causing a sharp rise in the temperature of the heating surface, triggering chip frequency reduction or even damage.

[0004] Secondly, boiling bubbles adhere to the heating surface, forming a "gas film layer," which severely degrades the heat transfer coefficient. During immersion boiling, a large number of bubbles nucleate, grow, and eventually detach at the heating micro-protrusions. However, when the heat flux density is high or the coolant flow disturbance is insufficient, the detachment frequency of bubbles cannot keep up with the generation rate. A large number of bubbles will accumulate and merge on the chip surface, forming a continuous or semi-continuous vapor insulation layer (i.e., the "gas film layer"). This gas film layer has an extremely low thermal conductivity, hindering direct contact between the liquid and the heating surface, resulting in a sharp increase in heat transfer resistance. Existing technologies mainly rely on natural buoyancy or forced circulation to flush away the bubbles, but under high heat flux density and complex chip layouts, the passive bubble desorption efficiency is extremely limited, severely restricting the overall heat dissipation capacity and stability of the system.

[0005] Third, the pressure regulation response is lag-induced, the process oscillates violently, and there is a lack of accurate gas-liquid two-phase volume change model support. To adjust the system boiling point, existing solutions typically use a vacuum pump to extract non-condensable gases or fill with compressed air to change the gas phase pressure of the sealed cavity. However, the interior of a two-phase immersion liquid-cooled cavity is a typical gas-liquid-solid three-phase coupled thermodynamic system. Its pressure changes are not only affected by the gas mass input / output, but also by the coupling effects of multiple factors such as coolant temperature, vapor partial pressure changes caused by the intensity of boiling, and thermal expansion of liquid volume. Existing control strategies lack accurate mathematical modeling of the dynamic evolution of gas phase volume, liquid phase volume, and vapor partial pressure within the cavity, making it difficult to tune the PID control parameters of pressure regulation actuators (such as variable frequency vacuum pumps and proportional control valves). In actual operation, this often results in large pressure overshoot, slow convergence speed, or even constant-amplitude oscillations. This not only fails to respond quickly to transient changes in chip power consumption, but may also cause the sealed cavity to be subjected to alternating pressure stress, increasing the risk of leakage.

[0006] In summary, existing immersion two-phase liquid cooling systems have significant shortcomings in dynamic power consumption adaptation, enhanced boiling heat transfer, and precise pressure control. Therefore, there is an urgent need in this field for a technical solution that can adaptively and precisely control the operating pressure of the two-phase liquid cooling system according to the real-time power consumption requirements of the chip, and actively intervene in bubble desorption behavior, thereby achieving efficient and stable heat dissipation across the entire operating range.

[0007] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0008] The purpose of this invention is to provide a technical solution that can adaptively and precisely adjust the operating pressure of a two-phase liquid cooling system according to the real-time power consumption requirements of the chip, and actively intervene in the bubble desorption behavior, thereby achieving efficient and stable heat dissipation across the entire operating range.

[0009] To achieve the above objectives, the present invention provides the following solution: A pressure-adaptive immersion two-phase liquid cooling system includes: A sealed cavity containing a low-boiling-point coolant, in which the server is immersed; A pressure regulating mechanism, connected to the sealed cavity, is used to regulate the gas phase pressure inside the sealed cavity; An ultrasonic generator is disposed inside the sealed cavity and is used to apply ultrasonic waves to the coolant on the surface of the server to promote the desorption of boiling bubbles. A temperature monitoring module, installed on the server, is used to acquire the real-time temperature of the server's chips. T c ; The power consumption acquisition module is used to acquire the power consumption data of the chip in real time. P ; A pressure monitoring module is installed inside the sealed cavity to acquire the actual pressure value inside the sealed cavity in real time. P act ; The controller is communicatively connected to the pressure regulating mechanism, the ultrasonic generator, the temperature monitoring module, the power consumption acquisition module, and the pressure monitoring module, respectively. The controller is configured as follows: The real-time power consumption data of the chip obtained by the power consumption acquisition module. P According to the formula

[0010] Calculate the target boiling temperature ,in T max This is the preset maximum allowable junction temperature of the chip. The equivalent thermal resistance between the chip junction and the coolant. A The effective heat exchange area of ​​the chip; According to the target boiling temperature The mapping relationship between the pressure and temperature characteristics of the coolant is used to calculate the target pressure value required inside the sealed cavity. p tar The mapping relationship satisfies

[0011] in a , b , c These are constants related to the physical properties of the coolant; According to the target pressure value p tar The actual pressure value fed back by the pressure monitoring module p act The deviation between the actual pressure value and the target pressure value is sent to the pressure regulating mechanism as a first control signal to make the actual pressure value approach the target pressure value. Furthermore, the controller is also configured to monitor the chip temperature fed back by the temperature monitoring module. T c With preset safe temperature threshold T safe Based on the comparison result, a second control signal is sent to the ultrasonic generator to start or adjust the output power of the ultrasonic generator. Q us . A boiling control method using the above-mentioned pressure-adaptive immersion two-phase liquid cooling system includes the following steps: Real-time acquisition of power consumption data of server chips immersed in low-boiling-point coolant. P and temperature data T c and the actual pressure value inside the sealed cavity. p act ; According to the real-time power consumption data P According to the formula:

[0012] Calculate the target boiling temperature that matches the real-time power consumption data. ,in T max This is the preset maximum allowable junction temperature of the chip. The equivalent thermal resistance between the chip junction and the coolant. A The effective heat exchange area of ​​the chip; based on the target boiling temperature

[0013] The mapping relationship between the pressure and temperature characteristics of the coolant is used to calculate the target pressure value required inside the sealed cavity. p tar The mapping relationship satisfies

[0014] in, a , b , c These are constants related to the physical properties of the coolant; According to the target pressure value p tarCompared with the actual pressure value p act The deviation between the actual pressure and the target pressure is adjusted in a closed loop to make the actual pressure value approach the target pressure value. Furthermore, based on the chip temperature T c With preset safe temperature threshold T safe Based on the comparison results, the output power of the ultrasonic generator located inside the sealed cavity is activated or adjusted. Q us Ultrasonic waves are applied to the coolant on the server surface to promote the desorption of boiling bubbles from the server surface.

[0015] Optionally, the equivalent heat transfer resistance It is composed of a chip package thermal resistance, an interface material thermal resistance, and a convective heat transfer thermal resistance connected in series, and the controller is based on the real-time power consumption of the chip. P and measured chip temperature T c Dynamically update the equivalent heat transfer resistance. The value is obtained and updated in the following way:

[0016] in For the current actual pressure p act The corresponding coolant saturation temperature.

[0017] Optionally, when the controller performs closed-loop regulation of the gas phase pressure inside the sealed cavity, it employs an incremental PID control algorithm, and the first control signal is the control quantity increment Δ of the pressure regulating mechanism. u ( k The calculation formula is as follows:

[0018] in k At the current sampling time, The pressure deviation at the current moment, K p , K i , K d These are the proportional, integral, and derivative coefficients, respectively. The controller also considers the real-time power consumption change rate of the chip. For the integral coefficients K i Perform adaptive adjustment, and adjust the relationship as follows:

[0019] in K i0 The baseline value for the integral coefficient, α This is the preset adaptive gain factor.

[0020] Optionally, the pressure regulating mechanism includes a vacuum pump, an air compressor, a proportional regulating valve, and a gas storage tank. The vacuum pump and the air compressor are respectively connected to the sealed cavity through the proportional regulating valve. The controller adjusts the opening of the proportional regulating valve and controls the start and stop of the vacuum pump and the air compressor to achieve bidirectional regulation of the gas phase pressure inside the sealed cavity.

[0021] Optionally, the ultrasonic generator includes multiple ultrasonic transducers distributed at different positions inside the sealed cavity. The controller independently controls the start-up, shutdown, and output power level of each ultrasonic transducer based on the layout of the server's heating elements and real-time heat distribution information.

[0022] Optionally, the controller adjusts the temperature based on the chip. T c With preset safe temperature threshold T safe The comparison results determine the output power of the ultrasonic generator. Q us The method for determining it is as follows: when T c ≤ T safe hour, Q us =0 or maintain minimum standby power; when T c > T safe At that time, according to the formula

[0023] Calculate the output power, where Q us,min The preset minimum effective ultrasonic power, β γ is the power regulation coefficient, γ is the exponential factor, and the output power is... Q us The power shall not exceed the maximum permissible power of the ultrasonic generator. Q us,max .

[0024] Optionally, the controller is further configured to: when the real-time power consumption data of the chip undergoes a step change, adjust the power consumption based on the gas volume inside the sealed cavity. Vg Liquid phase volume V l and the current temperature of the coolant T l Estimate the steady-state pressure inside the sealed cavity under thermal equilibrium conditions. p ss The estimation formula is:

[0025] Where △ P chip This represents the step change in the power consumption of the chip. This is the derivative of the coolant saturation pressure with respect to temperature. This is an empirical correction factor; the controller uses the steady-state pressure value. p ss As the target pressure value p tar The correction benchmark is used to adjust the target pressure value to

[0026] Where λ is a relaxation factor between 0 and 1 to suppress overshoot and oscillation of the pressure regulation mechanism.

[0027] Optionally, the sealed cavity is further provided with a condensation reflux device, which is connected to the gas phase region of the sealed cavity. This device is used to condense the vapor generated by the evaporation of the coolant into a liquid state and reflux it back into the sealed cavity. The flow rate of the cooling medium in the condensation reflux device is determined by the controller based on the actual pressure value inside the sealed cavity. p act and the chip's real-time power consumption data P Adjustments are made, and the adjustment relationship is as follows:

[0028] in The mass flow rate of the cooling medium. Based on basic traffic, This is the flow regulation coefficient. P ref Reference power consumption value ，h fg The latent heat of vaporization of the coolant.

[0029] Optionally, the controller is further configured to: before performing the closed-loop adjustment, generate a power consumption prediction value using a first-order exponential smoothing method based on the historical sequence of the chip's real-time power consumption data. The prediction formula is:

[0030] in k At the current sampling time, For smoothing coefficients, P ( k () represents the actual power consumption at the current moment. The predicted power consumption at the current moment; based on the predicted power consumption value. Pre-calculate the target pressure value for the next moment. p tar ( k +1), and at the actual pressure value p act Before deviating from the current target value, a feedforward control signal is sent to the pressure regulating mechanism to compensate for the mechanical lag of the pressure regulating mechanism.

[0031] Compared with the prior art, the present invention has the following beneficial effects: This invention acquires chip power consumption data in real time through a controller, constructing a complete closed-loop control link of "chip power consumption prediction → target boiling temperature calculation → target pressure calculation → active pressure regulation." This achieves real-time dynamic matching between boiling temperature and chip power consumption, effectively resolving the contradiction between energy waste during low-load over-boiling and insufficient boiling under high-load conditions in traditional fixed-pressure operation. Energy efficiency is significantly improved across all operating conditions. Simultaneously, this invention incorporates an ultrasonic generator inside the sealed cavity. The controller dynamically adjusts the ultrasonic output power based on the comparison between the chip temperature and a preset safety threshold, utilizing the ultrasonic cavitation effect and micro-... The jet actively disrupts the bubble and gas film layers attached to the heating surface, significantly enhancing the boiling heat transfer coefficient and overcoming the technical challenge of low efficiency in passive bubble desorption. Furthermore, this invention accurately models the volume changes of the gas-liquid two phases within the sealed cavity, estimating the steady-state pressure value of thermal equilibrium when chip power consumption undergoes a step change. A relaxation factor is used to feedforward correct the target pressure. Combined with an incremental PID algorithm, adaptive adjustment of the integral coefficient based on the rate of change of power consumption, and feedforward control based on power consumption prediction, overshoot and oscillations during pressure regulation are effectively suppressed, significantly improving the speed and stability of pressure control. In summary, this invention achieves the dual goals of precise and controllable boiling temperature and continuous and efficient wetting of the heat exchange surface, maintaining efficient, stable, and reliable heat dissipation performance across the entire power consumption fluctuation range of the chip. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. 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.

[0033] Figure 1 This is a schematic diagram of the pressure-adaptive immersion two-phase liquid cooling system provided in an embodiment of the present invention.

[0034] Figure 2 This is a schematic diagram of the boiling control method provided in an embodiment of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] The purpose of this invention is to provide a technical solution that can adaptively and precisely adjust the operating pressure of a two-phase liquid cooling system according to the real-time power consumption requirements of the chip, and actively intervene in the bubble desorption behavior, thereby achieving efficient and stable heat dissipation across the entire operating range.

[0037] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Example 1: Specific structure and control implementation of a pressure-adaptive immersion two-phase liquid cooling system: This embodiment provides a pressure-adaptive immersion two-phase liquid cooling system suitable for server clusters in data centers with a single rack power consumption of 10kW to 50kW.

[0039] I. System Hardware Configuration: The sealed cavity is welded from 304 stainless steel, with an effective internal volume of 1.2m³. 3 The design pressure is 0.2MPa~0.8MPa (absolute pressure). The chamber has a gas phase outlet at the top and a liquid phase reflux port at the bottom. 3M coolant is selected. TM Novec TM 7100 engineering fluid has a normal boiling point of 61℃ and a latent heat of vaporization. h fg =112kJ / kg, the saturated vapor pressure-temperature relationship satisfies the following equation:

[0040] That is, the constant in the formula takes the value of a =10.52, b =2512.3, c = -48.78 (temperature unit °C).

[0041] The server motherboard is submerged in coolant, with a 5mm–10mm coolant layer remaining above the chips. The temperature monitoring module uses a surface-mount PT1000 RTD, attached to the chip surface (or embedded within the chip's heatsink), with a measurement accuracy of ±0.5℃. The power consumption acquisition module reads the server's power management bus (PMBus) and outputs the chip's power consumption in real time. P (Unit: W), sampling frequency: 10Hz. The pressure monitoring module uses an absolute pressure piezoresistive sensor (range: 0-1MPa, accuracy: 0.1%FS), installed in the gas phase region of the cavity.

[0042] The pressure regulation mechanism includes: a variable frequency vacuum pump (pumping speed 4L / s), an air compressor (discharge pressure 0.8MPa), two electric proportional control valves (Cv values ​​of 0.1 and 0.5 respectively), and a 50L gas storage tank. The vacuum pump and compressor are connected to the gas phase port of the chamber via proportional control valves. The controller controls the opening degree of the proportional valves (0-100%) and the start / stop and speed of the vacuum pump / compressor via a 4-20mA signal, achieving bidirectional and precise regulation of the chamber pressure.

[0043] The ultrasonic generator consists of six piezoelectric ultrasonic transducers (frequency 40kHz, maximum power 50W / each), evenly distributed at the bottom and side walls of the cavity, each aimed at high heat flux density areas such as the server's CPU and GPU. Each transducer is powered by an independent driver, and the controller independently controls its output power level (0-100%) via an RS-485 bus.

[0044] The condensation reflux device is a water-cooled plate heat exchanger, installed on the gas phase outlet pipe of the cavity. The cooling water flow rate is controlled by an electric regulating valve, and the maximum cooling capacity is 30kW.

[0045] II. Specific configuration and working logic of the controller: The controller uses an industrial-grade PLC (Programmable Logic Controller) or an embedded industrial computer, and has the following built-in computing modules: (1) Calculation of target boiling temperature: The controller reads the chip power consumption in real time. P ( k According to the formula:

[0046] in, T max Set to 85℃ (the chip's maximum allowable junction temperature). A For the effective heat exchange area of ​​the chip, a value of 4 × 10 is used for a typical CPU (size 20mm × 20mm). -4 m 2 Equivalent thermal resistance The initial value is 0.15 K / W (including package thermal resistance of 0.05 K / W, thermal interface material thermal resistance of 0.03 K / W, and boiling convection thermal resistance of 0.07 K / W). The controller adjusts the temperature every 10 seconds based on the measured chip temperature. T c Saturation temperature corresponding to the current pressure Dynamic updates :

[0047] The updated value is then filtered by a first-order low-pass filter before being used in subsequent calculations to suppress measurement noise.

[0048] (2) Target pressure calculation: Based on the saturated vapor pressure-temperature relationship of the coolant, Substitute:

[0049] in a =10.52, b =2512.3, c = -48.78, thus obtaining the target absolute pressure (in kPa). For example, if the target boiling temperature is 50℃, then the calculated value is... p tar ≈46.7 kPa (absolute pressure).

[0050] (3) Pressure closed-loop PID control: The controller uses an incremental PID algorithm, with the control quantity increment Δ u ( k The calculation is as follows:

[0051] in, The initial PID parameters are tuned to... Kp =2.5, Ki 0 = 0.8 Kd =0.3. The controller monitors the rate of change of power consumption in real time. When the absolute value of the rate of change is greater than 10 W / s, the integral coefficient is adaptively adjusted.

[0052] Pick α =0.02s / W, to enhance the response capability to transient power consumption. Control quantity Δ u ( k After being limited, it is mapped to the change in the opening of the proportional valve or the speed command of the vacuum pump / compressor.

[0053] (4) Steady-state pressure feedforward correction (suppressing overshoot): When a step change in power consumption Δ is detected P When the chip power suddenly increases from 200W to 400W, the controller estimates the steady-state pressure after thermal equilibrium:

[0054] The revised As the setpoint for the PID controller, it can effectively suppress overshoot from the usual 30% to within 5%.

[0055] (5) Graded control of ultrasonic power: The controller is based on the chip temperature Tc With preset safety threshold T With a safe temperature of 75℃, dynamically adjust the ultrasonic output power. Q us: when Tc At ≤75℃, Q us=0 (or maintain standby power of 5W to keep the transducer preheated). when T c For temperatures above 75℃, the following nonlinear function is used for calculation:

[0056] Pick Q us,min =20W, β =0.5W / ℃ γ , γ=1.5. When T c At 80℃, Q us ≈20+0.5×5 1.5 ≈20 + 0.5 × 11.18 ≈ 25.6 W; when T c When it rises to 90℃, Q us ≈20+0.5×15 1.5 ≈20 + 0.5 × 58.1 ≈ 49W, but not exceeding the maximum allowable power. Q us,max =50W (per transducer). The controller distributes the total power to each transducer according to the heat distribution ratio.

[0057] (6) Adjustment of condensate reflux flow rate: The controller adjusts the cooling water flow rate based on actual pressure and power consumption.

[0058] Pick =0.5kg / s (corresponding to a basic heat dissipation of 10kW). μ=0.9 (dimensionless) P ref =1000W, h fg =112kJ / kg. When the total power consumption of the chip... P At 3000W, the additional bandwidth is 0. The total flow rate is approximately 0.516 kg / s.

[0059] (7) Power consumption prediction feedforward control: The controller samples power consumption every 100ms and uses first-order exponential smoothing to predict power consumption at the next moment. Take the smoothing constant .according to Calculate in advance p tar ( k +1), and immediately sends a feedforward compensation signal (superimposed on the PID output) to the proportional valve, causing the pressure regulating mechanism to act in advance, thereby compensating for the mechanical lag of the vacuum pump and valve (approximately 200–500 ms). Actual measurements show that this feedforward can reduce the response delay from approximately 1.2 seconds to less than 0.5 seconds.

[0060] III. Typical Operating Condition Examples: Assume the server initially operates stably at power consumption. P =1000W, chip temperature T c =70℃, current pressure p act =60kPa (corresponding to a saturation temperature of approximately 56℃). At this time... T c At temperatures below 75℃, ultrasonic waves are turned off.

[0061] When the workload suddenly increases, the power consumption jumps to [a certain level] within 2 seconds. P =3000W. The controller immediately calculates the target boiling temperature:

[0062] A negative value indicates that the coolant is outside its adjustable range. The controller will then set the target boiling point to the coolant's minimum effective temperature (e.g., 10°C), corresponding to the target pressure. p tar The equation yielded approximately 6.5 kPa. Simultaneously, since the actual pressure of 60 kPa was much higher than the target, the controller activated the vacuum pump for rapid evacuation and utilized feedforward correction to avoid overshoot. The chip temperature rapidly rose to 80°C, triggering ultrasonic waves to operate at approximately 25W power to assist in bubble desorption. The entire adjustment process reduced the pressure to approximately 7 kPa within 3 seconds, and the chip temperature eventually stabilized at 82°C (below [a certain value]). T max(At 85℃), the system enters a new steady state.

[0063] Example 2: Boiling control method flow based on the same system.

[0064] This embodiment is described in detail as follows: Figure 2 The boiling control method shown is applicable to the system hardware in Example 1.

[0065] Step S1: Data Acquisition The controller synchronously acquires the chip power consumption at a sampling rate of 50Hz. P ( k Chip temperature T c ( k ), cavity pressure p act ( k The signal is then subjected to median filtering to remove spike noise.

[0066] Step S2: Calculation of target boiling temperature: Calculate according to the above formula And add safety constraints: like Then it is forced to be set to 10℃ (to avoid air being sucked in due to excessively low pressure). like If so, the temperature is forced to be 80℃ (to avoid excessive pressure in the cavity exceeding the pressure design).

[0067] Step S3: Target pressure calculation and smoothing: Will Substituting into the equation, we get p tar ( k The setpoint is then subjected to a first-order low-pass filter (filter coefficient 0.2) to prevent frequent actuator movements caused by sudden changes in the setpoint.

[0068] Step S4: Pressure Regulation Decision: Calculate the deviation .like If the condition is stable, the current actuator state is maintained; otherwise, the process switches to PID calculation.

[0069] Step S5: Incremental PID output and adaptive integral: Δ is calculated according to the incremental PID formula in Example 1. u ( k ), and at the same time according to Real-time adjustment K i The control cycle is 200ms.

[0070] Step S6: Feedforward Compensation: Every 1 second based on power consumption prediction Calculate pre-adjustment pressure p tar,ff ( k +1), and add it as a feedforward to the PID output to change the valve position in advance.

[0071] Step S7: Determine the ultrasonic power: according to T c ( k )and T safe Comparison, calculated as a piecewise function Q us ( k ), and distribute them to each transducer. Simultaneously, increase the hysteresis interval ( T c The ultrasound should only be turned off when the temperature drops below 72°C to prevent frequent start-stop cycles.

[0072] Step S8: Condensation flow rate adjustment: The cooling water flow rate is adjusted synchronously to meet the steam condensation requirements and ensure stable steam partial pressure within the cavity.

[0073] Step S9: Iterate through the loop: Repeat steps S1 to S8 to achieve adaptive control under all operating conditions.

[0074] Step S10: Anomaly Protection: like p act Exceeding 0.7 MPa or below 0.1 MPa, or T c If the temperature exceeds 90°C, the controller will immediately issue an audible and visual alarm and trigger an emergency pressure relief or shutdown protection action.

[0075] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0076] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A pressure-adaptive immersion two-phase liquid cooling system, characterized in that, include: A sealed cavity containing a low-boiling-point coolant, in which the server is immersed; A pressure regulating mechanism, connected to the sealed cavity, is used to regulate the gas phase pressure inside the sealed cavity; An ultrasonic generator is disposed inside the sealed cavity and is used to apply ultrasonic waves to the coolant on the surface of the server to promote the desorption of boiling bubbles. A temperature monitoring module is installed on the server to obtain the chip temperature of the server in real time. A power consumption acquisition module is used to acquire the power consumption data of the chip in real time. A pressure monitoring module is installed inside the sealed cavity to acquire the actual pressure value inside the sealed cavity in real time. The controller is communicatively connected to the pressure regulating mechanism, the ultrasonic generator, the temperature monitoring module, the power consumption acquisition module, and the pressure monitoring module, respectively. The controller is configured to: calculate a target boiling temperature matching the real-time power consumption data of the chip obtained by the power consumption acquisition module; calculate the target pressure value required inside the sealed cavity based on the mapping relationship between the target boiling temperature and the physical properties of the coolant; send a first control signal to the pressure regulating mechanism based on the deviation between the target pressure value and the actual pressure value fed back by the pressure monitoring module, so that the actual pressure value approaches the target pressure value; and the controller is further configured to send a second control signal to the ultrasonic generator based on the comparison result between the chip temperature fed back by the temperature monitoring module and a preset safe temperature threshold, so as to start or adjust the output power of the ultrasonic generator.

2. A boiling control method using the pressure-adaptive immersion two-phase liquid cooling system as described in claim 1, characterized in that, Includes the following steps: Real-time acquisition of power consumption and temperature data of server chips immersed in low-boiling-point coolant, as well as the actual pressure value inside the sealed cavity; Calculate the target boiling temperature that matches the real-time power consumption data based on the real-time power consumption data. Based on the mapping relationship between the target boiling temperature and the physical properties of the coolant, the target pressure value required to be achieved inside the sealed cavity is calculated. Based on the deviation between the target pressure value and the actual pressure value, the gas phase pressure inside the sealed cavity is adjusted in a closed loop so that the actual pressure value approaches the target pressure value. Furthermore, based on the comparison between the chip temperature and the preset safe temperature threshold, the output power of the ultrasonic generator located inside the sealed cavity is activated or adjusted to apply ultrasonic waves to the coolant on the server surface, promoting the desorption of boiling bubbles from the server surface.

3. The boiling control method according to claim 2, characterized in that, The step of calculating the target boiling temperature matching the real-time power consumption data specifically includes: The real-time power consumption data is input into a preset power consumption-temperature mapping model to calculate the target boiling temperature of the coolant that ensures the server chip temperature does not exceed the preset safe temperature range and the boiling heat transfer efficiency is optimal under the real-time power consumption. The power consumption-temperature mapping model at least associates the thermal resistance parameter between the saturation temperature of the coolant and the junction temperature of the chip.

4. The boiling control method according to claim 2, characterized in that, The step of calculating the target pressure value required inside the sealed cavity based on the mapping relationship between the target boiling temperature and the physical properties of the coolant specifically includes: Based on the physical property correspondence between the saturated vapor pressure and temperature of the coolant, the saturated vapor pressure value corresponding to the target boiling temperature is taken as the target pressure value. The target pressure value increases with the increase of the real-time power consumption data of the chip and decreases with the decrease of the real-time power consumption data.

5. The boiling control method according to claim 2, characterized in that, The pressure regulating mechanism includes a vacuum pump, an air compressor, a proportional regulating valve, and a gas storage tank. The vacuum pump and the air compressor are respectively connected to the sealed cavity through the proportional regulating valve. The controller adjusts the opening of the proportional regulating valve and controls the start and stop of the vacuum pump and the air compressor to achieve bidirectional regulation of the gas phase pressure inside the sealed cavity.

6. The boiling control method according to claim 2, characterized in that, The ultrasonic generator includes multiple ultrasonic transducers, which are distributed in different positions inside the sealed cavity. The controller independently controls the start-up, shutdown, and output power level of each ultrasonic transducer based on the layout of the server's heating elements and real-time heat distribution information.

7. The boiling control method according to claim 2, characterized in that, The controller is further configured to: when the chip temperature is lower than a preset first temperature threshold, control the ultrasonic generator to operate at a first power level or remain off; when the chip temperature is higher than the first temperature threshold and lower than a preset second temperature threshold, control the ultrasonic generator to operate at a second power level, the second power level being higher than the first power level; and when the chip temperature is higher than the second temperature threshold, control the ultrasonic generator to operate at a third power level, the third power level being higher than the second power level.

8. The boiling control method according to claim 2, characterized in that, The controller is further configured to: when the real-time power consumption data of the chip undergoes a step change, estimate the steady-state pressure value inside the sealed cavity under thermal equilibrium based on the gas phase volume, liquid phase volume and the current temperature of the coolant inside the sealed cavity, and use the steady-state pressure value as a correction reference for the target pressure value to suppress overshoot and oscillation of the pressure regulation mechanism.

9. The boiling control method according to claim 2, characterized in that, The sealed cavity is also equipped with a condensation reflux device, which is connected to the gas phase region of the sealed cavity. The condensation reflux device is used to condense the vapor generated by the evaporation of the coolant into a liquid state and reflux it back into the sealed cavity. The flow rate of the cooling medium in the condensation reflux device is adjusted by the controller according to the actual pressure value inside the sealed cavity and the real-time power consumption data of the chip.

10. The boiling control method according to claim 2, characterized in that, The controller is further configured to: generate a power consumption prediction value based on the changing trend of the chip's real-time power consumption data before performing the closed-loop regulation, and pre-adjust the target pressure value based on the power consumption prediction value to compensate for the mechanical lag of the pressure regulation mechanism.

Images