The power density of electronic components grows continuously, and the subsequent heat accumulation and temperature increase inevitably affect electronic equipment’s stability, reliability and service life. Therefore, achieving efficient cooling in limited space has become a key problem in updating electronic devices with high performance and high integration. Two-phase immersion is a novel cooling method. The computational fluid dynamics (CFD) method is used to investigate the cooling performance of two-phase immersion cooling on high-power electronics. The two-dimensional CFD model is utilized by the volume of fluid (VOF) method and Reynolds Stress Model. Lee’s model was employed to calculate the phase change rate. The heat transfer coefficient along the heated walls and the shear-lift force on bubbles are calculated. The simulation data are verified with the literature results. The cooling performance of different coolants has been studied. The results indicate that the boiling heat transfer coefficient can be enhanced by using a low boiling point coolant. The methanol is used as the cooling medium for further research. In addition, the mass flow rate and inlet temperature are investigated to assess the thermal performance of two-phase immersion cooling. The average temperature of the high-power electronics is 80°C, and the temperature difference can be constrained to 8°C. Meanwhile, the convective heat transfer coefficient reaches 2740 W/(m^{2}·°C) when the inlet temperature is 50°C, and the mass flow rate is 0.3 kg/s. In conclusion, the results demonstrated that two-phase immersion cooling has provided an effective method for the thermal management of high-power electronics.

At present, power electronics play an irreplaceable role in a vast array of applications-from computers to data centers, battery chargers to new energy vehicles, and even aerospace industries. However, with the development of electronic equipment in the direction of integration, miniaturization and modularization, the problem of heat dissipation has more and more influence on its performance, reliability and working life. If heat dissipation is not good, high temperature will be generated in electronic devices, affecting the performance and even leading to device failure. For every 2°C increase above the maximum operating temperature of an electronic component, the failure frequency increases by 10% [

Various cooling methods, such as jet-impingement [

The immersion cooling techniques include two kinds: single-phase cooling and two-phase cooling. Implementing two-phase immersion cooling systems enables greater heat transfer efficiency through the boiling process compared to single-phase immersion cooling, allowing for greater power densities with two-phase immersion cooling. Furthermore, the cooling infrastructure required to support two-phase immersion cooling is typically less complex, as additional adiabatic cooling beyond a dry cooler is not necessary. During the two-phase cooling process, the coolant boils and thus exists in both a liquid and gas phase. Wang et al. [

Heat transfer enhancement of two-phase liquid immersion cooling through surface treatment has been studied. El-Genk et al. [

From the above analysis, immersion cooling, especially with phase change, represents a paradigm shift in high-power electronics cooling. The main object of this paper is to study the cooling effect of two-phase immersion on high-power electronics. On the basis of CFD method, the heat and mass transfer process is simulated. Lee model is selected as mass transfer mechanism. To ensure the safety operation of electronic devices, the coolant should be dielectric firstly. Considered the conditions for working fluid, three different liquids (deionized water, ethanol and methanol) were chosen as the coolant. The flow boiling behavior and heat transfer efficiency is evaluated for different coolants. The effects of different operating conditions on the heat transfer coefficient are carried out.

In order to study the thermal performance of two-phase immersion cooling on high-power electronics, a simplified two-dimensional model is developed, as shown in

Materials | Density (kg/m^{3}) |
C_{p} |
Thermal conductivity |
Viscosity |
---|---|---|---|---|

Methanol (liquid) | 785 | 2534 | 0.202 | 0.0005495 |

Methanol (vapor) | 1.43 | 1820 | 0.0163 | 1.35 × 10^{−5} |

HPE | 5370 | 325 | 120 | – |

The grid is generated by ICEM CFD, and quadrilateral mesh is used, as shown in

For reasonable simplification of the numerical calculation, several assumptions used in this study are as follows [

(1) The liquid and vapor phases are immiscible.

(2) The radiation heat transfer process is not considered.

(3) The high-power electronics are assumed to be a homogeneous and continuous medium.

(4) Physical properties of coolant variations with temperature and pressure are not considered because these have little change over the operating conditions.

(5) It is assumed that the heated surface is smooth, and cavity effects are neglected for nucleation.

In multiphase flows, finite volume numerical solutions are more challenging than for single-phase flows. In order to determine appropriate numerical methods, interface tracking, associated with mass transfer rate and surface tension at the interface by curvature and surface area, is the critical point and needs much attention [

The average properties of two-phase flow at liquid and vapor interface are expressed as follows:

where _{p} and

The Navier-Stokes equations for flow with phase change can be written as,

Continuity equation,

Momentum equation,

Continuum Surface Force (CSF) is used to model surface tension forces with wall adhesion. is calculated as follows [

where

Energy equation,

In the energy equation, a source term _{h} accounts for the extra heat transfer involved in boiling and condensation by multiplying the rate of mass transfer and latent heat, which is given by,

Mass transfer across liquid-vapor interfaces must be accurately modeled in phase change simulations. The phase change and mass transfer equation proposed by Lee is derived from the Hertz-Knudsen equation. It has the advantages of a simple format, easy calculation and high reliability, which has been widely used. The mass transfer rates per unit volume in Lee model are calculated as,

For evaporation,

For condensation,

However, the evaporation and condensation coefficient are difficult to determine in Lee model, which is generally grounded on the researcher’s experience or experimental data and lacks theoretical foundations. In this study, the values are referenced in the previous flow boiling studies [

The Reynolds stress model (RSM) retains the Reynolds stress equation, which is able to better reflect the characteristics of turbulent flow. The transport equations are shown below:

Moreover, the RSM has a clear advantage that automatically considers the buoyancy effect and near-wall effect. Mimouni et al. [

At the initial stage, the overall temperature is set as inlet temperature. The liquid fills the entire fluid domain at the beginning. The overall temperature at the beginning is equal to the inlet temperature. Non-slip boundary condition is applied at solid walls. The influence of wall adhesion was considered by specifying the three-phase contact angle. Rather than imposing the boundary condition at the wall itself, the contact angle at which the fluid is in contact with the wall is used to adjust the surface normal in cells near the wall. The main boundary conditions are listed in ^{−4} for the residuals of the continuity equation and velocity components and 10^{−6} for the residuals of the energy are assumed. The time step for the simulation is set to 5 × 10^{−5} s.

Position | Boundary conditions | Explanation |
---|---|---|

Inlet | Mass flow inlet | The liquid fraction of inlet is 1 |

Outlet | Pressure outlet | The reverse temperature is equal to inlet temperature |

Walls | Convection boundary condition | ^{2}·°C), |

HPE | Heat source | ^{3} |

Operating conditions | Values |
---|---|

Inlet mass flow rate (kg/s) | 0.2, 0.3, 0.4, 0.5 |

Inlet temperature (°C) | 40, 45, 50, 55 |

Four mesh amounts of 15571, 23465, 32951 and 47928 were selected for comparison. According to the simulation results, the average wall temperatures of HPE were 81.63°C, 80.17°C, 79.31°C and 79.74°C, respectively. As can be seen, the wall temperature almost leveled off after the grid number of 32951. Consequently, the grid number of 32951 Quad cells is selected for the CFD simulation.

The local heat transfer coefficient on the surface of HPE is achieved from the following expression,

where _{w,iy} is the surface temperature of HPE (_{f,y} is the bulk fluid temperature, which is defined as the sectional mean temperature at a different location, _{i,y} is the local convection coefficient, and _{w,iy} and _{f,y} are obtained from simulation results. Further, the average heat transfer coefficient

By consulting a great deal of literature, there is a lack of experimental research on the two-phase immersion cooling methods on high-power electronics. A numerical model based on the mentioned mathematical model, which used the two-phase immersion technique in concentrating photovoltaic cooling, is established according to the literature [

The cooling performance of deionized water, ethanol and methanol has been compared with corresponding operation conditions _{in} = 50°C and

The calculation and analysis showed that the ^{2}·°C), respectively. It can be concluded that compared with single-phase immersion cooling, two-phase immersion cooling is a way to improve the heat transfer efficiency effectively.

The inlet subcooling of the coolant has an important influence on the flow boiling process. _{in} increases, the average temperature of the HPE also increases gradually. At _{in} = 40°C, 45°C, 50°C and 55°C, _{HPE} are around 73.9°C, 76.6°C, 79.9°C and 85°C, respectively. However, as long as the _{HPE} does not exceed the temperature limit, the high-power electronics can be operated safely, and there is no need to blindly pursue a low cooling temperature. Meanwhile, higher _{in} is able to reduce energy consumption for secondary cooling of the coolant. _{in}. Higher inlet temperature enhances boiling heat transfer efficiency, and the increasing rate _{in}. The reason is that the subcooling degree has a significant effect on subcooled boiling heat transfer. The decreasing subcooling degree can reduce the difference in heat transfer temperature. Thus, the bubbles generated on the heated surface can grow and coalesce. The heat transfer process will be enhanced, which lead to an increase in the heat transfer coefficient. When the coolant temperature around the HPE is s heated to the saturated temperature, the increasing trend of

After a steady state has been achieved, a volume of fraction for vapor between 0 and 1 is conducted to capture vapor-liquid mass transfer features around vapor bubbles and heated walls. As shown in _{in} at steady state (

The thermal stress is formed in the operating process because of temperature differences, which may lead to HPE’s failure. Hence, the temperature uniformity of HPE is one of the main indicators in measuring the cooling performance of two-phase immersion cooling method. The temperature distribution along the HPE’s left surface from the bottom to the top for different _{in} at _{in}. As _{in} is 40°C, 45°C, 50°C and 55°C, the maximum surface temperature difference is 5.8°C, 6.9°C, 7.1°C and 8.9°C, respectively.

In the process of flow boiling, the mass flow rate of coolant has an important influence on the flow boiling state. _{in} = 50°C. The temperature of HPE is basically in a steady state with minor fluctuations after 15 s. In the meantime, the rising mass flow rate leads to the _{HPE} showing a downward trend while the magnitude of the increase descends for larger _{f} changes small.

Bubble behaviors, such as the formation of bubbles, departure, sliding, and coalescence, are considered critical heat transfer mechanisms of flow boiling on the HPE’s surface.

The temperature distribution of the HPE’s left surface for different mass flow rate at _{in} = 50°C is presented in

The main objective of this paper is to study the application of the two-phase immersion cooling method on the thermal management of high-power electronics. Detailed 2D predictions of interfacial behavior and heat transfer characteristics of HPE have been achieved through CFD simulation. The effects of different cooling medium are investigated. In addition, the influence of different operating conditions has been discussed in detail. The research findings can be used to guide the optimal operating conditions design of two-phase immersion cooling system. More concretely, we get the following main conclusions:

1) The VOF model combined with Lee model and Reynolds Stress Model is utilized in the CFD simulation. The reliability of the CFD model has been verified with the experimental results in the literature, which proves that the equation model is correctly selected;

2) The cooling performance of different coolants, such as deionized water, ethanol and methanol has been studied. For deionized water, the cooling process has no change in phase. Meanwhile, for ethanol and methanol, the phase change occurs under corresponding operating conditions. The calculated heat transfer coefficient indicates that low-boiling coolant can control the HPE at a reasonable temperature and improve heat transfer efficiency;

3) The influences of different inlet temperatures on the cooling performance are analyzed. The temperature of HPE declines and the _{HPE} stays below 80°C, when _{in} is less than 50°C at

4) The effects of two-phase immersion cooling at different mass flow rates have also been studied. The rising mass flow rate leads to the _{HPE} showing a downward trend while the magnitude of the increase descends for a larger mass flow rate. The temperature difference is around 7°C, when m ranges from 0.3 to 0.5 kg/s.

As mentioned, the performance of the two-phase immersion cooling systems can be better understood from an academic viewpoint, and the results can be useful for further academic study and potential industrial assessments. The field of two-phase immersion cooling has a lot of scope for future work. The impacts of sloshing and vibrations on heat transfer performance need to be explored. Also, the effects of different coolants on major components and the reliability of two-phase immersion cooling technology should be evaluated.

Time, s

Heat transfer flux, W/m^{2}

Local heat transfer coefficient, W/(m^{2}·°C)

Average heat transfer coefficient, W/(m^{2}·°C)

Temperature, °C

Temperature difference, °C

_{p}

Specific heat capacity, J/(kg·°C)

Thermal conductivity, W/(m·°C)

Density, kg/m^{3}

Mass flow rate, kg/s

Phase fraction

Viscosity, Pa·s

Surface tension, N·m

Liquid

Vapor

Bulk fluid

Length of the HPE

Wall

Inlet

Outlet

Computational fluid dynamics

High-power electronics

Volume of fluid

Reynolds stress model

The simulation work was completed in cooperation with Lanzhou University of Technology-Ansys Joint Simulation Laboratory, China.

The authors are grateful for financial support from the Key Laboratory of Multiphase Flow Reaction and Separation Engineering of Shandong Province, China (Grant No. 2021MFRSE-C01), the Natural Science Foundation of Gansu Province, China (No. 22JR5RA269) and Fujian Province Science Foundation for Youths, China (No. 2020305069).

The authors confirm contribution to the paper as follows: study conception and design: Liqun Zhou, Yongtong Li; numerical simulation: Liqun Zhou, Weilin Yang, Shi Lin; analysis and interpretation of results: Weilin Yang, Chaojie Li; draft manuscript preparation: Liqun Zhou, Weilin Yang. All authors reviewed the results and approved the final version of the manuscript.

Data is available on request from the authors.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.