Sort:
Open Access Issue
Experimental test and analysis on a liquid cold plate heat exchanger used for high-computing-power data centers
Experimental Technology and Management 2026, 43(6): 241-247
Published: 20 June 2026
Abstract PDF (1.7 MB) Collect
Downloads:0
Objective

Conventional air cooling in data centers cannot rapidly dissipate high heat and struggles to control the temperature rise of IT equipment or high-density chips, leading to severe temperature non-uniformity and increased downtime risk in server racks. Owing to its high cooling efficiency and low operating cost, the liquid cold plate heat exchanger (LCPHE) has become the preferred solution for electronic component cooling in high-computing-power data centers.

Methods

In this study, a microchannel LCPHE with a heat dissipation capacity of 350 W was designed and manufactured to meet the cooling requirements of a data center. An experimental setup was built to test the heat transfer and fluid flow performance of the LCPHE. The test system primarily consists of two liquid cold plates, a constant temperature water tank, an electrical heating module, a power module, connecting pipes, and a data monitoring and acquisition module. Deionized water was selected as the working medium. Temperature and pressure sensors were installed at the inlet and outlet of the liquid cold plates, and a mass flow meter was positioned at the outlet of the circulation pump. By adjusting the opening of the control valve, the effect of the circulating flow rate on the LCPHE performance was measured. Meanwhile, a numerical model was developed to simulate the thermal–flow coupling performance of the tested liquid cold plate. The effects of three commonly used coolants, namely deionized water, propylene glycol solution, and ethylene glycol solution, on the heat transfer performance and flow pressure drop of the LCPHE were compared, and the optimal coolant for the liquid cooling system was identified.

Results

As the coolant circulation flow rate increased from 0.8 L/min to 1.4 L/min, the fluid temperature at the outlet of the cold plate decreased from 50.2 ℃ to 46.6 ℃, and the pressure drop across the cold plate increased from 17.72 kPa to 48.88 kPa. After thorough comparison and analysis, the standard kε turbulence model was found to exhibit high prediction accuracy and was deemed acceptable for performance simulation of the LCPHE. With increasing solution concentration, both the heat transfer coefficient and fluid flow resistance increased. In addition, the comprehensive performance index of the LCPHE also increased with solution concentration. From multiple perspectives, ethylene glycol solution proved to be the optimal coolant choice.

Conclusions

Given that limited research and teaching experiments on liquid cold plates have been conducted to date, the present study establishes a comprehensive liquid cold plate test bench for teaching and research through systematic design, system debugging, variable operating condition testing, and numerical simulation validation. An in-depth investigation of the operational performance of the liquid cooling system in data centers was conducted, and the potential for system improvement and application was evaluated. The outcomes not only provide technical support for the low-carbon, efficient operation of green data centers but also contribute to advancing scientific research and teaching practice reform in the field of liquid cooling for data centers.

Issue
System design and experimental study of single-phase immersion liquid cooling for high-computing-power data center
Experimental Technology and Management 2025, 42(4): 37-41
Published: 20 April 2025
Abstract PDF (732.7 KB) Collect
Downloads:49
[Objective]

With the rapidly increasing demand for high computing power and intelligent computing from artificial intelligence, big data, and cloud platforms, the cooling capacity of traditional air cooling technology for IT cabinets has reached its limit. The renewal and replacement of diverse computing infrastructure, such as high density, high computing power, and high thermal output, have effectively promoted the development of liquid cooling technology. Liquid cooling technology effectively improves the traditional form of air cooling and can meet the precise cooling needs of high-density cabinets and chip levels. Depth research and development of liquid cooling technology are crucial for reducing data center energy consumption and improving energy utilization efficiency. Thus far, no teaching or research experiments on single-phase immersion liquid cooling have been conducted in Chinese universities.

[Methods]

A comprehensive single-phase immersion liquid cooling test bench for teaching and research was constructed in this study. The detailed thermal performance was analyzed, and the associated calculation was carried out to select the circulation pump, plate heat exchanger, cooling tower, heating device, and other heat transfer equipment to establish the single-phase immersion liquid cooling system. The coolant was comparatively selected from fluorinated liquid, deionized water, and mineral oil based on the fluid thermal physical properties, fluid motion characteristics, heat exchange performance, and operational stability. Three circulation mass flow rates were designed to compare the performance difference of the proposed single-phase immersion liquid cooling unit. The circulation mass flow rate, fluid pressure, liquid height, and power consumption were monitored and utilized to control and reflect the detailed operation of the single-phase immersion liquid cooling system. After reasonable design, comparative selection, and device connection, both the primary side cooling water cycle and the secondary side coolant cycle were assembled. After the gas injection pressure and water injection tests, the airtightness and compressive strength of the established experimental system were confirmed. Meanwhile, system debugging was conducted to test the accuracy and sensitivity of the monitoring system.

[Results]

With the circulation mass flow rate of cooling water varying from 4.4 m3/h to 6.4 m3/h, the temperatures of the inlet and outlet of the coolant and cooling water have been reduced. For pressure loss, the pressure difference on the coolant side was nearly uninfluenced, whereas the maximum pressure drop was approximately 56.5 kPa on the cooling water side, with a mass flow rate of 6.4 m3/h. The power usage efficiency of the proposed single-phase immersion cooling system was varied in the range of 1.08–1.09. The coefficient of performance of the cooling system decreased from 6.4 to 5.68, with the cooling water mass flow rate increasing from 4.4 m3/h to 6.4 m3/h. The efficiency of the circulation was evaluated and determined to be approximately 20% for three operation cases.

[Conclusions]

Generally, the performance of the single-phase liquid cooling and heat dissipation system utilized for high-computing-power data centers is thoroughly explored and analyzed, and its improvement space and application potential are evaluated. On the one hand, the present work can provide technical support for the low-carbon and efficient operation of green data centers. On the other hand, the present work can further improve the teaching practice reform of liquid cooling research-oriented experiments in data centers.

Research Article Issue
Influence factors of the numerical model build-up on fluid sloshing
Experimental and Computational Multiphase Flow 2022, 4(4): 435-444
Published: 13 January 2021
Abstract Collect

It is important to predict fluid sloshing in cryogenic fuel storage tanks with an accurate numerical model. In the present paper, a computational fluid dynamics (CFD) model was established to simulate fluid sloshing. The fluid sloshing experiments in an apparent vessel were adopted to be the benchmark and used to validate the numerical model. As most of investigations were conducted with some certain model settings, few are involved on the influence factor on numerical model build-up. Based on the selected sloshing experiments, effects of the numerical time step and phase change factor on fluid sloshing were investigated. The results showed that the time steps of 0.001 and 0.002 s and the phase change factor of 0.1 s-1 could meet the requirement of fluid sloshing prediction. With some valuable conclusions being obtained, the present study may supply some effective references for numerical model building on fluid sloshing.

Total 3