Simulation Study of Influencing Factors of Immersion Phase-Change Cooling Technology for Data Center Servers
Energies (MDPI AG) · Tiantian Zhao, Rongfeng Sun et al.
This study combines CFD (boiling and condensation) with a validation experiment to examine two-phase immersion cooling of server chips: the fluorinated fluid HFE-7100 boils on the chip and condenses on a water-cooled condenser plate.
CC BY 4.0 — share with attribution
On this page
Overview
Immersion phase-change cooling uses the latent heat of the cooling liquid by putting it in direct contact with the heat-generating chip, and can meet the cooling needs of today's high-heat-flux data centres. A cited reference states that above 100 W/cm² heat flux, air cooling and heat-pipe cooling are no longer applicable. The study varies condenser cooling-water inlet temperature and velocity and the spacing between server boards, tracking chip temperature, temperature spread and a uniformity coefficient.
FC-40 and FC-77 have boiling points too high to boil while the chip stays below its 85 °C safe limit, and HFE-7000 boils even at idle, raising condenser water-pump energy, so all three were excluded and HFE-7100 was used. Against experiment at 20 °C cooling-water inlet and 1 m/s, the model's steady chip surface temperature differed by no more than 2.5 °C. Flow-rate cases used a cooling-water inlet temperature of 24 °C with velocities of 0.6, 0.8, 1 and 1.2 m/s; steady chip temperature was only weakly affected by flow rate, while time to stabilise fell as flow increased.
Increasing board spacing from 5 mm to 15 mm cut maximum temperature by 5.06%, reduced temperature difference from 16.5 °C to 11 °C and lowered the uniformity coefficient by 52.6%. Going from 15 mm to 25 mm spacing reduced maximum temperature by only 1.07%, with temperature difference falling from 11 °C to 10.3 °C, so 15 mm was judged optimal. Lower water inlet temperature mainly lowers steady chip temperature, while higher flow mainly shortens the transient.
Limits: a single fluid (HFE-7100), specific power conditions and a small-scale model; results do not transfer directly to other fluids or server geometries. Boiling point is decisive in fluid selection: a fluid is needed that boils below the chip safe limit but does not boil continuously at idle.
Key points
- FC-40 and FC-77 have boiling points too high to boil while the chip stays below its 85 °C safe limit; HFE-7000 boils even at idle; all three were excluded and HFE-7100 was used.
- Against experiment at 20 °C cooling-water inlet and 1 m/s, the model's steady chip surface temperature differed by no more than 2.5 °C.
- Flow-rate cases used a cooling-water inlet temperature of 24 °C with velocities of 0.6, 0.8, 1 and 1.2 m/s; steady chip temperature was only weakly affected by flow rate, while time to stabilise fell as flow increased.
- Increasing board spacing from 5 mm to 15 mm cut maximum temperature by 5.06%, reduced temperature difference from 16.5 °C to 11 °C and lowered the uniformity coefficient by 52.6%.
- Going from 15 mm to 25 mm spacing reduced maximum temperature by only 1.07% (temperature difference 11 °C to 10.3 °C), so 15 mm was judged optimal.
- A cited reference states that above 100 W/cm² heat flux, air cooling and heat-pipe cooling are no longer applicable.
Using it in design
- Select the two-phase immersion fluid by boiling point: it must boil below the 85 °C chip limit but not boil continuously at idle and raise condenser pump energy (HFE-7100 was chosen).
- Keep server board spacing at about 15 mm; 5 to 15 mm cut maximum temperature by 5.06%, 15 to 25 mm only 1.07%.
- Use condenser water inlet temperature as the variable for steady chip temperature and flow rate for transient time.
Monthly newsletter · free
Keep your reading list current
Once a month: new library sources, changes to Turkish regulations and upcoming data centre events.
SoSo CAD
Check cooling capacity alongside your design
Review zone heat loads, available capacity and the capacity balance after the largest unit is lost in SoSo CAD.
Request a demo
Who should read it
Cooling engineers and researchers designing two-phase immersion fluid and board spacing.
Document contents
- 1. Introduction
- 2. Establishment of a Simulation Model (2.1 Selection of Cooling Liquid, 2.2 Physical Model, 2.3 Numerical Model: VOF model, turbulence models, evaporation-condensation model; 2.5 Grid Independence Verification, 2.6 Model Validation, 2.7 Evaluation Index)
- 3. Results and Discussion (3.2 The Effect of Different Cooling Water Inlet Flow Rates on Cooling Performance)
- 4. Conclusions
Read the document
immersion-phase-change-server-simulation-2023.pdf
CC BY 4.0 — share with attribution
The document is provided unaltered because its licence permits redistribution.
Publication details
- Authors
- Tiantian Zhao; Rongfeng Sun; Xukai Hou; Jikai Huang; Wenguang Geng; Jianguo Jiang
- Publisher
- Energies (MDPI AG)
- Year
- 2023
- Volume
- 16
- Issue
- 12
- Pages
- 4640
- Document language
- English
- Source type
- Scientific
- Access
- Open licence — downloadable
- Licence
- CC BY 4.0 — share with attribution
- Licence evidence
- www.mdpi.com
Original licence note
PDF p.1 states the article is open access under the Creative Commons Attribution (CC BY 4.0) licence; MDPI publishes all articles under CC BY 4.0.
Citation
Tiantian Zhao; Rongfeng Sun; Xukai Hou; Jikai Huang; Wenguang Geng; Jianguo Jiang (2023). Simulation Study of Influencing Factors of Immersion Phase-Change Cooling Technology for Data Center Servers. Energies (MDPI AG). 16(12), 4640. https://doi.org/10.3390/en16124640
Events on this subject
- SC26 – International Conference for High Performance Computing, Networking, Storage and Analysis
- NVIDIA GTC
- ISC High Performance
All content and rights belong to the respective authors and publishers. SoSo CAD provides only the citation, its own summary and a link to the source, and is not responsible for the sources' content.