| Summary: | The motivation of this study is to address the demand of more efficient data centres as the traditional method of cooling using air conditioning is no longer effective with the increasing density of server racks in a data centre. This study aims to investigate the effectiveness of immersion cooling on the behaviour of high performance servers. The focus of this paper will be on the Two-Phased Immersion Cooling (TPIC) method with a single motherboard server which will be submerged into a tank filled with liquid coolant Novec 7100 cooled by an external circulator unit. The experiment is done by adjusting different parameters such as condenser temperature, time and Central Processing Unit (CPU) load to understand the effects on cooling. Based on the experimental results, the use of TPIC method without forced convection decreases the temperature by 29.6% as compared to air cooling. In addition, TPIC with forced convection has shown to improve the cooling performance by an additional 8%. A prolonged test was also conducted to further understand the effects of cooling for an extended period. These results have shown the importance of certain parameters which would considerably affect the performance of the cooling in a TPIC setup. Several recommendations for future work are also presented at the end of the report. These includes the need to explore the use of different liquid coolant, monitoring the temperature of the liquid coolant in the tank and the power consumption for the overall system, inclusion of new parameters - flow of pump and lastly a thorough experiment with the prolonged temperature in different conditions as set in this paper. This study ultimately concludes that immersion cooling is much more efficient in removing heat from the server as compared to air cooled.
|
|---|