Server performance optimization for single-phase immersion cooling data center

Abstract

With the development of data center, air based cooling method should be updated due to the high energy consumption. Therefore, finding a high efficient cooling method becomes more and more urgent. Single-phase immersion cooling has received more attentions due to the low energy consumption. To improve the performance, more parameters and their interactive impacts should be considered. Therefore, this paper used seven key parameters to optimize server performance by using single-phase immersion cooling method, including fin height (A), fin spacing (B), thermal conductivity (C), height ratio of baffle (D), outlet area (E), inlet flowrate (F), and inlet temperature (G). The maximum temperature and pressure loss are used as the performance indicators. With the validated CFD model and response surface method (RSM), the impact of single parameter, multi-parameter, and weights of maximum temperature and pressure loss are analyzed. Results showed that, the regression equations for maximum temperature and pressure loss can well predict and optimize the server performance. In addition to the impact of single parameter, the interactive impact between height ratio of baffle and flowrate, fin spacing and flowrate, fin height and flowrate, fin spacing and height ratio of baffle, should be carefully considered when optimizing maximum temperature; the interactive impact of outlet area and flowrate should be considered when optimizing pressure loss. The weight could largely affect the optimal result and optimal value, which should be carefully designed for application. The differences in maximum temperature and pressure loss could achieve as high as 19.8 °C and 18.6 Pa when the weight of maximum temperature varied from 0.1 to 10.