| AI-based energy saving in data center has a very wide application prospect.For this purpose,a digital model of the server room is required to provide training data for AI energy saving algorithms.Usually,a server room should maintain at a constant temperature,so that it is necessary to adjust the cooling air inlet parameters in real time under various working conditions.Therefore,numerical simulation study of the thermal environment of the server room under variable working conditions is of great significance for energy saving.In this paper,a data center in Changzhou was taken as the research object and the following research was carried out in the present paper.Rack servers,as the core equipment in the server room,are necessary to be simplified in numerical simulations,and a porous media model was employed in modeling the airflow through the servers.Three different CFD models were established for the cold aisle above a perforated plate and the adjacent server rack.Measurements were carried out to get the boundary conditions,and the velocity and temperature fields were obtained by simulation and compared with the experimental measurements.The optimal model showed a better performance in terms of efficiency and computation cost.Subsequently,the transient calculations of the optimal model under variable working conditions were performed to analyze the temperature variation inside the rack with respect of time.The applicability of the porous media model was verified by a single rack model,and the method was applied to a server room in a large data center in Changzhou.Numerical simulation under standard operating conditions was performed and the results were compared with measurement result.The air outlet model was used at perforated plate to describe the pressure decay and flow jet phenomenon.The fan model and porous media model were used at the rack to model the function of the fun in the server and the pressure drop as the cooling air through the server,respectively.The air flow and temperature fields were calculated and solved,and the simulation results were compared with the experimental measurement data.The following results were obtained: the average relative error between the simulation and measurement results for the velocity and temperature in the cold aisle was 11.4% and 3.3%,and the average relative error between the simulation and measurement results for the temperature in the hot aisle was 2.3%.Finally,the numerical simulation of the room for variable conditions showed that when adjusting the inlet air temperature,the rack outlet air temperature decreased linearly with the inlet air temperature.Due to the influence on convection heat transfer coefficient,increasing the air flow rate resulted in the decrease of the rack outlet temperature,but it was no longer a linear relationship.A critical air flow rate was found that when the inlet air was larger than the critical value,the rack exit temperature drop rate gradually became smaller.Finally,according to the return air control method,the return air temperature was set at 30℃,and different control schemes of the inlet air parameters under variable working conditions were investigated.The results obtained could provide a theoretical model for the energy-saving control scheme. |
