An open-source and experimentally guided CFD strategy for predicting air distribution in data centers with air-cooling

Data centers are generally over cooled to ensure the trouble free running. In a data center with air-cooling, it is crucial to investigate the air distribution and temperature field to aid the design, which ensures the stable operation with less energy consumption. Computational fluid dynamics (CFD) is perfect for such a mission, but the cost is high and the user is offered with little customization as commercial software are mostly adopted. Therefore, this study presented a systematic investigation on making use of open-source software, including OpenFOAM and paraView to realize geometry preparation, mesh generation, experimentally guided numerical setup and solution, and results visualization. A JAVA program was developed to ensure the case preparation and simulation in just one command. Self-adapted momentum sources were developed to realize the desired flow rate through the servers. The strategy was validated and demonstrated by a pilot data center from Alibaba cloud, Alibaba group. The developed solver predicted the air temperature in both cold and hot isles of a data center with mean error of 0.7 K. This work initiated a starting point for achieving automated CFD simulation of data centers with open-source tools.