Application of GM(0,N) Model in Analysis of Data Center Design Factors

Abstract

The paper uses the Computational Fluid Dynamics (CFD) simulation to evaluate the performance of data center. Five design parameters are adopted: cooling capacity, server cabinet density, server placement, floor-to-ceiling height, and aisles width. CFD method was used to analyze flow distribution in the server cabinet and simulate the impact factors on the overheating cabinet. Firstly, the various design parameters were obtained from the CFD simulation results. Then, gray GM (0, N) was applied to the weight mathematical model using different design parameters. Finally, It is found that the server cabinet density is the most influent parameter, while less effect on the floor-to-ceiling height. The comparison of the data with actual conditions indicates that the theoretical coefficients of various locations are identical with the results of similarity classification. The proposed mathematical models in this paper are reasonable and feasible for evaluating the performance of data center design.