Towards energy smart data centers: Simulation of server room cooling system

Abstract

Cooling is an extremely important process in modern data centers. Cooling systems of server rooms ensure appropriate operation conditions to IT systems, such as servers and data storage, but, on the other side, they consume a lot of energy. Current control systems, which are installed in data centers and are responsible for thermal management of the facilities, are following conservative control strategies that maintain constant thermal conditions irrespective of computer load and outside temperature, thus efficient use of energy has not been appropriately addressed by them. In this paper, a method of optimizing energy consumption while maintaining an acceptable level of thermal comfort for CPUs in the server room is proposed. In the proposed method, a behavioral thermal model for server room should be created first, and then thermal behavior of the server room would be simulated under different circumstances using its thermal model, in order to find an optimum control strategy capable of retaining balance between thermal comfort and efficient use of energy. The effectiveness of the proposed method has been investigated via simulating a typical server room using MATLAB and SIMULINK and the results are demonstrated.