Energy efficient decision making in data centers with multiple cooling methods

Cooling systems consume around 40% of modern data centres' total energy consumption, thus reducing the energy waste in this sector will have positive environmental impact. There exist several cooling methods appropriate for particular conditions. Since environmental conditions, such as air temperature, change on seasonal bases, no single cooling method can be claimed to be the best. In contrast, data centres with multiple cooling methods can perform more efficiently. However, the decision-making process of which cooling system is the most appropriate one should be an automated process. In this paper, a decision-making process based on simulation is proposed. The simulation tool comprises of a mathematical model and a multi agent control. The mathematical model simulates the thermal behaviour of SICS ICE data centre, which is a real facility located in Northern Sweden. The main aim of the simulation is to calculate thermal conditions and energy consumption of different cooling methods in cold and hot seasons. The result then will be used by multi-agent control to choose the most appropriate cooling method.