Due to high energy demands of data centers and the energy crisis throughout the world, efficient heat transfer in a data center is an active research area. Until now major emphasis lies upon study of air flow rate and temperature profiles for different rack configurations and tile layouts. In current work, we consider different hot aisle (HA) and cold aisle (CA) configurations to study heat transfer phenomenon inside a data center. In raised floor data centers when rows of racks are parallel to each other, in a conventional cooling system, there are equal number of hot and cold aisles for odd number of rows of racks. For even number of rows of racks, whatever configuration of hot/cold aisles is adopted, number of cold aisles is either one greater or one less than number of hot aisles i.e. two cases are possible case A: n(CA) = n(HA) + 1 and case B: n(CA) = n(HA) − 1 where n(CA), n(HA) denotes number of cold and hot aisles respectively. We perform numerical simulations for two (case1) and four (case 2) racks data center. The assumption of constant pressure below plenum reduces the problem domain to above plenum area only. In order to see which configuration provides higher heat transfer across servers, we measure heat transfer across servers on the basis of temperature differences across racks, and in order to validate them, we find mass flow rates on rack outlet. On the basis of results obtained, we conclude that for even numbered rows of rack data center, using more cold aisles than hot aisles provide higher heat transfer across servers. These results provide guidance on the design and layout of a data center.
High-energy heavy-ions physics: from RHIC to LHC
