Three computational methods for analysing thermal airflow distributions in the cooling of data centers

This paper develops computational models to study thermal airflow distribution when cooling a data center. The thermal airflow distribution study employs three computational approaches, namely finite element, finite volume and lattice Boltzmann methods, which are respectively implemented via commercial Multiphysics software, opensource CFD code and home grown GPU based code. The results focus on comparison of the three methods, all of which include models for turbulence, when applied to 2 rows of racks with cool air supplied by a plenum. Advantages and disadvantages of the three computational methods are addressed in terms of application to thermal management of data centers.