Rapid prediction of exergy destruction in data centers due to airflow mixing

Abstract

ABSTRACTA novel lumped parameter approach is introduced to predict data center exergy destruction due to airflow mixing, resulting in a speedup of several orders of magnitude compared to detailed computational fluid dynamics (CFD) simulations. Both lumped parameter and detailed CFD methods agree within 8.5% for 11 test cases on an example data center. A significant time-saving design strategy is also introduced using two detailed CFD simulations to predict bulk flow parameters, and then applying lumped parameter analysis on flow-independent design parameters. The strategy shows agreement within 0.39% when computer room air conditioner (CRAC) supply temperature is varied from 12–20°C.