Balancing Cooling and IT Airflow with Dampers in Ceiling-Ducted Hot-Aisle Containment in Data Centers

Physical containment is often used to separate cold and hot airflow streams for reliable and efficient cooling in data centers. In a common implementation, all (or most) data-center-pod hot aisles are ducted directly to the ceiling plenum. Since each contained IT pod typically has different IT airflow and leakage characteristics, the amount of cooling airflow drawn through each pod varies significantly. Therefore, it is usually necessary to oversupply cooling airflow at the data-center level to ensure that sufficient airflow is drawn through each pod. This practice wastes considerable energy due to increased cooling-fan power and reduced cooling-system efficiency. A much closer match between IT and cooling airflow can be achieved by employing adjustable dampers installed between the pod hot aisle and the ceiling plenum. In order to analyze the efficiency improvements afforded by the damper systems, we developed a software tool based on a computationally efficient flow network model (FNM) of the airflow in ceiling-ducted containment systems. This paper describes the FNM and a closed-form technique for predicting optimal damper settings which balance pod airflows. Finally, the FNM can also be used in the development of damper-control systems based on pressure measurements for real facilities.