Quantifying Air Flow Rate Through A Server In An Operational Data Center And Assessing The Impact Of Using Theoretical Fan Curve

Abstract

Cooling is one of the major cost factors in data centres that accounts for 37% of the total data centre electricity consumption. Demands on cooling systems have increased substantially in recent years, due to increasing server densities. To avoid the overheating of servers, cooling aisles are over provisioned, but little attention is paid to the actual scenario at the server level. Cooling requirements of the servers is calculated on the basis of fan manufacturer’s published data for the air flow capacity of the server fan and system impedance data obtained from the server designer put together. Thus, the air flow rate through the system is assumed to be the theoretical air flow rate delivered by the server fans that can overcome the system impedance of the server. The following experimental work analyses the impact of different flow conditions on the volumetric flow rates through the server .Operating points are obtained at various fan duty cycles. The air flow rates through the server at various operating points are quantified in ideal and operational data centre conditions. The study predicts that this data will be useful to quantify the actual air flow rate across the racks and might be useful in controlling the facility parameters that will modulate the airflow distribution and in turn the power consumption due to cooling.