Optimal supply air temperature with respect to data center operational stability and energy efficiency in a row-based cooling system under fault conditions

Abstract

This is the first study of data center row-based cooling strategies under fault conditions. An optimal supply air temperature (SAT) for row-based cooling is proposed to ensure energy efficiency and operational reliability. The appropriate SAT for emergency operation is determined by evaluating how long the allowable IT equipment's inlet air temperature (IAT) can be maintained by only running the fan of the computer room air handling (CRAH) unit when the chilled water supply is interrupted. Three cases with SAT values of 20 °C, 22 °C, and 24 °C are evaluated in a numerical analysis, and the maintenance potential of the IAT is assessed by analyzing 600 s of transients from the time of cooling failure. In all cases, the IAT is almost the same as the SAT under normal cooling, but in cooling failure, the time to reach downtime beyond the allowable IAT is 246s in Case 1, 133s in Case 2, and 85s in Case 3. Response strategies to cooling failure include increasing uninterruptible power supply (UPS) capacity and applying a thermal buffer tank. The study concludes that it is reasonable to set the SAT of row-based cooling to 22 °C or lower and to install a thermal buffer tank in the chilled water system for effective downtime delay. These results are based on a rack power density of 7 kW/rack.