Rack Level Study of Hybrid Cooled Servers Using Warm Water Cooling with Variable Pumping for Centralized Coolant System

Abstract

As the worldwide demand for the data centers grows, so does the size and load placed on data centers which leads to the applied constraints on power and space available to the operator. Cooling power consumption is a major contributor of the total energy consumption of the system. In the process of optimization of cooling energy consumption per performance unit, liquid cooling technology has emerged as one of the most viable solutions. In this rack level study, 2OU (Open U) hybrid cooled web servers are tested for an evaluation of warm water cooling in centralized coolant system. Effects of higher inlet temperatures of the coolant in terms of device temperatures as well as IT and cooling power have been observed as a part of the evaluation. The study discusses the significance of variable pumping in centralized coolant system for its more efficient use. The experimental setup for cooling consists of 1/3rd sized mini rack capable of housing up to eleven liquid cooled web servers and two heat exchangers. The cooling configuration is centralized and has two redundant pumps placed in series with heat exchanger at the rack. CPUs of each server are liquid cooled with using passive micro-channel cold plates while rests of the components are air cooled. Synthetic load has been generated on each servers for thermal stress testing and observed performance characteristics such as device temperatures and cooling power consumption of servers. Centralized redundant pumps are separately powered using an external DC power supply unit. The pump speed is varied with variable voltage supply ranging from 11V to 17V across the armature. The experimental testing is carried out at higher inlet temperatures ranging from 25°C to 45°C which falls within the ASHRAE liquid cooled envelope W4. Variable pumping at higher inlet temperatures has been achieved to evaluate to operating temperatures of device components for reliability and reduction in operational cooling power consumption of the servers.