Enhancing heat transfer processes in 2U-Rack Servers: A study case employing thermal simulation and the Taguchi method

Abstract

The data center industry consumes large quantities of energy due to the consumption spawned by Information Technology Equipment (ITE) and air conditioning systems. Servers constitute the predominant portion of ITE within a data center. Relatedly, diverse optimization strategies are required to minimize data centers' energy consumption. This paper used a 2U-Rack Server, specifically, the R261-3C0 server as a study case to enhance heat transfer processes by minimizing the temperature variation in the high thermal-stress components and optimizing airflow patterns. The Taguchi Method is used for this purpose. A thermal simulation is conducted using Computational Fluid Dynamics (CFD) techniques through the Ansys Fluent code. Seven design variables were proposed as control factors, submitting them to the noise factors. New parameters such as the motherboard lifting, the displacement of heatsinks and the CPUs, and the fan speed variation will be introduced among these control factors. Among the noise factors, we considered the fluid velocity at the server inlet region and the Operating Point of the Fan (OPF). After applying the Taguchi Method, the control factors and their optimal levels were obtained. The thermal simulation presented a consistent performance. The optimal configuration of the control parameters generated a 10 °C decrease in the server's overall temperature, an air circulation enhancement by the vector field behavior verification, and a 17.11 % decrease in the server's thermal variation. The optimized model stays in its verification phase, and the validation phase should be accomplished in future work.