Document Type

Conference Paper

Rights

Available under a Creative Commons Attribution Non-Commercial Share Alike 4.0 International Licence

Disciplines

Electrical and electronic engineering

Publication Details

ITherm: IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems.

doi:10.1109/ITHERM.2017.7992572

Abstract

Data centres account for approx. 1.3% of the world's electricity consumption, of which up to 50% of that power is dedicated to keeping the actual equipment cool. This represents a huge opportunity to reduce data centre energy consumption by tackling the cooling system operations with a focus on thermal management. This work presents a novel Data Centre Air Flow Model (DCAM) for temperature prediction of server inlet temperatures. The model is a physics-based model under-pinned by turbulent jet theory allowing a reduction in the solution domain size by using only local boundary conditions in front of the servers. Current physics-based modeling approaches require a solution domain of the entire data centre room which is expensive in terms of computation even if a small change occurs in a localised area. By limiting the solution domain and boundary conditions to a local level, the model focuses on the airflow mixing that affects temperatures while also simplifying the related computations. The DCAM model does not have the usual complexities of numerical computations, dependencies on computational grid size, meshing or the need to solve a full domain solution. The input boundary conditions required for the model can be supplied by the Building Management System (BMS), Power Distribution Units (PDU), sensors, or output from other modeling environments that only need updating when significant changes occur. Preliminary results validated on a real world data centre yield an overall prediction error of 1.2°C RMSE. The model can perform in real-time, giving way to applications for real-time monitoring, as input to optimise control of air conditioning units, and can complement sensor networks.

DOI

https://doi.org/10.1109/ITHERM.2017.7992572

Download


Share

COinS