Experimental and numerical analysis for potential heat reuse in liquid cooled data centres

• Published in: Energy conversion and management 2016-03, Vol.112, p.135-145
• Main Authors: Carbó, Andreu, Oró, Eduard, Salom, Jaume, Canuto, Mauro, Macías, Mario, Guitart, Jordi
• Format: Article
• Language: English
• Subjects: Aplicacions de la informàtica; Consum; Data centers; Data centre; Eficiència energètica; Energia; Energies; Energy consumption; Energy management; Energy use; Experimentation; Informàtica; IT load; Liquid cooling; Liquids; Mathematical models; Models matemàtics; Numerical modelling; Reuse; Water temperature; Àrees temàtiques de la UPC
• ISSN: 0196-8904; 1879-2227
• DOI: 10.1016/j.enconman.2016.01.003

•The potential heat reuse of a liquid data centre has been characterized.•Dynamic behaviours of a liquid cooled data centre have been studied.•A dynamic energy model of liquid cooling data centres is developed.•The dynamic energy model has been validated with experimental data.•Server usage and consumption relation was developed for different IT loads. The rapid increase of data centre industry has stimulated the interest of both researchers and professionals in order to reduce energy consumption and carbon footprint of these unique infrastructures. The implementation of energy efficiency strategies and the use of renewables play an important role to reduce the overall data centre energy demand. Information Technology (IT) equipment produce vast amount of heat which must be removed and therefore waste heat recovery is a likely energy efficiency strategy to be studied in detail. To evaluate the potential of heat reuse a unique liquid cooled data centre test bench was designed and built. An extensive thermal characterization under different scenarios was performed. The effective liquid cooling capacity is affected by the inlet water temperature. The lower the inlet water temperature the higher the liquid cooling capacity; however, the outlet water temperature will be also low. Therefore, the requirements of the heat reuse application play an important role in the optimization of the cooling configuration. The experimental data was then used to validate a dynamic energy model developed in TRNSYS. This model is able to predict the behaviour of liquid cooling data centres and can be used to study the potential compatibility between large data centres with different heat reuse applications. The model also incorporates normalized power consumption profiles for heterogeneous workloads that have been derived from realistic IT loads.