Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling

ABSTRACT Data centers worldwide face rising electricity consumption, with a substantial portion used for cooling IT components. Traditional air‐cooling methods have become increasingly inefficient as server power densities rise, causing a transition towards liquid‐cooling technologies. Direct liquid...

Full description

Saved in:
Bibliographic Details
Main Authors: Maximilian Stahlhut, Markus Sekulla, Peter Großöhme, Axel Auweter, Thorsten Urbaneck
Format: Article
Language:English
Published: Wiley 2025-07-01
Series:Energy Science & Engineering
Subjects:
Online Access:https://doi.org/10.1002/ese3.70116
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850113408346619904
author Maximilian Stahlhut
Markus Sekulla
Peter Großöhme
Axel Auweter
Thorsten Urbaneck
author_facet Maximilian Stahlhut
Markus Sekulla
Peter Großöhme
Axel Auweter
Thorsten Urbaneck
author_sort Maximilian Stahlhut
collection DOAJ
description ABSTRACT Data centers worldwide face rising electricity consumption, with a substantial portion used for cooling IT components. Traditional air‐cooling methods have become increasingly inefficient as server power densities rise, causing a transition towards liquid‐cooling technologies. Direct liquid‐cooled servers operate at higher coolant temperatures than air‐cooled systems, significantly reducing cooling demands and improving the potential for waste heat reuse. However, elevated coolant temperatures may increase server power consumption, reduce heat transfer to the coolant, and potentially affect IT equipment performance. To systematically quantify these effects, a comprehensive analysis based on benchmark tests was conducted on ten liquid‐cooled servers under varying coolant inlet temperatures (25°C–50°C) and IT loads. Results demonstrated that server performance remained unaffected within this temperature range. Nonetheless, raising the average cooling temperature from 30°C to 50°C increased server power consumption by 8.5% and reduced heat transfer to the coolant by 14.6% due to thermal losses. Based on these experimental results, a simulation model was developed and implemented in TRNSYS. This model enables the prediction of the behavior of direct liquid‐cooled servers and the assessment of practical applications for waste heat utilization.
format Article
id doaj-art-34f9cfcfa5cd4d248eb6de93e2514782
institution OA Journals
issn 2050-0505
language English
publishDate 2025-07-01
publisher Wiley
record_format Article
series Energy Science & Engineering
spelling doaj-art-34f9cfcfa5cd4d248eb6de93e25147822025-08-20T02:37:10ZengWileyEnergy Science & Engineering2050-05052025-07-011373605361910.1002/ese3.70116Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and ModelingMaximilian Stahlhut0Markus Sekulla1Peter Großöhme2Axel Auweter3Thorsten Urbaneck4University of Technology Chemnitz, Professorship Applied Thermodynamics Chemnitz GermanyUniversity of Technology Chemnitz, Professorship Applied Thermodynamics Chemnitz GermanyMegware Computer Vertrieb und Service GmbH Chemnitz GermanyMegware Computer Vertrieb und Service GmbH Chemnitz GermanyUniversity of Technology Chemnitz, Professorship Applied Thermodynamics Chemnitz GermanyABSTRACT Data centers worldwide face rising electricity consumption, with a substantial portion used for cooling IT components. Traditional air‐cooling methods have become increasingly inefficient as server power densities rise, causing a transition towards liquid‐cooling technologies. Direct liquid‐cooled servers operate at higher coolant temperatures than air‐cooled systems, significantly reducing cooling demands and improving the potential for waste heat reuse. However, elevated coolant temperatures may increase server power consumption, reduce heat transfer to the coolant, and potentially affect IT equipment performance. To systematically quantify these effects, a comprehensive analysis based on benchmark tests was conducted on ten liquid‐cooled servers under varying coolant inlet temperatures (25°C–50°C) and IT loads. Results demonstrated that server performance remained unaffected within this temperature range. Nonetheless, raising the average cooling temperature from 30°C to 50°C increased server power consumption by 8.5% and reduced heat transfer to the coolant by 14.6% due to thermal losses. Based on these experimental results, a simulation model was developed and implemented in TRNSYS. This model enables the prediction of the behavior of direct liquid‐cooled servers and the assessment of practical applications for waste heat utilization.https://doi.org/10.1002/ese3.70116data centerexperimentheat reuseliquid coolingmodeling
spellingShingle Maximilian Stahlhut
Markus Sekulla
Peter Großöhme
Axel Auweter
Thorsten Urbaneck
Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling
Energy Science & Engineering
data center
experiment
heat reuse
liquid cooling
modeling
title Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling
title_full Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling
title_fullStr Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling
title_full_unstemmed Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling
title_short Data Centers With Direct Liquid‐Cooled Servers: Experimental Analysis and Modeling
title_sort data centers with direct liquid cooled servers experimental analysis and modeling
topic data center
experiment
heat reuse
liquid cooling
modeling
url https://doi.org/10.1002/ese3.70116
work_keys_str_mv AT maximilianstahlhut datacenterswithdirectliquidcooledserversexperimentalanalysisandmodeling
AT markussekulla datacenterswithdirectliquidcooledserversexperimentalanalysisandmodeling
AT petergroßohme datacenterswithdirectliquidcooledserversexperimentalanalysisandmodeling
AT axelauweter datacenterswithdirectliquidcooledserversexperimentalanalysisandmodeling
AT thorstenurbaneck datacenterswithdirectliquidcooledserversexperimentalanalysisandmodeling