Comparing Performance Metrics of Partial Aisle Containments in Hard Floor and Raised Floor Data Centers Using CFD

Overview

The study numerically investigates the effect of partial aisle containment between server racks in hard-floor and raised-floor configurations. The room measures 6.484 × 7.000 × 3.150 m with a switchable hard-floor or raised-floor setup; perforated tiles had 56% and 77% open area and were modelled as fully open because open area was at least 50%; supply temperature was 18.3 °C. Validation used velocity profiles at heights of 0.6, 1.0, 1.4, 1.8 and 2.2 m at five locations in both open-aisle configurations, with discretisation error estimated with the GCI on three tetrahedral grids. The Reynolds stress model was used instead of k-epsilon because k-epsilon is less accurate in low-velocity regions.

At operating supply flow, RCI was 100% for both floor types while RTI was only about 40% (hard floor) and 50% (raised floor), meaning the room was well cooled but energy-inefficient because of oversupply. With supply cut by 50%, RTI rose above 80%, and recirculation appeared in all containment variants while RCI still looked acceptable.

At 50% flow, the Capture Index of end rack R1 was 44.66% on the open hard floor versus 86% on the open raised floor (96.88% with a raised floor and side cover), showing local problems that averaged metrics hide. The authors argue for combining temperature-based and airflow-based metrics. Airflow management improved in the raised-floor configurations, where a top or side cover fully prevented recirculation.

Limits: a small test room with uneven rack load; results depend on this room geometry, and only two supply flow rates (100% and 50%) were examined.

Key points

1. The room measures 6.484 × 7.000 × 3.150 m with a switchable hard-floor or raised-floor setup; perforated tiles had 56% and 77% open area (modelled as fully open since at least 50%); supply temperature was 18.3 °C.
2. Validation used velocity profiles at heights of 0.6, 1.0, 1.4, 1.8 and 2.2 m at five locations in both open-aisle configurations; discretisation error was estimated with the GCI on three tetrahedral grids.
3. The Reynolds stress model was used instead of k-epsilon because k-epsilon is less accurate in low-velocity regions.
4. At operating supply flow, RCI was 100% for both floor types while RTI was only about 40% (hard floor) and 50% (raised floor): well cooled but energy-inefficient because of oversupply.
5. With supply cut by 50%, RTI rose above 80%, and recirculation appeared in all containment variants while RCI still looked acceptable.
6. At 50% flow, the Capture Index of end rack R1 was 44.66% on the open hard floor versus 86% on the open raised floor (96.88% with a raised floor and side cover).

Using it in design

• In airflow assessment do not rely on RCI alone; RCI of 100% coexisted with RTI of 40-50%, showing oversupply and bypass. Report RTI and per-rack Capture Index together.
• When reducing supply flow (50%), recognise that recirculation can appear and averaged metrics can hide it; check per-rack CI for end racks.
• In raised-floor layouts, consider a side or top cover for partial containment; on a raised floor these fully prevented recirculation.

Who should read it

Cooling engineers evaluating partial aisle containment and supply-flow reduction using airflow metrics.

Document contents

1. 1. Introduction
2. 2. Method (2.1 Geometry, 2.2 Boundary Conditions, 2.3 Validation, 2.4 Grid Study, 2.5 Simulation Settings, 2.6 Performance Metrics)
3. 3. Results and Discussion (3.1 Grid Study, 3.2 Validation, 3.3 Flow Field, 3.4 Comparison of Setups at Operating Conditions, 3.5 Comparison of Setups with 50% Supply Flow Rate)
4. 4. Conclusions

Related standards

• ASHRAE TC 9.9 Thermal Guidelines for Data Processing Environments

Read the document

Publication details

Authors

Emelie Wibron; Anna-Lena Ljung; T. Staffan Lundström

Publisher

Energies (MDPI AG)

Year

2019

Volume

12

Issue

8

Pages

1473

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.3390/en12081473

Licence

CC BY 4.0 — share with attribution

Licence evidence

www.mdpi.com

Citation