Effects and optimization of airflow on the thermal environment in a data center

Overview

The study addresses rising data-centre energy consumption by optimising airflow organisation, comparing three cooling systems that are usually considered in isolation. The room is 16.2 m × 9.45 m × 5.125 m (about 153 m²) with 55 cabinets of 600 × 1200 × 2200 mm at 6 kW each, laid out in three rows in a hot/cold aisle pattern. In the raised-floor case, two perimeter units of up to 230 kW and 45,900 m³/h each supply 20 °C air through a 0.64 m raised floor and 0.6 × 0.6 m floor grilles at 50% open area.

The raised-floor and in-row cases were each run open, with cold-aisle containment (CAC) and with hot-aisle containment (HAC), giving seven cases. Air supply efficiency (ASE), air return efficiency, a recirculation index, RTI, SHI and RHI were tabulated. Raised floor without containment gave ASE 65.69%, RTI 77.94% and SHI 0.141; with CAC, ASE 85.57% and SHI 0.0027; with HAC, ASE 90.25%, RTI 90.94% and SHI 0.0024. Without containment, cabinet No. 19 had a mean inlet temperature of 30.1 °C and a maximum of 36.7 °C, and ten cabinets went above the 27 °C specification limit.

In the in-row case, three units of 51.1 kW sensible and 10,200 m³/h each gave ASE 71.29% and SHI 0.145 without containment, and with CAC ASE 92.16%, ARE 99.88% and SHI 0.0015. With 6 kW heat-pipe rear-door coolers, cabinet inlet temperatures averaged about 21.5 °C and varied by no more than 2 °C within a cabinet, and the room stayed at about 22 °C without aisle containment.

Limits: the hall was still being built, so the model was checked against a second CFD code rather than measurements: COMSOL against 6SigmaRoom gave a mean cabinet inlet temperature error below 10%. There is no measurement validation.

Key points

1. The room is 16.2 m × 9.45 m × 5.125 m (about 153 m²) with 55 cabinets of 600 × 1200 × 2200 mm at 6 kW each in three rows in a hot/cold aisle pattern.
2. Raised-floor case: two perimeter units of up to 230 kW and 45,900 m³/h each, 20 °C supply, a 0.64 m raised floor and 0.6 × 0.6 m floor grilles at 50% open area.
3. Raised floor without containment: ASE 65.69%, RTI 77.94%, SHI 0.141. With CAC: ASE 85.57%, SHI 0.0027. With HAC: ASE 90.25%, RTI 90.94%, SHI 0.0024.
4. Without containment, cabinet No. 19 had a mean inlet temperature of 30.1 °C and a maximum of 36.7 °C, and ten cabinets went above the 27 °C limit.
5. In-row case: three units of 51.1 kW sensible and 10,200 m³/h each. Without containment ASE was 71.29% and SHI 0.145; with CAC, ASE 92.16%, ARE 99.88% and SHI 0.0015.
6. With 6 kW heat-pipe rear-door coolers, cabinet inlet temperatures averaged about 21.5 °C and varied by no more than 2 °C within a cabinet; the room stayed at about 22 °C without aisle containment.
7. Validation was code-to-code only: COMSOL against 6SigmaRoom gave a mean cabinet inlet temperature error below 10%, because the room was still under construction.

Using it in design

• Make aisle containment the default in raised-floor or in-row cooling: containment raised ASE from 65.69% to 85.57-90.25% and brought SHI close to zero.
• Show temperature-limit exceedance by cabinet (ten cabinets above 27 °C without containment) and check the worst cabinet (30.1 °C mean, 36.7 °C maximum) separately.
• Where aisle separation is not possible in high-density cabinets, consider heat-pipe rear-door coolers (for 6 kW cabinets, inlet about 21.5 °C) as an option.
• Label results that rest only on code-to-code validation in a hall under construction as preliminary design, not firm values.

Who should read it

Cooling designers comparing raised-floor, in-row and rear-door cooling arrangements for new data halls.

Document contents

1. 1 Introduction
2. 2 Materials and methods (2.1 Physical model, 2.2 Numerical simulation: mathematical model, boundary conditions and solution settings, grid independence study and model)
3. 3 Results and discussion (3.1 Option 1: precision air conditioners delivering air from the raised floor: without containment, CAC, HAC, quantitative analysis; 3.2 Option 2: horizontal air supply of inter-column air conditioners: without containment, CAC, HAC, quantitative analysis; 3.3 Option 3: backplane air conditioning and temperature field analysis)
4. 4 Conclusion

Related standards

• ASHRAE TC 9.9 Thermal Guidelines for Data Processing Environments

Read the document

Publication details

Authors

Di Jiang

Publisher

Frontiers in Built Environment (Frontiers Media SA)

Year

2024

Volume

10

Pages

1362861

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.3389/fbuil.2024.1362861

Licence

CC BY 4.0 — share with attribution

Licence evidence

www.frontiersin.org

Citation