CFD Analysis of Data Center Hall Cooling Performance under Normal and Failure Modes with Control Strategies and Airflow Leakages

Overview

The study asks how a dynamically controlled, non-raised-floor hall with hot-aisle containment (HAC) cools under normal and failure modes, with control strategy and air leakage. The hall is 41 m × 20 m (810 m²), 5.7 m floor-to-ceiling, with 308 cabinets of 0.6 × 1.2 × 2.4 m and 1920 kW IT (6.23 kW per cabinet) in 14 rows with HAC. Nine ACUs of 274.20 kW each were modelled, giving 2467.80 kW in normal mode and 1919.40 kW with two ACUs offline in failure mode, against a room load of about 2000-2018 kW. Each ACU has four fans (19.5 m³/s maximum per ACU) and a 25 °C off-coil setpoint.

Fan control used delta-T (return setpoint 37 °C) and a 3 Pa hot/cold differential-pressure floor, with fan speed limited to 40-100%. Leakage was modelled as 5% cabinet gaps and 0.561 mm HAC panel gaps. The grid study selected 7,288,454 cells with the standard k-epsilon model and wall functions. Eight cases crossed normal and failure modes with and without control strategy and leakage.

In normal mode with control and leakage, peak cabinet inlet temperatures ranged from 25.97 to 30.11 °C, and ACUs ran at only 74.07-86.30% of available capacity. In failure mode with leakage (Cases 1 and 3), peak inlet temperatures reached 34.82 and 34.66 °C, with some cabinets above 32 °C; without leakage (Cases 2 and 4), peaks were 27.27 and 27.15 °C. The control strategy cut cooling power by about 9.89% in normal mode but made little difference in failure mode, where fans and valves were already near 100%.

Limits: simulation only, with no field measurement, and a design-stage model; leakage sizes (5% cabinet gaps, 0.561 mm panel gaps) are assumptions. The finding is that the leak-free model stays compliant in failure mode while realistic leakage breaks compliance, so failure-mode assurance given without modelling leakage may be optimistic.

Key points

1. The hall is 41 m × 20 m (810 m²) with 5.7 m floor-to-ceiling, 308 cabinets of 0.6 × 1.2 × 2.4 m and 1920 kW IT (6.23 kW per cabinet) in 14 rows with HAC on a non-raised floor.
2. Nine ACUs at 274.20 kW each give 2467.80 kW in normal mode and 1919.40 kW with two ACUs offline in failure mode, against a room load of about 2000-2018 kW.
3. Fan control used delta-T (return setpoint 37 °C) and a 3 Pa hot/cold differential-pressure floor, fan speed limited to 40-100%; leakage was modelled as 5% cabinet gaps and 0.561 mm HAC panel gaps.
4. The grid study selected 7,288,454 cells with the standard k-epsilon model and wall functions.
5. In normal mode with control and leakage, peak cabinet inlet temperatures ranged from 25.97 to 30.11 °C, and ACUs ran at only 74.07-86.30% of available capacity.
6. In failure mode with leakage (Cases 1 and 3), peak inlet temperatures reached 34.82 and 34.66 °C with some cabinets above 32 °C; without leakage (Cases 2 and 4), peaks were 27.27 and 27.15 °C.
7. The control strategy cut cooling power by about 9.89% in normal mode but made little difference in failure mode, where fans and valves were already near 100%.

Using it in design

• Do thermal verification of failure mode (two ACUs offline) with realistic air leakage (cabinet gaps, HAC panel joints); a leak-free model can comply while the leaky model exceeds 32 °C.
• Use fan and valve control (delta-T and differential-pressure floor) for normal-mode energy savings, but assume control gives no gain in failure mode (fans already near 100%).
• Specify sealed cabinet gaps and HAC panel joints as part of the redundancy design and inspect leakage at commissioning.

Who should read it

Cooling designers verifying normal and failure mode thermal performance of slab-floor, hot-aisle-contained halls.

Document contents

1. 1. Introduction (1.1 Literature Review, 1.2 Role of CFD in Data Center Design, 1.3 Raised Floor Versus Non-Raised Floor Data Hall, 1.4 Scope of Study, 1.5 Novelty and Contribution)
2. 2. CFD methodology (governing equations, turbulence modelling, near-wall treatment and wall functions, thermal modelling and buoyancy, numerical methodology and convergence, validity of CFD simulation, mesh independence study)
3. 3. Data center working principle and modes of operation
4. 4. ACU construction, position and control strategy
5. 5. Design data for CFD simulation, design considerations and simulation scenarios
6. 6. Normal Mode and Failure Mode
7. 7. Results (normal mode steady state operation, failure mode operation)
8. 8. Conclusions

Related standards

• ASHRAE TC 9.9 Thermal Guidelines for Data Processing Environments

Read the document

Publication details

Authors

Sushil Ashok Surwase; Suribabu Badde; R. Balakrishnan

Publisher

Journal of Engineering Research and Sciences (Journal of Engineering Research and Sciences)

Year

2026

Volume

5

Issue

1

Pages

9

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.55708/js0501002

Licence

CC BY-SA 4.0 — share with attribution under the same licence

Licence evidence

www.jenrs.com

Citation