Energy performance analysis of multi-chiller cooling systems for data centers concerning progressive loading throughout the lifecycle under typical climates

Overview

The rising demand for cooling energy is a global concern, but existing studies lacked a comprehensive analysis of the energy performance of widely used multi-chiller cooling systems in air-cooled data centres throughout their lifecycle, especially under progressive loading. Because IT load ramps up over the facility's life, the plant spends most of its life at part load, where chiller staging steps, oversized towers and switchover wet-bulb thresholds change free-cooling hours and system COP. Chilled-water supply is set at 13 °C, giving 20-22 °C supply air in line with the recommended 20-27 °C space temperature, with a 10 °C air temperature difference across computer rooms.

Free-cooling hours at part-load ratio (PLR) 1 ranged from 5921 h in Ulanqab to only 158 h in Hong Kong. The wet-bulb switchover from partial to full free cooling is 11 °C at PLR 0.3 but 8 °C at PLR 1, because towers are effectively oversized at low load; free-cooling hours can therefore differ by up to 1442 h between part-load ratios at one site. Cooling-system COP varies by up to 6 across part-load ratios, a PUE difference of up to 0.14, with sharp COP drops at PLR 0.25, 0.5 and 0.75 when extra equipment stages on.

With a progressive load ramp from 30% to 90% (Rasmussen profile), the lifecycle-average cooling-system COP in Beijing is only 11.7, 2.9 lower than the design COP. Climate performance is mainly set by the annual wet-bulb distribution, and extremely cold regions need freeze protection that can lower efficiency.

Limits: the study is confined to Chinese climates and specific equipment test data and typical staging controls; results may change with other equipment choices and control logic. It still shows that design-point COP is much higher than the lifecycle average.

Key points

1. Chilled-water supply is set at 13 °C, giving 20-22 °C supply air in line with the recommended 20-27 °C space temperature, with a 10 °C air temperature difference across computer rooms.
2. Free-cooling hours at PLR 1 ranged from 5921 h in Ulanqab to only 158 h in Hong Kong.
3. The wet-bulb switchover from partial to full free cooling is 11 °C at PLR 0.3 but 8 °C at PLR 1, because towers are effectively oversized at low load; free-cooling hours can differ by up to 1442 h between part-load ratios at one site.
4. Cooling-system COP varies by up to 6 across part-load ratios (a PUE difference of up to 0.14), with sharp COP drops at PLR 0.25, 0.5 and 0.75 when extra equipment stages on.
5. With a progressive load ramp from 30% to 90% (Rasmussen profile), the lifecycle-average cooling-system COP in Beijing is only 11.7, 2.9 lower than the design COP.
6. Climate performance is mainly set by the annual wet-bulb distribution; extremely cold regions need freeze protection that can lower efficiency.

Using it in design

• Calculate cooling plant efficiency from the lifecycle-average COP reflecting progressive IT load ramp rather than the design-point COP (in Beijing 2.9 lower than design).
• Add part-load review of chiller staging steps and towers (COP drops at PLR 0.25, 0.5, 0.75) to design verification; staging and tower size shift the free-cooling switchover threshold.
• Include the efficiency penalty of freeze protection in cold climates in the design energy model.

Who should read it

Cooling designers evaluating multi-chiller waterside-economiser plants under progressive loading and climate.

Document contents

1. 1. Introduction
2. 2. Methodology (2.1.2 Typical operation modes, 2.3.1 Cooling system designs for different climate conditions, 2.3.2 Data center energy flow and performance metrics)
3. 3. Energy performance under full-range loads and progressive loading (3.2.1 Free cooling hour under full-range cooling loads, 3.2.2 Energy performance under full-range cooling loads, 3.2.3 Impact of climate condition on energy performance, 3.3 Life-cycle energy performance under progressive loading)
4. 4. Discussion
5. 5. Conclusions

Related standards

• ASHRAE TC9.9 Thermal Guidelines
• ASHRAE 90.4

Read the document

Publication details

Authors

Yingbo Zhang; Hangxin Li; Shengwei Wang

Publisher

Building Simulation (Springer Science and Business Media LLC)

Year

2024

Volume

17

Issue

10

Pages

1693-1708

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.1007/s12273-024-1167-9

Licence

CC BY 4.0 — share with attribution

Licence evidence

doi.org

Citation