Energy Efficiency Measurement Method and Thermal Environment in Data Centers—A Literature Review

Overview

The increase in data-centre facilities has led to higher energy consumption and a larger carbon footprint, prompting improvements in thermal environments for energy efficiency and server lifespan. Existing literature often overlooks categorising equipment for PUE, power-efficiency measurement limitations and employee thermal comfort. Thirty-nine papers and nine websites were reviewed. Averaged over the compiled breakdowns, IT equipment is 44.8% of total power, which works out to a PUE of about 2.23 ('Average' on The Green Grid scale); none of the analysed facilities fell into the 'Efficient' or 'Very efficient' PUE bands.

Published studies use inconsistent or vague categories (for example 'Power Supply System', 'Power Conversion', 'Other Equipment'), so assigning loads to IT versus facility power needed author judgement. PUE does not account for data-centre type: a hyperscale site can report a lower PUE than an SMB site that carries office loads, without its servers being more efficient. The authors call for every PUE assessment to state the facility type (edge, SMB, enterprise, colocation, hyperscale) and for IT/facility categorisation to be more detailed. Partial PUE (pPUE) is cited as a way to assess clearly bounded segments of a facility, in contrast to a single whole-site PUE.

Free-cooling feasibility depends on facility type: airside and waterside free cooling are rated 'not feasible' for Telco Edge, Commercial Edge and SMB (urban, space-limited, polluted air) and 'ideal' for Colocation and Hyperscale. Measured indoor data-hall conditions across five sites ranged from 15 to 34 °C and 17 to 68% RH, and only the US site stayed fully within the ASHRAE recommended range (18-27 °C; allowable 15-32 °C depending on class). No study of staff thermal comfort in data centres was found, so computer-lab data were used as a proxy; in one lab PMV was 1.1-1.4, outside the ISO ±0.5 comfort band.

Limits: the review rests on 39 papers and 9 websites, category assignments needed author judgement, and results depend on the quality of published breakdown data. Indoor air temperatures averaged 16.5 °C with 19% relative humidity, within the ASHRAE allowable range.

Key points

1. Averaged over the compiled breakdowns, IT equipment is 44.8% of total power, giving PUE = 1/0.448 = 2.23 ('Average' on The Green Grid scale); none of the analysed facilities fell into the 'Efficient' or 'Very efficient' bands.
2. Published studies use inconsistent or vague categories ('Power Supply System', 'Power Conversion', 'Other Equipment'), so assigning loads to IT versus facility power needed author judgement.
3. PUE does not account for data-centre type: a hyperscale site can report a lower PUE than an SMB site that carries office loads without its servers being more efficient; the authors call for every PUE assessment to state the facility type.
4. Partial PUE (pPUE) is cited as a way to assess clearly bounded segments of a facility.
5. Free-cooling feasibility depends on facility type: airside and waterside free cooling are 'not feasible' for Telco Edge, Commercial Edge and SMB and 'ideal' for Colocation and Hyperscale.
6. Measured indoor conditions across five sites ranged from 15 to 34 °C and 17 to 68% RH, and only the US site stayed fully within the ASHRAE recommended range (18-27 °C).
7. No study of staff thermal comfort in data centres was found; in one computer lab PMV was 1.1-1.4, outside the ISO ±0.5 band.

Using it in design

• In PUE comparisons state the facility type (edge, SMB, enterprise, colocation, hyperscale) and assign IT/facility power categories per The Green Grid definitions in detail; PUE across types is not directly comparable.
• Consider partial PUE (pPUE) for bounded segments of a facility.
• Assess free-cooling suitability by facility type and location (hard in urban, space-limited small facilities; ideal with colocation and hyperscale).
• Add thermal comfort (PMV ±0.5) for data-hall staff to design criteria; no data-centre-specific research exists.

Who should read it

Designers and operators measuring PUE, distinguishing facility types and assessing data-hall thermal environment.

Document contents

1. 1. Introduction (1.1 Growth of Data Centers and Energy Concerns, 1.2 Importance of Power Metrics, 1.3 Research Gaps, 1.4 Research Objectives)
2. 2. Research Methodology (2.1 Process of Literature Review, 2.2 Power Efficiency Metric in Data Centers, 2.3 Alternative Metrics to Measure Data Center Power Efficiency)
3. 3. Results and Discussion (3.1 Energy Usage and Efficiency Measurement in Data Centers, 3.2 The Role of Data Center Classification in Energy Efficiency Measurement: limitations of PUE without classification, evaluating passive cooling feasibility; 3.3 Thermal Environment in Data Centers, 3.4 Thermal Comfort in Computer Rooms)
4. 4. Conclusions

Related standards

• ASHRAE Thermal Guidelines for Data Processing Environments

Read the document

Publication details

Authors

Zaki Ghifari Muhamad Setyo; Hom Bahadur Rijal; Naja Aqilah; Norhayati Abdullah

Publisher

Energies (MDPI AG)

Year

2025

Volume

18

Issue

14

Pages

3689

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.3390/en18143689

Licence

CC BY 4.0 — share with attribution

Licence evidence

www.mdpi.com

Citation