Data Center Cooling Choices Shift Water Impacts Across the Grid: An Integrated Water-Energy Model for Sustainable Data Center Development

Overview

Data centres are being developed at an unprecedented pace, yet their energy and water impacts, and the spatial and temporal distribution of those impacts, remain poorly characterised. Existing assessment frameworks rely on facility efficiency metrics and average grid water intensity factors, suppressing the temporal impacts of data-centre load and generation availability. The model compares a water-cooled chiller, an air-cooled chiller and a hybrid airside economiser / water-cooled plant for a 200 MW IT facility in Michigan, and also looks at multi-gigawatt build-out siting cases.

For a fixed 200 MW IT load, moving from a water-cooled to an air-cooled chiller cut direct water by 97% (6.45 to 0.21 million m³/yr) and raised facility electricity by 32% (3.08 to 4.06 TWh). The same switch raised indirect water by 32% (4.48 to 5.91 million m³) but cut total water by 44% (10.93 to 6.12 million m³); the indirect share went from 41.0% to 96.6%. The hybrid airside economiser / water-cooled plant had the lowest electricity (2.76 TWh) and indirect water (3.98 million m³), with 3.20 million m³ direct water; its total of 7.19 million m³ was 34% below the water-cooled chiller.

Hybrid mode used the economiser only when daily mean temperature was at or below 27 °C and mean RH at or below 70%, which held for only 25% of the year in Michigan. Indirect water intensity varied by about 50% across months, depending on which generators supplied the added load. Higher electricity demand from air cooling raised depletion in subbasins away from the data-centre site, which facility-level reporting would miss.

The finding is that cooling configuration shifts water consumption from the facility to electricity generation, annual-average grid water factors hide large monthly variation, and water consumption magnitude alone cannot determine water impact. Limits: a single region (Michigan) and an open-data model; values depend on this grid mix and the cooling assumptions.

Key points

1. For a fixed 200 MW IT load, moving from a water-cooled to an air-cooled chiller cut direct water by 97% (6.45 to 0.21 million m³/yr) and raised facility electricity by 32% (3.08 to 4.06 TWh).
2. The same switch raised indirect water by 32% (4.48 to 5.91 million m³) but cut total water by 44% (10.93 to 6.12 million m³); the indirect share went from 41.0% to 96.6%.
3. The hybrid airside economiser / water-cooled plant had the lowest electricity (2.76 TWh) and indirect water (3.98 million m³), with 3.20 million m³ direct water; its total of 7.19 million m³ was 34% below the water-cooled chiller.
4. Hybrid mode used the economiser only when daily mean temperature was at or below 27 °C and mean RH at or below 70%, which held for only 25% of the year in Michigan.
5. Indirect water intensity varied by about 50% across months, depending on which generators supplied the added load.
6. Higher electricity demand from air cooling raised depletion in subbasins away from the data-centre site, which facility-level reporting would miss.

Using it in design

• In cooling selection, calculate direct and indirect water together: an air-cooled chiller cut direct water 97% but raised the indirect share to 96.6% and electricity by 32%.
• Assess water impact with monthly varying intensity (about 50% variation) rather than an annual-average grid water factor; which generators serve the load matters.
• Check the hybrid economiser's usage thresholds (27 °C, 70% RH) against project climate data; in Michigan they held for only 25% of the year.

Who should read it

Planners and policy specialists weighing direct and indirect water impacts in data-centre siting, cooling selection and water reporting.

Document contents

1. 1. Introduction
2. 2. Methods (IT-load representation, industry data: PUE and WUE metrics, grid operations and generator dispatch, generator water consumption)
3. 3. Case study (Michigan electricity system grid data)
4. 4. Results
5. 5. Discussion: water consumption magnitude alone cannot determine water impact
6. 6. Conclusion

Related standards

• ISO/IEC 30134-9 (WUE)
• ISO/IEC 30134-2 (PUE)

Read the document

Publication details

Authors

Garrett Alston; Nancy Love; Rabab Haider

Publisher

arXiv (University of Michigan preprint)

Year

2026

Document number

arXiv:2609.25437

Document language

English

Source type

Scientific

Access

Open licence — downloadable

DOI

10.48550/arXiv.2609.25437

Licence

CC BY 4.0 — share with attribution

Licence evidence

arxiv.org

Citation