As artificial intelligence drives unprecedented growth in data center construction, discussions about sustainability often focus on electrical consumption. While power remains a critical consideration, water availability is becoming an equally important constraint in many regions.
Large-scale data centers can consume significant amounts of water through cooling towers and evaporative cooling systems. In water-constrained regions, the ability to reduce or eliminate cooling water consumption may become a competitive advantage for developers, owners, and operators.
Fortunately, advances in dry cooling, hybrid cooling strategies, immersion cooling, and thermal storage provide practical opportunities to dramatically reduce water consumption without sacrificing reliability.
Historically, many data centers have relied on cooling towers and evaporative cooling because they offer strong thermal performance and can improve overall facility efficiency. The downside is water consumption.
As data center development expands into regions already facing water stress, owners are increasingly evaluating Water Usage Effectiveness (WUE) alongside traditional metrics such as Power Usage Effectiveness (PUE).
Key questions include:
One promising approach is the use of dry coolers as the primary heat rejection system, combined with limited water-assisted operation during extreme ambient conditions.
Under normal operating conditions, the dry coolers reject heat directly to the outdoor air without consuming water. During the hottest hours of the year, a water-assisted mode can provide additional cooling capacity.
This strategy changes the role of water in the cooling system. Instead of relying on continuous evaporative cooling, water becomes a supplemental resource used only when needed.
Potential benefits include:
Many cooling systems are designed around a single peak design condition. However, the hottest few hours of the year often drive substantial increases in equipment size, infrastructure cost, and water consumption.
A more effective approach is to evaluate hourly and seasonal climate data to understand how often water assistance is truly required.
Important considerations include:
For some facilities, water-assisted cooling may only be required during a small fraction of annual operating hours. By reviewing the seasonal design approach temperatures, owners may be able to avoid water use entirely during off-peak months and significantly reduce water use during shoulder seasons.
The effectiveness of dry cooling depends heavily on location. A cooling strategy that works well in Toronto may not be appropriate in Phoenix, Houston, or Miami.
Cooler climates may allow dry coolers to operate without water assistance for much of the year. Hot and dry climates may benefit from nighttime operation, thermal storage, and limited water assist. Hot and humid climates require careful evaluation because high wet-bulb temperatures can reduce the effectiveness of evaporative strategies.
Every project should be evaluated using local hourly weather data rather than generic assumptions.
Thermal storage is often discussed as an energy cost reduction strategy, but it can also play an important role in reducing water consumption.
Many data centers experience their most challenging cooling conditions during hot afternoon periods when ambient temperatures are highest, utility demand charges may be elevated, dry cooler performance is reduced, and water-assisted cooling is most likely to be required.
Thermal storage allows cooling to be produced during nighttime hours when outdoor conditions are more favorable and chiller operation may be more efficient. That stored cooling capacity can then be discharged during peak daytime conditions.
Potential benefits include:
Perhaps most importantly, thermal storage can reduce the number of hours that dry coolers require water assistance. Instead of sizing the entire system around the hottest hour of the year, stored cooling capacity can shave the daytime peak load and allow the facility to operate dry for a larger percentage of annual hours.
As rack densities increase, immersion cooling and other liquid cooling strategies are becoming more relevant.
Unlike traditional air-cooled data halls, immersion cooling systems may be able to operate at higher fluid temperatures. This can create an opportunity to reject heat directly through dry coolers without relying on chilled water systems or evaporative cooling during many operating hours.
Potential benefits include:
For high-density AI workloads, the combination of immersion cooling and dry coolers may become one of the most effective paths toward low-water data center cooling.
Reducing water consumption is rarely free. Dry coolers often require more fan energy than cooling towers, and water-saving strategies may increase electrical consumption under certain conditions.
The goal should not be to optimize a single metric in isolation. Owners should evaluate the full system, including:
The best cooling strategy often balances water use, energy use, cost, resilience, and long-term operational flexibility.
When evaluating a data center cooling strategy, owners should ask:
Data center cooling decisions have long-term implications for water use, energy cost, reliability, and project feasibility. Small changes in design assumptions can significantly affect annual water consumption and operating cost.
An independent technical review can help owners evaluate:
If you are evaluating cooling options for a new or existing data center, I provide independent technical reviews focused on improving cooling efficiency, reducing water consumption, and identifying practical opportunities for long-term operational savings.
Contact Fassbender Energy Advisory to discuss your data center cooling strategy.
```
Copyright © 2010-2024 CosmoLogic LLC. TRACE 700 and eQUEST are ™ of Trane Inc. and James J. Hirsch respectively. Energy-Models.com is built in San Francisco, CA and Slinger, WI USA.