As data centers contend with rising rack densities and increasing cooling loads, water consumption has become an increasingly important consideration in cooling-system selection. Yet water efficiency is often evaluated using one deceptively simple metric: gallons per minute at design conditions.

That number is useful, but incomplete.

A design-point figure represents water consumption near peak ambient conditions, typically around the hottest conditions for which the system is designed. It provides a straightforward way to compare equipment, but it does not answer the question that matters most for facility planning: How much water will this system consume over a full year at this specific site?

The answer depends heavily on climate.

Two Approaches to Adiabatic Cooling

Consider two common adiabatic heat-rejection architectures.

A once-through, mist-cooled system uses staged nozzles to introduce water into the entering airstream. As the water evaporates, it lowers the air temperature before it reaches the heat exchanger. The system can operate completely dry when ambient conditions allow, activating stages of adiabatic assistance only as temperatures rise.

A recirculating, media-cooled system passes incoming air through wetted evaporative media. Water is typically supplied from a basin and recirculated through the media, with a portion periodically discharged as blowdown to manage dissolved solids.

Both approaches can reduce water consumption compared with traditional evaporative heat rejection by half or more, according to Uptime Institute research. But comparing them only at peak conditions can obscure how differently they operate across thousands of hours each year.

What the Design Point Doesn't Show

In a recent NIMBUS analysis, the two architectures were modeled across six U.S. climates, Los Angeles, Miami, New York City, Chicago, Phoenix, and Atlanta, spanning marine, hot-humid, mixed-humid, cool-humid, hot-dry, and warm-humid climate zones at four cooling duty points.  Each site used its ASHRAE 0.4% design condition and ASHRAE's published climatic weather dataset (2017 edition), with each system evaluated using its actual hardware and operating characteristics rather than generic assumptions.

At each location's peak design condition, the once-through mist system used less water than the recirculating system in every scenario modeled. The median difference was 17%, ranging from approximately 5% in milder climates to nearly 50% in the hottest, driest climate evaluated.

That is a meaningful result, but it represents only a snapshot of system operation.

A year contains 8,760 hours, and most of them do not resemble the 0.4% design condition just described. Actual equipment, however, must operate across the entire range of dry-bulb and wet-bulb conditions a site experiences.

For an annual water budget, those other hours matter.

Climate Changes the Annual Equation

To capture them, the same six climates were evaluated using an hour-by-hour simulation. Each hour was assessed using site-specific dry-bulb and wet-bulb conditions, with water consumption calculated according to each system's operating logic.

Across every climate and duty point modeled, the once-through system consumed less water annually than the recirculating system. The median annual difference was 21%, ranging from 8% in milder climates to 41% in the hottest, driest location.

The reason comes down largely to operating hours and control strategy.

When conditions allow, a staged once-through system operates dry and consumes no adiabatic water. Water use begins only after ambient conditions cross the system's activation threshold and then increases in stages as additional cooling assistance is required. A recirculating system has different operating characteristics, including basin circulation and blowdown while operating in evaporative mode.

Across a full year, those differences accumulate; and climate determines how frequently each system operates in water-consuming modes.

Chicago and Atlanta illustrate the impact. At the same 750-ton duty point, the modeled once-through systems used identical hardware and had the same peak water draw: 11.5 gallons per minute. Based solely on that specification, their water requirements appear identical.

Annually, they were not.

The modeled Chicago installation consumed approximately 251,000 gallons per year, while Atlanta consumed approximately 443,000 gallons, about 76% more water from identical equipment. The difference was driven by operating hours. Atlanta experienced approximately 1,300 hours above the modeled adiabatic activation point, compared with roughly 775 hours in Chicago.

Same equipment. Same peak flow. Very different annual water budget.

Ask for the Number That Matters

For data center developers, operators and engineers evaluating adiabatic cooling, design-point water consumption remains an important specification. It simply should not be the only one.

Before selecting a heat-rejection architecture based on water consumption, ask vendors for two numbers: 1) design-point water draw and 2) modeled annual water consumption using hourly weather data for the actual project location.

The first indicates what the system may require near-peak conditions. The second provides a much clearer picture of what the facility is likely to consume, and pay for, over a year of operation.

That distinction is increasingly important as water availability influences operating costs, sustainability objectives, permitting and community conversations surrounding new data center development.

When building a water budget, equipment selection is only part of the equation. Climate determines how that equipment will actually operate.