Data center adiabatic and evaporative cooling field guide

Adiabatic and evaporative cooling, and the water-for-energy bet

Evaporative cooling is using the evaporation of water to cool air, or to cool the condenser that rejects a data center's heat, by driving the temperature down toward the wet-bulb instead of the dry-bulb. Adiabatic cooling is the same physics aimed at one specific job: pre-cooling the air entering an air-cooled condenser or a dry cooler so it rejects more heat on a hot day. Both spend water to save compressor work.

That is the whole bet in one sentence. The compressor is the most expensive part of mechanical cooling, and anything that lets it run less or not at all cuts the cooling energy hard. Evaporation gives you a way to reach a colder effective temperature without paying the compressor to do it. In a dry climate that bet pays well. In a humid one it pays little, because the air cannot take up much more water.

The catch is that you pay in water instead of energy, and that has become the central argument in data center design. Evaporative cooling drives the energy metric, PUE, down while it drives the water metric, WUE, up. Whether that is the right trade depends on the climate, the cost and availability of water and power at the site, and what the community around the site can spare. The free-cooling guide covers the economizer side of this, and the cooling-systems overview maps where evaporative rejection fits among the families.

How does evaporation cool the air?

Evaporation cools because turning liquid water into vapor takes energy, and that energy comes out of the air as heat. Each pound of water that evaporates absorbs roughly a thousand BTU of latent heat, pulling it from the surrounding air and dropping the air temperature. No compressor, no refrigerant. Just a phase change doing the work.

The limit on how far the air cools is the wet-bulb temperature, not the dry-bulb. Dry-bulb is the temperature a plain thermometer reads. Wet-bulb is what a thermometer reads with a wet wick over the bulb, so evaporation is already cooling it, and it represents the coldest the air can get by evaporation alone. The drier the air, the bigger the gap between dry-bulb and wet-bulb, and the more cooling evaporation can deliver. On a 100°F desert afternoon at low humidity, the wet-bulb might sit near 65°F, so evaporation has a 35°F window to work in. On a 90°F humid Gulf afternoon, the wet-bulb can be 78°F, leaving almost nothing.

No real device reaches the wet-bulb exactly. How close it gets is the effectiveness of the media or the approach of the tower, and it is always short of the theoretical floor. That floor is set by the weather, not by the equipment, and that is the single fact that decides whether evaporative cooling is worth building on a given site.

Dry-bulb

Air temperature read by a standard thermometer, before any evaporation

Wet-bulb

The lowest temperature reachable by evaporation alone, set by the air's temperature and humidity

Latent heat

The energy absorbed when water changes from liquid to vapor, roughly 1000 BTU per pound

What is the difference between direct and indirect evaporative cooling?

Direct evaporative cooling sprays or wicks water straight into the supply air, so the air leaves cooler and wetter, with moisture added. Indirect evaporative cooling evaporates water on one side of a heat exchanger and cools the IT air on the other side, so the supply air leaves cooler and dry, with no moisture added. That difference decides which one belongs in a data hall.

Direct evap is simple and effective and is common in warehouses and shops where humidity does not matter. In a data center it does matter. The supply air carries the added moisture straight to the racks, and a hall has a humidity band it has to stay inside per the ASHRAE thermal guidelines. Push direct evap too hard and you raise the dew point at the equipment, which is exactly the variable the white space is trying to control.

Indirect evap, often built as an indirect-direct evaporative cooling unit, IDEC, solves that by keeping the two air streams apart. Outdoor scavenger air gets wetted and runs across an air-to-air heat exchanger. The IT air runs across the dry side, gives up its heat, and never touches the water. The IT air stays at the moisture it started with, so the hall's humidity stays where the design put it. For most data center work the rule is short: if the moisture reaches the IT air, you have the wrong system. Indirect keeps the IT air dry, and that is why it dominates large evaporative installations.

Adiabatic pre-cooling on the condenser and dry cooler

The most common use of evaporation in a modern data center is not cooling the IT air at all. It is pre-cooling the air going into an air-cooled condenser or a dry cooler so the equipment rejects more heat on the hottest days. This is adiabatic pre-cooling, and it is a capacity boost you turn on only when the weather forces it.

An air-cooled chiller or a dry cooler rejects heat to outdoor air, and its capacity falls as that air gets hotter. On a design-day afternoon the unit can run short of the heat it has to dump. Spray a fine mist into the intake air, or run that air through a wetted pad, and the air entering the coil drops several degrees toward the wet-bulb. The condenser sees cooler air, the head pressure falls, the compressor works less, and the rated capacity comes back. The same trick lets a dry cooler hold a colder supply on a hot day, extending the hours it can carry the load without a chiller.

The reason this approach wins is restraint. The unit runs dry most of the year and spends water only on the handful of hot hours that need it. You get the dry-cooler water profile most of the time and the evaporative capacity exactly when the weather takes it away. That is the pattern behind most hybrid plants, and it is the one that keeps annual water use low while protecting the design-day rejection.

The evaporative types in data center use

Four shapes of evaporative cooling show up in data centers, and they differ in where the water meets the air and whether the IT air gets wet. Naming the type tells you both the water exposure and the humidity risk.

The cooling tower is evaporative by nature: it rejects the heat from a chilled-water or condenser-water loop by evaporating a slice of the circulating water, and it is the workhorse of large water-cooled plants. Indirect evaporative units cool the IT air through a heat exchanger without wetting it. Direct evaporative air handlers wet the supply air and are used where the hall can tolerate the added moisture, which is rare for IT space. Adiabatic pre-coolers sit on the intake of air-cooled condensers and dry coolers and only run on hot days. The cooling-systems overview places these against the air and liquid families; this guide stays on the evaporative mechanism itself.

Why is the wet-bulb temperature the floor?

The wet-bulb is the floor because it is the coldest evaporation can reach. Evaporation cools by adding water vapor to air, and once the air is saturated it cannot take up more, so the temperature stops falling at the wet-bulb. Every evaporative device, tower, pad, or spray, is chasing that number and lands somewhere above it.

How far above is the approach. A cooling tower's approach is the gap between the cold water it produces and the ambient wet-bulb, and a smaller approach means a better, usually larger, tower. The same idea on the air side is the effectiveness of the pad or the exchanger, the fraction of the dry-bulb-to-wet-bulb gap the unit actually captures. Neither one beats the wet-bulb, and you size the equipment for the design wet-bulb at the site, not the dry-bulb.

This is where evaporative cooling lives or dies, and it is climate, not catalog. Pull the design wet-bulb from the ASHRAE climatic data for the location and check it against the supply temperature the hall needs. If the design wet-bulb plus a realistic approach lands above the temperature the racks require, evaporative cooling alone cannot hold the hall on the worst days, and you need a chiller for the trim. The free-cooling guide works the same wet-bulb math from the economizer-hours angle.

Does evaporative cooling use a lot of water?

Yes, evaporative cooling consumes water by design, because evaporation is how it works. The number to watch is water usage effectiveness, WUE, the liters of water used per kilowatt-hour of IT energy, a metric The Green Grid defined to sit alongside PUE. A heavily evaporative site can run well above a liter and a half per kilowatt-hour over a year; a dry-cooled or chilled site with little evaporation runs far lower. The water goes three ways: evaporation that does the cooling, drift carried off as droplets, and blowdown dumped to keep the dissolved solids from concentrating.

Here is the trade that defines the whole debate. The fastest way to cut PUE is to lean on evaporation, which spikes WUE. The fastest way to cut WUE is to run dry with air-cooled equipment, which raises energy use and PUE. Power and water pull against each other, and a site has to pick its blend based on what power and water cost there, what the grid's own water footprint is, and what the local supply can give up.

So treat PUE and WUE as one decision, not two. A design that brags about a low PUE while quietly running a high WUE has not solved the problem, it has moved it from the meter to the watershed. Report both, hold both to a target the site can actually defend, and remember that the numbers shift with the climate and the design, so the right blend in Phoenix is not the right blend in Dublin.

The water debate and the community

Data center water use has moved from a back-of-house engineering detail to a public fight, and any evaporative design now ships with that scrutiny attached. Reporting through 2025 and 2026 has put hard numbers on it: a large share of planned US data centers sit in areas that have seen drought, and the AI build-out is concentrating new load in places where water is already stretched.

The pushback is real and it has teeth. Communities have protested permits, and at least one government temporarily pulled a major project's authorization after public pressure over water. For an evaporative design that means the WUE number is no longer just an efficiency figure. It is a permitting and reputational fact that an owner has to be ready to defend with where the water comes from, whether it is potable or reclaimed, and what happens to the watershed in a dry year.

The engineering responses are the ones in this guide: lean on indirect and adiabatic so water is spent only when the weather demands it, recover and reuse water where the site allows, and reach for non-potable or reclaimed supply so the facility is not drinking from the municipal tap a town depends on. None of that makes evaporation free of water. It makes the water spend defensible, which on a contested site is the difference between a project that gets built and one that does not.

Water treatment, scale, and blowdown

Evaporative systems leave their dissolved minerals behind. Water evaporates pure and the salts stay in the loop, so the concentration climbs every cycle. Without treatment that ends in scale on the heat-transfer surfaces, corrosion of the metal, and biological growth in the warm wet basin. A treatment program is not optional on an evaporative system; it is the thing that keeps the system working and safe.

The lever is cycles of concentration, how many times the makeup water is concentrated before it is dumped. You measure conductivity or total dissolved solids and bleed off a slice of the loop, the blowdown, to hold the concentration in range, then add fresh makeup. Run too few cycles and you waste water on excess blowdown. Run too many and you scale the fill and the exchanger. The water chemistry of the makeup sets where that balance sits, which is why the treatment regime is site-specific and gets dialed in during commissioning, not copied from another plant.

Treatment also means chemistry: scale and corrosion inhibitors, and a biocide program to keep the biology down. That biology is not just an efficiency issue. It is the Legionella issue, which the next section treats on its own because it is a health hazard, not a maintenance line item.

Is evaporative cooling a Legionella risk?

Yes. Any system that evaporates water and throws a mist, cooling towers and evaporative condensers especially, can grow Legionella bacteria in the warm water and aerosolize it into the air around the building, where people can breathe it in. This is a documented public-health hazard, and it is the reason an evaporative system carries a water management program rather than just a maintenance checklist.

The framework is ASHRAE Standard 188, which calls for a documented water management program for buildings with cooling towers: a written plan, defined control points, scheduled water-chemistry and Legionella testing, physical inspection of the tower components, and records that prove it all happened. The treatment program from the previous section is the chemistry side; the 188 program is the management system that makes sure the chemistry, the testing, and the cleaning actually get done on a schedule and get documented.

Drift eliminators are the physical control that keeps the aerosol down, and they are covered next. The blunt version: an evaporative system without a real, documented Legionella program is a liability, not a cooling system. Treat the program as part of the equipment, fund it, and keep the records, because when a case is investigated the records are what get pulled first.

Drift, carryover, and drift eliminators

Drift is the water that leaves an evaporative system as fine droplets carried off in the air stream, separate from the vapor that does the cooling. It matters for two reasons. The droplets are lost water, and on a tower or evaporative condenser they are the main way Legionella gets out into the air, because a droplet can carry bacteria where pure vapor cannot.

Drift eliminators are the baffles that catch those droplets and drain them back before the air leaves the unit. A good set cuts drift to a tiny fraction of the circulating water, on the order of thousandths of a percent, and that is both a water-saving and a public-health number. Damaged, fouled, or displaced eliminators are a direct failure: more water lost and more aerosol released. Inspecting them is a line item in the 188 program for exactly that reason.

Carryover on the air side is the direct-evap version of the same problem. If a direct evaporative section is overdriven or its media is failing, liquid water can be carried into the supply air as droplets rather than absorbed as vapor. In a data hall that means water on the IT air, which is why direct evap needs the media sized and the air velocity held in range, and why indirect designs sidestep the issue entirely by keeping the wet side away from the IT air.

Pad, spray, and high-pressure fog media

The evaporative media is how water gets surface area to evaporate from, and the three common forms trade cost against water control and maintenance. The choice shapes both performance and the upkeep burden.

Cellulose pads are corrugated sheets the air passes through while water trickles down them. They are cheap, give high effectiveness, and are forgiving, but they hold water, foul with dust and minerals, and become a biological surface that has to be cleaned and eventually replaced. Spray systems atomize water into the air stream through nozzles, which lets you stage and meter water finely, at the cost of nozzles that clog and a pump that has to be maintained. High-pressure fog pushes water through fine nozzles at high pressure to make a mist that evaporates almost completely in the air, giving precise control and low carryover when it is set up right, but it is the most sensitive to water quality and nozzle fouling and demands the cleanest water of the three.

Water quality drives the maintenance on all of them. Hard or dirty water scales pads, plugs nozzles, and shortens the interval between cleanings, which is one more reason the treatment program and the media choice get decided together, not in isolation.

Hybrid: dry cooler plus adiabatic plus mechanical trim

The hybrid plant is the design that wins on most modern projects because it spends water only when it has to. The shape is a dry cooler or air-cooled unit as the base, an adiabatic stage for the hot hours, and a mechanical stage, DX or a chiller, as the trim for the days the weather beats both.

It runs in stages that follow the weather. When it is cold, the unit runs fully dry, rejecting heat to ambient air with no water and no compressor, which is the free-cooling region the economizer guide covers. As the day warms past the point where dry rejection falls short, the adiabatic stage wets the intake and buys back capacity with a little water and still no compressor. Only on the hottest hours, when the wet-bulb is too high for evaporation to carry the load alone, does the mechanical trim kick in.

That sequencing is the point. Most of the year the plant uses no water and little energy. Water gets spent on a small slice of hours, and the compressor on a smaller slice still. You get a low PUE and a controlled WUE from the same machine, which is why the dry-cooler-plus-adiabatic-plus-trim arrangement has become the default answer when a site has to balance both meters.

Does evaporative cooling work in humid climates?

It works far better in dry climates than humid ones, because the gap between dry-bulb and wet-bulb is what evaporation has to work with, and humidity closes that gap. In a hot dry climate the wet-bulb stays low even when the dry-bulb is high, so evaporation reaches deep and a site can run mostly evaporative with little compressor help. In a warm humid climate the wet-bulb sits close to the dry-bulb, evaporation buys you little, and you lean on mechanical cooling instead.

This is why evaporative-heavy designs cluster in dry and high-desert regions and why the same design moved to a humid coast underperforms its brochure. The hours where evaporation alone carries the load track the local wet-bulb across the year, and those hours are what justify the water plumbing and the treatment program in the first place. A site with few low-wet-bulb hours pays the water and maintenance cost of an evaporative system without getting the energy payback that makes it worth it.

Run the bin analysis for the actual location before committing. Count the annual hours below the wet-bulb your design needs, weigh the energy those hours save against the water they cost, and let the site's climate, not a vendor's headline number from a different region, pick the system. Climate decides this one, and it decides it differently in every market.

The energy payback and low PUE

Where the climate cooperates, the energy payback is the reason evaporative cooling exists. Taking the compressor out of the loop for most of the year is the single largest cut available to a data center's non-IT energy, and evaporation is what extends the no-compressor hours past where plain dry rejection runs out.

An economizer can hold the hall on outdoor air when it is cold. Evaporation widens that window into the warmer hours by reaching below the dry-bulb toward the wet-bulb, so a site that would need the chiller at a 75°F dry-bulb afternoon can ride on evaporation instead. Across a year in a favorable climate that turns into a large share of hours with the compressor off and a low annual PUE, which is the headline that sells these systems.

Keep the framing honest, though. The low PUE is real and it is earned by spending water, so it is half of a two-meter result. The free-cooling guide covers how the economizer and the changeover are controlled to capture those hours; this guide's job is to remember that the energy win and the water cost are the same decision seen from two sides.

Controls and the water-saving mode

The controls decide when water gets spent, and a sloppy sequence is where water budgets quietly blow up. The core rule is to wet the system only when the wet-bulb makes it worth it and to stage the water rather than dumping it all at once.

A good sequence reads the wet-bulb, not just the dry-bulb, because the wet-bulb is what sets the available cooling. It runs the plant dry as long as dry rejection can hold the load, brings on the evaporative or adiabatic stage only as the load approaches the dry limit, and modulates the water to hold the supply setpoint instead of running full spray. Many modern units offer a water-saving or dry-priority mode that biases toward energy at the cost of a little water, or the reverse, so the operator can lean the plant toward whichever resource is tighter that season.

Stage it and meter it, and the same hardware can run lean. Skip the staging and the plant wets early, wets hard, and posts a WUE nobody can explain at the year-end review. The control sequence is where the energy-versus-water blend actually gets set, so it belongs in the design intent and gets proven in commissioning, not left to a default the vendor shipped.

AI density: evaporation, liquid cooling, and water

AI racks have sharpened the water question because their density pushes the industry toward liquid cooling, and liquid cooling changes where evaporation fits. Direct-to-chip and other liquid loops can deliver both a low PUE and a low WUE, which breaks the old energy-versus-water tradeoff at the rack. That has made liquid the fast-growing answer for high-density AI halls.

Liquid cooling at the chip still has to reject its heat somewhere, though, and that heat rejection is often still air-side or evaporative outside the building. So evaporation does not disappear with liquid cooling; it moves to the rejection plant, where the same WUE-versus-PUE and Legionella questions apply. The cooling-systems overview covers how density picks the capture method; the point here is that even an AI hall on liquid usually leans on evaporative or hybrid rejection for the hot hours, so the water program in this guide still applies.

Commissioning an evaporative system

Commissioning proves three things on an evaporative system: that it actually delivers the cooling the design claimed, that the water program is real and documented, and that the staging spends water the way the sequence intended. Skip any of the three and the system runs wrong quietly, usually wetter and more expensive than the design promised.

On the performance side, verify the evaporative effectiveness or the tower approach against the design wet-bulb, ideally on a warm day when the system is actually working, not on a mild day when nothing is loaded. Confirm the supply temperature holds across the staging points and that the changeover from dry to wet to mechanical happens at the wet-bulb thresholds in the sequence, not earlier. On the water side, confirm the treatment system, the cycles of concentration, the blowdown control, and the 188 water management program are in place and documented before the system carries load, not as a punch-list item after.

The number that gets faked is WUE. A plant can post a clean PUE in commissioning and a quietly high WUE because the staging was never tuned and the unit wets early. Trend both meters through a range of weather during commissioning, confirm the water-saving controls do what they claim, and hand over a system whose water spend matches the design intent, because that is the number the site will have to defend later.

Maintenance and winterization

Evaporative systems are wet, warm, and exposed, which makes them maintenance-heavy in a way dry equipment is not. The upkeep falls into four buckets, and the system degrades fast if any one is neglected.

The media comes first: clean and replace pads on schedule, clear and replace fouled nozzles, and keep the fog or spray system producing the droplet size it was designed for. Water treatment is continuous: hold the cycles of concentration, keep the chemistry and biocide dosing on program, and keep the testing and records current under the 188 plan. Drift eliminators get inspected and repaired, because damaged ones lose water and release aerosol. And in any climate that freezes, the system has to be winterized, with basins heated or drained and dry-operation provisions confirmed, because a frozen basin or a burst line takes the plant out exactly when a cold snap should be giving you free cooling.

The thing operators let slide is the testing cadence on the water program, because nothing visibly breaks when a sample gets skipped. It breaks later, as scale, as a corrosion leak, or as a Legionella finding, and by then the cheap, scheduled fix is no longer available.

What to document

An evaporative system is judged on two records: that it cools as designed and that the water is managed. Capture both, by method, so a reviewer or an inspector can reconstruct the decision and the program.

Record the cooling method and where the water meets the air, the design wet-bulb and the effectiveness or approach the unit was commissioned to, the staging thresholds, the treatment regime and target cycles of concentration, the blowdown control, the 188 water management program and its testing schedule, and the annual WUE alongside the PUE. If the design chose indirect over direct to protect the hall humidity, write down why, because the next person will wonder about the extra heat exchanger.

Common mistakes

• Using direct evaporative cooling on the IT air where humidity matters, instead of indirect, and pushing the hall dew point out of the ASHRAE band.
• Reporting a low PUE while ignoring the WUE, so the water cost never gets weighed against the energy savings.
• Running an evaporative system with no Legionella program and no documented ASHRAE 188 water management plan.
• Wetting the system when the wet-bulb is too high to gain anything, spending water for no cooling benefit.
• Leaving drift eliminators damaged or fouled, so the system loses water and releases aerosol.
• Skipping winterization in a freezing climate and losing the plant to a frozen basin or a burst line.
• Running no staging or water-saving controls, so the plant wets early and posts a WUE nobody can explain.

Field checklist

Standards and references

The ASHRAE TC 9.9 thermal guidelines set the temperature and humidity envelope the IT air has to stay inside, which is the constraint that pushes data centers toward indirect over direct evaporative cooling. ASHRAE also publishes the climatic design data, including the design wet-bulb you size evaporative equipment to, and the equipment chapters that cover cooling towers and evaporative devices. Pull the design conditions for the actual site and confirm them against the project basis of design.

Water usage effectiveness, WUE, comes from The Green Grid, which defined it in a white paper as the water companion to PUE. Treat both as targets the design holds, not external mandates, and read them together. For Legionella, ASHRAE Standard 188 is the controlling document, requiring a documented water management program for buildings with cooling towers; many jurisdictions adopt or reference it, and local public-health rules can be stricter, so confirm what the AHJ requires.

Equipment performance, the effectiveness curves, the drift rate, the recommended water chemistry and cycles of concentration, comes from the manufacturer's data for the unit you actually buy, and it governs the specifics. Uptime Institute and industry reporting frame the water-and-energy tradeoff at the operations level. Cite the standard that controls the point, hold WUE, PUE, and the wet-bulb to the climate and the design rather than to a single headline number, and verify the adopted edition before relying on a section.

Units, terms, and conversions

Evaporative cooling spans a few metrics and unit systems, so the same idea reads differently across a design narrative, a vendor sheet, and an operations report.

WUE is usually liters of water per kilowatt-hour of IT energy, and some US sources use gallons; one liter is about 0.264 gallons. PUE is dimensionless, total facility energy divided by IT energy. Wet-bulb and dry-bulb appear in °F on US drawings and °C elsewhere; the conversion is °C times 9/5 plus 32. Approach and effectiveness both describe how close a unit gets to the wet-bulb, one as a temperature gap and one as a percentage. Cycles of concentration is a unitless ratio of the makeup chemistry to the loop chemistry.

WUE

Water usage effectiveness, water used per unit of IT energy, the water companion to PUE

PUE

Power usage effectiveness, total facility energy divided by IT energy

Approach

The gap between the temperature a unit produces and the ambient wet-bulb

Cycles of concentration

How many times makeup water is concentrated before blowdown dumps it

Blowdown

Water bled from the loop to hold dissolved solids in range

Drift

Water carried off as droplets, a water loss and the main Legionella aerosol path

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is free cooling in a data center?
• How are data centers cooled?
• What is WUE?
• How does an evaporative cooler work?
• What is a CDU in a data center?