Humidification and dehumidification control field guide for HVAC

Humidity control, and the two jobs it has to balance

Humidity control is keeping the water vapor in the air inside a target band, by adding moisture when the air is too dry and removing it when the air is too wet. Humidification is the adding. Dehumidification is the removing. In most commercial buildings one of those dominates by season, but in a tight critical space you can need both in the same week, sometimes the same day.

The mistake that runs through almost every bad humidity job is treating it as one number on one sensor. It is not. It is a moisture balance across a whole room, with people, outside air, equipment heat, and several pieces of conditioning gear all pushing the air in different directions. The sensor reads one spot. The room is not one spot.

Get the control variable right and most of the rest follows. Get it wrong, and you build a room where one unit humidifies while another dehumidifies, the energy bill climbs, and the space still drifts out of band. Most humidity callbacks are not a failed humidifier. They are a control strategy that was never commissioned and a sensor nobody calibrated.

What too-dry air does to a building and its electronics

Dry air is the quiet one. Nothing drips, nothing molds, so it gets ignored until something static-sensitive dies or the occupants start complaining. The first real cost is electrostatic discharge. As humidity falls, the air insulates better, charge builds on people and surfaces, and a body can carry several thousand volts before it ever feels a zap. That discharge into a circuit board, a memory module, or a drive can degrade or kill it, and the failure often shows up later as flaky behavior rather than a clean death you can trace.

Materials move too. Wood, paper, and adhesives give up moisture and shrink when the air dries out, which is why archives, museums, and print shops watch the low end as hard as the high end. Veneers check, paper curls, gaps open at joints.

Then there is the human side. Dry air pulls moisture from skin and airways, dries out the mucous membranes that catch airborne particles, and leaves people with scratchy throats and dry eyes that get blamed on everything except the air. None of this is dramatic on any single day. It is the slow tax of running a space drier than it should be, and in a data center the ESD piece is the one that writes the check.

What too-wet air does

Wet air fails louder. The headline risk is condensation: when a surface drops below the dew point of the surrounding air, water forms on it, and that water finds the worst possible places. A cold chilled-water pipe sweats onto a ceiling tile. A cold supply duct drips inside a chase. A cold metal surface in an electrical enclosure grows a film you never see until the corrosion does its work.

Corrosion is the slow version of the same problem. High humidity accelerates oxidation on connectors, contacts, and circuit traces, and in spaces with airborne contaminants the moisture lets that chemistry run faster. In data halls the specific worry is conductive dust and hygroscopic salts that pull water out of humid air and bridge across contacts. That is a soft fault that comes and goes with the humidity, which makes it miserable to chase.

Mold needs sustained high humidity and a food source, and it does not need standing water to start. Keep surfaces and pockets of a building above roughly 70 percent surface relative humidity for long enough and you have given mold what it wants. On the comfort side, humid air makes a space feel warmer than the thermostat reads and loads the cooling coil with latent heat it has to wring out. The wet end is where the visible damage lives, so it tends to get the attention, but both ends cost money.

Why control dew point instead of relative humidity?

Dew point is the temperature at which air becomes saturated and water starts to condense, and it is a direct measure of how much water vapor the air actually holds. Relative humidity is a ratio: how full the air is compared to how much it could hold at its current temperature. That distinction is the whole reason dew point is the better control variable in any space with temperature gradients, and a data center is nothing but temperature gradients.

Here is the trap with relative humidity. The same air can read very different RH numbers at different temperatures while carrying the exact same moisture. Air at a 55 degree F dew point reads one RH in a 75 degree F cold aisle and a much lower RH in an 85 degree F hot aisle, and the rule of thumb is that RH drops roughly 2 to 3 percent for every 1 degree F the temperature rises at a fixed dew point. So you can have a perfectly stable, correct moisture level and still watch the RH sensors disagree because they sit in air at different temperatures.

Control on RH in that room and the controller chases the temperature differences as if they were moisture problems, adding and removing water that did not need touching. Control on dew point and you are controlling the thing that actually matters, the absolute moisture, which is the same everywhere the air mixes. One ENERGY STAR data center that switched its humidification setpoint from RH to dew point cut humidifier runtime from about 80 percent of the time to about 20 percent. Same room, same hardware, better variable.

Dry-bulb temperature

The air temperature a standard thermometer reads, with no moisture effect

Wet-bulb temperature

The temperature read by a thermometer with a wetted wick, reflecting evaporative cooling and moisture together

Dew point

The temperature at which air saturates and condenses; a direct measure of absolute moisture

Relative humidity (RH)

Water vapor present as a percentage of what the air could hold at its current temperature

What humidity should a data center be?

The reference is the ASHRAE Technical Committee 9.9 thermal guidelines, whose 2021 fifth edition is the one most current specs point to. For the recommended envelope, the moisture window is a dew point of about minus 9 degrees C (roughly 16 degrees F) on the low end up to 15 degrees C (59 degrees F) on the high end, with an upper bound of 60 percent relative humidity. That recommended band is where ASHRAE suggests running for the best balance of reliability and energy.

The allowable envelope is wider, and it varies by equipment class, A1 through A4. The low-moisture limit for the allowable range is set as the greater of a minus 12 degrees C dew point and 8 percent RH, and the upper RH runs from 80 percent for the tighter classes up to 90 percent for A4. Below a crossover near 25 degrees C the dew point floor governs the low end, and above it the 8 percent RH governs, which is the practical reason the guidelines moved to a combined dew point and RH definition instead of an RH-only one.

The direction of travel matters as much as the numbers. Each edition has widened the bands and leaned harder on dew point, because tight RH control burns a lot of energy for protection that wider ESD practices already provide. Do not size or set a room from memory of an older edition. Confirm the equipment class, pull the envelope from the adopted edition of the guideline, and let the project specification override it where it is stricter.

Steam or evaporative: which humidification method?

There are two families, and they split on where the energy to evaporate the water comes from. Isothermal humidification boils water into steam and adds that steam to the air, so it adds moisture without dropping the air temperature. Adiabatic, or evaporative, humidification sprays or wicks water and lets the air itself supply the heat to evaporate it, which cools the air as it humidifies. Both take roughly 1,000 BTU to evaporate a pound of water. The difference is who pays that 1,000 BTU.

With steam, you pay it directly as electricity, gas, or boiler steam to make the steam. With adiabatic, the air pays it, so every pound of water added pulls about 1,000 BTU of heat out of the airstream, which can drop the air more than 20 degrees F across a high-output stage. That free cooling is the whole reason adiabatic systems exist, and it is why the electrical energy to run one can be under 10 percent of an equivalent-capacity steam system.

So the choice is rarely about whether one humidifies better. It is about the temperature effect and the water hygiene. Steam is clean, precise, and indifferent to air temperature, which is why it wins where you cannot tolerate a temperature swing or a wetting risk. Adiabatic wins where you want the evaporative cooling and have a climate and a water treatment program that justify it. Pick on the temperature consequence and the water, not on a humidity spec sheet.

The steam humidifier in the duct

An isothermal humidifier delivers steam into the airstream through a dispersion tube or a multi-tube manifold mounted across the duct. The steam has to fully absorb into the air before it reaches anything downstream. That distance is the absorption distance, and it is the number people miss. Put a tube too close to a coil, a filter, a turn, or a duct-mounted sensor and the steam condenses on it, which means wetting, corrosion, microbial growth, and a humidity reading that lies.

The non-wetting requirement drives the layout. You size the manifold for the airflow and the moisture load, you give the steam its absorption distance in a straight run, and you keep the high-limit sensor far enough downstream that it reads absorbed moisture rather than a wet plume. A short-coupled manifold in a tight mechanical room is the classic field compromise that shows up later as a rusted coil and a saturation alarm.

The steam source is its own decision. Electric resistance and electrode-boiler units are common and clean for general use. Gas-fired units cut operating cost where gas is cheap. Where the building has a central steam plant, a steam-to-steam exchanger makes clean humidification steam without dumping boiler chemicals into the breathing air. For a hospital, a lab, or a clean space, that clean-steam separation is often a spec requirement, not a preference. Confirm what the project allows in the breathing zone before you select the source.

Evaporative and adiabatic humidification

Adiabatic systems come in three common forms: wetted media that the air passes through, high-pressure nozzles that atomize water into a fine fog, and ultrasonic units that vibrate water into a mist. All three rely on the air to evaporate the droplets, all three cool the air while they humidify, and all three live or die on water quality.

The energy case is strong. Because the air supplies the evaporation heat, the parasitic electrical load is a fraction of a steam system, and the evaporative cooling can offset a real chunk of the cooling plant in a dry climate. In a data center makeup-air or in-AHU application, that cooling is a feature you size for, not a side effect. This is also where adiabatic humidification overlaps with the free-cooling logic on an economizer, so the two have to be commissioned to work together rather than against each other.

The catch is hygiene, and it is not optional. You are spraying water into air people breathe, so scale, biofilm, and Legionella risk are all on the table. High-pressure and ultrasonic systems need treated water, typically reverse-osmosis or deionized, to avoid spraying mineral dust into the space and to control scale. Wetted media needs a bleed and a dry-down cycle so it is not sitting wet and warm between calls. Stagnant warm water in a humidifier is a Legionella incubator, full stop, and the maintenance program that prevents it is part of the design, not an afterthought the owner discovers later.

Cooling coil or desiccant for dehumidification?

There are two ways to pull moisture out of air. The cooling-coil method chills the air below its dew point so water condenses on the coil and drains away, then usually reheats the now-cold air back to a usable supply temperature. The desiccant method passes the air over a material, a wheel or a liquid, that adsorbs water vapor directly, no condensation required. The split is about how low a dew point you need.

Coil dehumidification is the default and it is fine down to a point. The practical floor is around a 45 degree F dew point on a DX system, because below that the coil surface approaches freezing and ices up, and the economics fall apart. Reheat is also a horizontal move on the psychrometric chart, so it does nothing to the absolute moisture, it only resets the temperature. You are spending energy to cool past where you needed, then spending more to heat back.

Desiccant takes over where you need a dry supply. Because adsorption does not depend on a cold surface, a desiccant wheel can deliver a supply dew point well below the coil's surface temperature, reaching the 40 degree F dew point and lower that coils cannot economically hit. Vendor data commonly shows a desiccant stage adding a 5 to 15 degree F lower dew point and cutting dehumidification energy meaningfully versus a coil-plus-reheat approach for the deep-drying duty. Remember the load split: a cold coil pulls the latent load by overcooling the air, removing more sensible heat than comfort alone needs, which is exactly why the reheat that adds that sensible heat back is a penalty.

That reheat penalty is one of the most expensive things an HVAC system can do, because you pay twice: once for cooling past the comfort setpoint to condense water, once for heating the cold air back up. The energy code knows it. ASHRAE 90.1 and the mechanical energy codes restrict simultaneous heating and cooling and limit new-energy reheat, with carve-outs for spaces that genuinely need humidity control, so a humidity-driven reheat scheme has to be justified, not assumed.

The smarter approaches cut the penalty rather than pay it. Use recovered heat for the reheat, from condenser heat or a runaround loop, so it is free energy instead of new energy. Reset the cooling setpoint so you only overcool when the latent load actually demands it. Or move the deep-drying duty to a desiccant stage so the coil is not driven cold just to wring out the last of the moisture. The field failure is a reheat coil running against a cooling coil all year because a humidistat was set tight and nobody questioned it. The two coils fight, the meter spins, and the space sits in band the whole time, so no one notices until an energy audit does. Confirm the reheat is allowed under the adopted energy code for the space type.

The case against tight humidity control in a data center

The modern view, and where most new data center design has landed, is that tight relative-humidity control is usually unnecessary and wasteful. The expanded ASHRAE allowable bands let many sites run with little or no humidification across much of the year, and the low-end RH worry that drove decades of over-humidification has been softened by better understanding of ESD risk.

The ESD logic is the part operators have to actually believe. Dry air does raise static charge, but the defense is not a humid room. It is grounding: equipment rated and tested for ESD per IEC 61000-4-2, conductive or static-dissipative flooring, grounded wrist straps and personal grounding when anyone handles boards. With those in place, the data published in support of the wider bands shows the humidity-to-ESD relationship is not a simple line, and the low-humidity risk to installed, energized equipment is far smaller than the old setpoints assumed.

So the trade is real money against a small, manageable risk. Every percent of humidification you avoid is energy and water you do not spend, and in a large hall that adds up fast. The defensible setup is dew point control on the wider band, solid ESD discipline for hands-on work, and humidification reserved for the genuinely dry hours, not run as a constant. Where a tenant or an insurer demands a tighter band, that requirement governs, but make sure it is a requirement and not a habit.

Humidity sensors and calibration drift

Every humidity sensor drifts. That is not a defect, it is the nature of the measurement, and it is the single most common reason a humidity loop misbehaves. Capacitive RH sensors lose accuracy over time as the sensing polymer ages and as contaminants, dust, oils, and volatile organics in the air deposit on it. Quality probes spec on the order of 1 percent RH of drift per year in clean air, and dirtier or wetter environments push that faster.

Annual calibration is the floor, not the goal. Calibrate at least once a year, and step up to semi-annual or quarterly where the air is dusty, wet, or swinging hard, with an acceptable drift band often around plus or minus 2 percent RH before a sensor needs correction or replacement. A sensor reading 2 to 3 percent off does not look broken. It just quietly biases the whole control loop, so the room runs wetter or drier than the screen claims.

Placement is half the battle. A sensor in a dead pocket, in a hot plume off equipment, or right downstream of a humidifier manifold reads its microclimate, not the space, and the controller acts on that lie. In a multi-unit room, two sensors reading different microclimates is exactly how units end up fighting. Mount sensors in representative, well-mixed air, label the last-calibration date, and treat an uncalibrated humidity sensor controlling expensive equipment as the liability it is. The classic field symptom is a single bad sensor making its own unit fight every other unit in the room.

Why are my CRAC units fighting?

CRAC fighting is when one unit humidifies while another in the same room dehumidifies, both running hard, both spending energy, and the net effect on the room is close to zero. It is one of the largest single sources of wasted energy in a computer room, and it is almost always a control-strategy problem, not a hardware fault.

The mechanism is the RH trap from earlier. Put several CRAC units on the same RH setpoint and let them see return air at different temperatures, which they always do in a real room, and they compute different RH from the same actual moisture. The unit seeing warmer return reads lower RH and humidifies. The unit seeing cooler return reads higher RH and dehumidifies. Neither is wrong by its own sensor. They are both chasing temperature differences they have mistaken for moisture, and they can do it around the clock. Demand fighting like this can cut system efficiency on the order of 20 to 30 percent.

Two fixes work. The clean one is to switch the control variable to dew point so every unit is acting on the same absolute moisture, and the fighting has nothing to feed on. The pragmatic one, where the controls cannot do dew point, is to take humidity control away from most units and assign it to a few, or to a central humidifier and a separate dehumidification stage, so the room has one voice on moisture instead of a committee. If you want the airflow side of why the return temperatures differ in the first place, that traces back to air distribution and balance, which is a separate job.

This is also the per-unit versus central question. The cheapest-looking arrangement, a humidifier and a reheat coil in every CRAC, is the one most likely to fight, because each unit becomes an independent humidity controller acting on its own local sensor, and it multiplies the maintenance: every unit now has a cylinder or media, a water feed, and a sensor to keep honest. Centralizing the moisture control is usually the better answer in a multi-unit room. One humidifier on the makeup or supply air, sized for the whole room, sets the moisture once, and the CRAC or CRAH units handle sensible cooling and airflow without each one voting on humidity. Where per-unit humidification stays, the room controller has to coordinate it with dew point control, a single master humidity demand, and lockouts. Moisture is a room property, so it wants one control authority, not one per box.

Water quality for humidification

The water feeding a humidifier decides how much maintenance the system needs and, for evaporative types, what ends up in the air. Hard tap water carries dissolved minerals that drop out as scale when the water evaporates or boils. In a steam cylinder that scale coats the heaters and fouls the electrodes, shortening cylinder life and cutting output. In an evaporative system those minerals get atomized into the space as fine white dust that settles on everything, including the equipment you are protecting.

That is why high-pressure and ultrasonic humidifiers usually run on treated water, reverse-osmosis or deionized, to strip the minerals before they can scale or dust. Steam units tolerate harder water but pay for it in cylinder replacement and blowdown frequency. Either way the makeup water rate, the treatment, and the drain all have to be designed, not improvised at the unit.

Two field points carry most of the trouble. First, deionized water is aggressive and will corrode the wrong materials, so the wetted parts have to suit it, which is a selection detail people miss when they swap a water source. Second, any humidifier holding warm standing water between calls is a microbial risk, so the design needs a bleed, a dry-down, or a drain-down cycle so water is not sitting stagnant. Skip the water-treatment plan and the owner inherits scaled cylinders, dusted equipment, or a hygiene problem, and usually all three.

Condensation and keeping surfaces above dew point

Condensation is a surface problem, not an air problem. Air at a given dew point will not condense anywhere until it touches a surface colder than that dew point, and then it condenses there. So the whole game of avoiding sweating is keeping cold surfaces above the room air's dew point, or keeping the dew point low enough that your coldest surfaces stay dry.

The usual offenders are predictable. Chilled-water piping and cold supply ducts that lost their insulation or never had a proper vapor barrier. Cold metal in enclosures and equipment near a chilled supply. Exterior walls and glazing in humid weather. Raise the room dew point with aggressive humidification and you can start sweating these surfaces even though the RH on the screen looks unremarkable, because the screen is reading air, not the cold pipe in the chase.

This is the direct argument against over-humidifying a critical space. Every bit of moisture you add raises the dew point and shrinks the margin between the air and your coldest surface. In a data center with chilled-water distribution overhead, that margin is exactly what keeps water off the gear. Insulate and vapor-seal the cold surfaces, hold the dew point with room to spare below the coldest surface temperature you actually have, and check the cold spots with a surface thermometer rather than trusting the wall sensor.

Comfort, museum, archive, and healthcare humidity

Outside the data hall, humidity targets are set by people and materials, and they are looser than critical-space numbers but not unimportant. For general occupied comfort, ASHRAE Standard 55 frames the comfort zone, and a common practical band is roughly 30 to 60 percent RH, with the high end held down to keep the space from feeling sticky and to limit mold and dust-mite conditions. Ventilation rates that bring in outside air, governed by ASHRAE 62.1, also load the latent system, so comfort humidity and ventilation are tied together.

Some occupancies tighten the band hard. Museums, archives, and rare-collection storage hold both temperature and RH in a narrow window and, more importantly, limit how fast they are allowed to swing, because it is the cycling that cracks veneers and embrittles paper, not a steady offset. Healthcare spaces, operating rooms in particular, carry their own RH ranges in the governing health-facility guidelines, often a defined minimum and maximum tied to infection control and equipment.

The lesson that carries across all of them is the same as the data center one. Define the band, decide whether dew point or RH is the control variable for that space, and respect the rate-of-change limit where one exists. A gallery that meets its average RH while bouncing 10 points a day is failing the collection even though the logbook average looks fine. Pull the actual requirement from the project program or the governing guideline rather than reaching for a generic comfort number.

Commissioning humidity control

Commissioning humidity is where the design either becomes real or stays a drawing. Start with the setpoints and the control variable: confirm the loop is controlling the variable the design intended, dew point where dew point was specified, and that the setpoint and band in the controller match the spec, not a default left in from startup.

Then prove the sensors before you trust anything they say. Verify each humidity sensor against a calibrated reference at the actual condition, not a bench number, and confirm placement in representative air. A commissioning that signs off on an uncalibrated sensor has certified a guess. With the sensors honest, drive the system through its range and watch the response: call for humidification and confirm it adds moisture and recovers without overshoot or wetting at the manifold; call for dehumidification and confirm it pulls the dew point down and that any reheat is sequenced and sourced as designed.

The test that catches the expensive failure is the no-fighting check. In a multi-unit room, watch every unit at once and confirm that none is humidifying while another dehumidifies at the same condition. Verify the high-limit humidity sensor downstream of the humidifier trips and protects against saturation. Verify lockouts and staging do what the sequence claims. Document the as-left setpoints, the calibration results, and the response, because the owner cannot maintain a control strategy that was never written down.

The owner-side maintenance

Humidity control is maintenance-heavy, and the owner is the one who lives with whatever the design left them. Steam cylinders and electrodes scale and wear out, so they are consumables on a schedule tied to the water hardness and the runtime, not parts you replace when they fail. Evaporative media fouls and needs cleaning or replacement, and its bleed and dry-down cycles have to keep working or the hygiene risk creeps back in.

Water treatment is the piece that gets dropped first and costs the most when it does. Reverse-osmosis and deionization systems need their membranes and resin serviced, and a humidifier quietly running on degraded treated water scales up or starts dusting the space without throwing an obvious alarm. The drains, the bleed valves, and the makeup controls all need to stay clear and functional.

Sensor calibration belongs on the same maintenance calendar as the filters, at least annually and more often in dirty air. The owner who calibrates humidity sensors on a real cadence keeps the control honest for the cost of a morning. The one who never does ends up running the room out of band, fighting units, or both, and paying for it on the meter the whole time. None of this is exotic. It is a calendar and the discipline to keep it.

What to document

Hand a humidity loop to the next technician with no commissioning record and they cannot maintain it, let alone explain why the room sits where it does. The record is what tells the next technician what the band is, what variable controls it, when the sensors were last trusted, and why the room is set the way it is.

For each controlled zone, capture the setpoint and band, the control variable, the humidification and dehumidification method serving it, the last sensor calibration and result, and the design dew point limits the space has to stay inside. Tie it to the spec or guideline the band came from, so a future change is a decision, not a guess. The table below is the minimum a commissioning record should leave behind.

Common mistakes

• Controlling tight relative humidity in a room with temperature gradients instead of controlling dew point.
• Leaving every CRAC on the same RH setpoint so units fight, one humidifying while another dehumidifies.
• Running humidity sensors for years without calibration and trusting a reading that has drifted out of band.
• Over-humidifying a critical space until cold pipes, ducts, and surfaces sweat below the room dew point.
• Overcooling to dehumidify and then reheating with new energy where the code restricts it and recovered heat was available.
• Skipping water treatment and hygiene on evaporative systems, inviting scale, mineral dust, and Legionella risk.
• Short-coupling a steam manifold so the steam wets a coil, a sensor, or a turn before it can absorb.

Field checklist

Standards and references

For data centers, the reference is ASHRAE Technical Committee 9.9 and its thermal guidelines for data processing environments, which define the recommended and allowable temperature and humidity envelopes by equipment class. Use the edition the project and the equipment manufacturer point to, currently the 2021 fifth edition for most new work, and confirm the class before reading a number off it. The envelope has widened across editions, so an old number is a wrong number.

For comfort and ventilation, ASHRAE Standard 55 frames the thermal comfort zone including humidity, and ASHRAE 62.1 governs ventilation rates that load the latent system. For energy, ASHRAE 90.1 and the adopted mechanical energy code restrict simultaneous heating and cooling and limit new-energy reheat, which is what governs a humidity-driven reheat scheme. Healthcare and other specialized occupancies carry their own governing guidelines with their own RH ranges.

Beyond the design standards, the equipment manufacturer's installation requirements, absorption distances, water-quality limits, and the project specification all control the actual install, and where they are stricter than a guideline, they win. For water hygiene on evaporative systems, follow the governing Legionella risk-management guidance for the jurisdiction. Confirm the adopted editions and any local amendments before citing a specific number on a submittal.

Units, terms, and conversions

Humidity shows up in several units across a drawing set, a controller, and an equipment sheet, and mixing them is how setpoints get entered wrong. Relative humidity is a percent referenced to the air's current temperature. Dew point is a temperature, in degrees F or C, and it is the absolute-moisture variable to control on. Grains of moisture per pound of dry air, and the metric grams per kilogram, are the absolute humidity ratio you see on a psychrometric chart and in load calculations.

Two conversions are worth carrying. Dew point and RH only convert to each other once you also know the dry-bulb temperature, which is the entire reason RH alone is ambiguous. And a humidifier's output is rated in pounds per hour of moisture (or kilograms per hour), which you match to the room's moisture load, not to a humidity percentage.

Isothermal humidification

Adding moisture as steam, with no drop in air temperature; energy comes from boiling the water

Adiabatic humidification

Evaporating water using the air's own heat, which cools the air as it humidifies

Latent vs sensible load

Latent is the energy in moisture; sensible is the energy in temperature. Humidity control is the latent side

Absorption distance

The duct length a steam plume needs to fully evaporate before reaching a surface or sensor

Grains per pound (gr/lb)

Absolute moisture: grains of water vapor per pound of dry air; the chart and load-calc unit

CRAC / CRAH

Computer room air conditioner (DX) and computer room air handler (chilled water), the room units that often carry humidity control

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is air balancing?
• What is an air-side economizer?
• What humidity should a data center be?
• What is the ASHRAE recommended temperature for a data center?
• What is delta-T in a data center?