PAU explained: why outdoor air gets a machine of its own
The air outside a Singapore office holds more water than a room unit can strip out on top of its own job. Larger buildings treat that outdoor air separately, before it reaches a fan coil. Delete the step and the damp arrives with nobody's name on it.
By Team Snowflake | Updated 8 Aug 2026
The job a pre-treatment unit does before the air arrives
A pre-treatment unit conditions outdoor air on its own, before that air meets anything in the occupied space. NEA's ACMV handbook lists it as a primary air handling unit, written PAU where it appears on paperwork, and gives its duty as cooling and dehumidifying outdoor fresh air. Nothing else passes through it.
Feeding it undiluted outdoor air is the reason it exists as a separate box. The same handbook notes that a PAU handles an airstream made up entirely of outdoor air, which carries a lot of moisture. Its coil therefore works on the wettest air in the building, and on nothing else.
The conventional alternative is to mix outdoor air into a much larger stream of return air and send the blend across one coil. That arrangement works against itself. Outdoor air of high moisture content joins a large volume of return air of low moisture content, and because the outdoor share is the smaller one, the blend comes out relatively dry. The handbook states the consequence plainly: the potential to dehumidify that mixed air drops, which is what pushes designs toward colder water and deeper coils.
So the separate machine buys undiluted air to work on. Water is easier to take out of air that is still carrying it. Blend outdoor air into a room's own recirculated stream and the concentration that made it removable has already thinned.
Why a fan coil cannot absorb it on the way past
A fan coil is not refusing the work. Its control loop is answering a different question. A technical paper on outdoor air systems in Singapore describes the units downstream of a PAU as reacting to internal heat loads instead of to the humidity of the airstream passing through them. They stop when the room reaches temperature, and whatever water is still in that air stays in the room.
Capacity is the second limit, and it hides better than the first. NEA's handbook shows two coils with identical total cooling capacity removing very different proportions of sensible and latent heat. It concludes that such a system can hold the preset space temperature while missing the desired relative humidity. A capacity figure being correct on a schedule says nothing about which half of the job that capacity does.
Both limits point the same way. Cooling and drying are related but they are not the same duty, and a box selected against a room's heat gain was never sized against the water arriving from outside it.
Why does outdoor air get counted as a separate load?
Two loads land in every conditioned space, and they originate in different places. The room makes one of them through its occupants, its lighting, its equipment and the sun on its glass. Outdoor air brings the other one in with it, and that share arrives whether the room is busy or empty.
NEA's handbook sets out how a chilled water design divides the two. Fan coil units are described as mainly carrying the sensible heat load of the spaces. A central primary air handling unit cools and dehumidifies the outdoor fresh air and delivers it through ducts. That conditioned air supplies the ventilation and carries away the moisture from the latent load of the spaces.
Read that division slowly and the second machine stops looking like duplication. One box holds a temperature inside a room. The other governs the moisture content of the air entering it. Neither can stand in for the other, because they are not attempts at the same outcome.
The wider literature names the arrangement. A dedicated outdoor air system is defined as two parallel systems. One is dedicated to the ventilation air and handles both the latent and sensible load of conditioning it. The second handles the mostly sensible loads generated indoors. Singapore practice reaches the same shape through the PAU.
Where pre-treatment is present, the handbook says the downstream coils are then designed to handle mainly sensible cooling and a relatively small latent load. That sentence is the part worth carrying away. The terminal equipment on a fit-out was selected on the assumption that the water had already been taken out upstream. Remove the assumption and every selection below it is wrong, while nothing on the schedule looks wrong.
| What generates it | Where it is meant to be handled | What shows up when it is not |
|---|---|---|
| What generates itHeat given off by people and equipment | Where it is meant to be handledThe unit serving that room | What shows up when it is notThe room drifts warm once it fills up |
| What generates itSun landing on glass through the afternoon | Where it is meant to be handledThe unit serving that room | What shows up when it is notOne side of the floor never quite settles |
| What generates itWater riding in with the outdoor air | Where it is meant to be handledA pre-treatment unit, upstream of both | What shows up when it is notAir that reads cold and still feels sticky |
| What generates itWater given off by the occupants themselves | Where it is meant to be handledWhichever coil is doing the drying | What shows up when it is notGrilles and glazing collecting droplets |
| What generates itVentilation air that arrives untreated | Where it is meant to be handledNothing was ever assigned to it | What shows up when it is notDamp that tracks headcount, not weather |
Why the second machine is not a spare
A tenant looking at two pieces of cooling equipment on one drawing usually reads redundancy into it. Redundancy means a duplicate capability held back in reserve. These two overlap in almost nothing.
Failure behaviour separates them cleanly, and it is the test to apply. Lose the units in the space and the room warms up while the air coming into it stays dry. Lose the pre-treatment unit and the room can hold its temperature perfectly while every surface in it turns damp. Two unlike complaints, two unlike machines, and only one of them reachable from inside the tenancy.
Where the unit sits, and what it hands its air to
A pre-treatment unit sits upstream of the space and never inside it. Everything it handles comes from outdoors, so it needs a duct to a louvre or an intake. It also needs floor area for a filter stage, a coil, and a fan strong enough to push treated air out to wherever it is owed.
Delivery goes one of two ways, and the handbook names both. Cold, dehumidified air leaving a PAU can be supplied straight to the conditioned spaces, or it can be fed into the inlet of the air handling units. Which of the two is in use decides where a shortfall becomes visible.
Straight delivery gives the treated air its own diffusers, running in parallel with whatever the room unit is blowing. Feeding an intake hides it completely. The outdoor air is already inside the supply air before it leaves the plant, and an occupant sees a single set of openings with no way to tell what is in them.
Premises carrying a large ventilation requirement are where these units concentrate. NEA's handbook offers hotel guest rooms as the example: many small spaces, each owed outdoor air, each with a terminal unit doing nothing beyond holding a temperature. Offices with dense occupancy, clinics and function rooms sit in the same category for the same reason.
None of this vocabulary describes a household split system. A flat's indoor unit takes room air, chills it and gives it back, with nothing upstream treating anything. Pre-treatment belongs to buildings where somebody was obliged to bring outdoor air in, and where the size of that obligation made it a machine of its own.
Telling treated air from recirculated air at the ceiling
Opening count is the first clue, and it needs no drawing. A space fed only by its own unit has a supply and a return belonging to the same box, and the air runs in a closed loop between them. A space also receiving pre-treated air has a supply opening with no matching return, because that air came from outdoors and leaves through an extract or through the building itself.
Behaviour once the room unit stops is the confirmation. Where treated outdoor air is delivered straight to a space, something keeps blowing after the terminal unit shuts down, often on the building's own schedule and not on the wall controller. Where everything goes quiet together, the space has no independent supply and whatever outdoor air it gets is arriving by some other route.
What a space does when the provision falls short
Undersizing or deleting a pre-treatment unit produces no fault anyone can point at. It produces a building that holds its temperature and cannot hold its moisture, and the complaints that follow get attributed to almost anything except the machine nobody built.
The failure path is documented. A Singapore technical paper describes engineers assuming the cooling downstream of a PAU will finish dehumidifying the outdoor air. It states that this is rarely what happens, because those units answer to heat and not to the moisture in front of them. It then follows the consequence through. When they ramp up on a temperature rise, the cold air they deliver takes the space air below the dew point. Condensation in the room follows, along with the mould problems that come with it.
What an occupier reports is never a shortfall on an outdoor air unit. It is a space that reads cool on a thermometer and sticky to everyone standing in it. It is skirting or gypsum staying damp at the perimeter long after anyone has stopped looking for a leak. It is a stack of small complaints that never point at a machine.
The thermostat is the wrong lever in both directions, which is worth settling before an office starts arguing about it. Set it lower and the terminal units run longer, pulling out more water while also bringing surfaces closer to the temperature where condensation forms. Set it higher to save energy and they run less, so more of the incoming water stays behind. Neither setting reaches the air arriving from outside.
Push back on a scope that answers a damp complaint by cleaning the units in the space. Servicing returns a coil to the duty it was selected for, and on a fit-out with pre-treatment in the design, that duty was mostly temperature. A quote written as though a clean coil will resolve the moisture has priced the wrong machine.
Complaints that get filed under the wrong machine
These arrive as separate tickets, raised by different people, usually weeks apart. Seen together they describe one condition, and the common thread is that each one is about water rather than about heat.
- A meeting room that turns clammy once it fills and recovers on its own when it empties.
- Droplets on diffuser faces and on the ceiling tiles immediately around them.
- A musty smell returning soon after every clean, strongest near the supply openings.
- Paper, labels and card stock that refuse to lie flat in one part of the floor.
- Mould appearing on the cool side of a partition while the warm side stays clear.
Questions that pin down the air arriving in a space
Establish whether anything treats outdoor air before it reaches the space, and establish it before a cooling scope is signed. Where a mechanical drawing exists the answer is on it. Where none exists, it is a short conversation with whoever manages the building, and the answer is either a named unit or silence.
Ownership follows immediately from the answer. A pre-treatment unit almost never sits inside a tenancy. It serves several occupiers at once through shared ductwork, so its filters, its coil and its controls belong to the building. An occupier with a moisture problem is therefore often reporting a condition on plant that is not theirs to open. Any scope written from inside the leased area will price around it without saying so.
Ask any contractor pricing a humidity complaint which machine the water is coming off. If every line in the answer names equipment inside the leased area, nothing in that quote has accounted for the air entering it, and the work will be paid for twice.
Three things are answerable on site with no paperwork at all. Whether any ceiling opening blows without a return grille anywhere near it. Whether the damp tracks headcount or tracks the weather outside. Whether neighbours on the same floor started noticing it at the same time. None of those name the machine, but together they say whether the problem stops at the tenancy boundary.
- Whether any unit outside the leased area was assigned to treat outdoor air for this floor.
- Which ceiling opening, if any, that unit's air actually arrives through.
- Who maintains it, under what arrangement, and where that arrangement stops.
- Whether the damp complaint follows occupancy or follows the weather.
- What the space was drawn to be used for, and what it is being used for now.
| What to ask | What a clear answer sounds like | Why it changes the scope |
|---|---|---|
| What to askIs outdoor air treated before it reaches this floor | What a clear answer sounds likeA named unit, with its own tag and duct | Why it changes the scopeDecides whether the moisture has an owner |
| What to askWhere does that unit's air enter the space | What a clear answer sounds likeIts own diffusers, or another unit's intake | Why it changes the scopeSets where a shortfall becomes visible |
| What to askWho holds the upkeep on it | What a clear answer sounds likeThe building, under its own arrangement | Why it changes the scopePuts those filters outside a tenancy scope |
| What to askWhat was it selected to deliver | What a clear answer sounds likeA duty stated on the mechanical schedule | Why it changes the scopeShows what the fit-out was allowed to add |
| What to askWhat has changed here since handover | What a clear answer sounds likeHeadcount, partitions, or a different use | Why it changes the scopeLoad added later sits outside the design |
Why the damp often shows up long after handover
Fit-outs change what a space does, and the outdoor air provision was sized against what the drawing said it would do. Turning a store room into a meeting room, or a quiet back office into a call floor, raises occupancy without anyone touching a single machine.
Occupancy is the variable that moves the water. More people means more outdoor air owed and more moisture given off inside, and neither of those reaches the cooling equipment as a signal it can act on. The gear is still delivering what it was selected for. What changed is the room it was selected against, and that mismatch surfaces as damp instead of heat.
Ready to get started?
Tell us what’s going on. Symptoms, setup, photos, anything we should know. We’ll assess and come back with the right next step.