Coanda effect: why aircon air travels along the ceiling
Air leaving a wall unit does not fall straight into the room. It runs along the ceiling first, and the far side of the room only gets cooled because of that. Break the run and the same machine covers half a space.
By Team Snowflake | Updated 9 Aug 2026
The ceiling is part of the air path
Air leaving an indoor unit meets the ceiling and stays on it. A stream running close to a surface lowers the pressure in the narrow gap between the two, and the higher pressure in the room presses the stream flat against that surface. The ASHRAE Fundamentals handbook defines the effect as a moving jet attaching to a parallel surface because of negative pressure developed between jet and surface.
Attachment is what carries cold air to the opposite end of a room. A stream moving through open space drags room air into itself from every direction, which slows it down and warms it up. Pin that same stream against a ceiling and it can only draw on the open face. Titus states the mechanism directly: air on the surface side can no longer mix in, so the amount of induction falls.
The size of that difference is the part worth carrying out of this. Three engineering references converge on the same figure, putting a free stream at roughly seventy percent of the reach of an attached one. ASHRAE converts catalogued attached-jet throw to free-jet throw by multiplying by 0.707. Price Industries and Nailor both publish the working version: cut the catalogue figure by about thirty percent once the outlet sits more than 2 ft from a surface.
So the ceiling performs a job here, and every published reach figure assumes it does. A far corner gets served along the ceiling, not across the middle of the room. That single fact reorders which questions deserve asking when one end of a space stays warm.
Why the effect is invisible from the floor
Nothing about a running unit shows the stream travelling. The air is clear, it moves above head height, and the only thing anyone notices is where it eventually lands on them. Households therefore build a mental picture of cold air pouring out and sinking, which is close to how it behaves once attachment ends and nothing like how it behaves before that.
The picture matters because it decides what gets blamed. A room understood as one box of air points at capacity. A room understood as a stream with a path points at the path, and the path is cheaper to inspect than a compressor.
Three conditions decide whether the stream holds
Attachment has entry requirements, and installation settles all three. The outlet has to sit close to the surface. It has to discharge at a shallow angle to it. Air has to leave fast enough to create the pressure difference in the first place.
Distance carries published thresholds. Titus places surface effect where a side wall outlet sits within a foot of the ceiling. Price Industries treats an outlet as free-standing once it is more than 2 ft from any surface, and Nailor gives the same threshold and the same reduction. Between those two figures the hold weakens by degrees, so this behaves as a slope and never as a switch.
Angle is the condition a household can move and rarely connects to reach. Titus puts the limit under 40 degrees between the jet and the surface for circular pattern diffusers, and somewhat less for other jet types. Past that angle the stream leaves on a trajectory of its own and the ceiling has no grip on it. Louvres pointed steeply into the room sit well outside the limit.
Velocity belongs on the list because the flow has to generate its own pressure difference. Price Industries words the condition plainly: the low pressure area forms when supply air velocity is sufficiently high. Take the fan down to its quietest setting and the same outlet on the same wall produces a weaker hold and a shorter run.
Wall units in Singapore flats satisfy all three by convention instead of by design. High on the wall, blowing roughly horizontal, at a fan speed set for cooling. The arrangement that gets chosen for pipe length and a tidy trunking run happens to be the arrangement attachment needs.
| Set at installation | What it does to the hold | What a household can still move |
|---|---|---|
| Set at installationHeight of the outlet below the ceiling | What it does to the holdWithin a foot holds; past 2 ft the data treats it as open air | What a household can still moveNothing, short of fresh pipework |
| Set at installationAngle the louvre sits at | What it does to the holdUnder 40 degrees attaches, steeper does not | What a household can still moveThe vane setting, on most units |
| Set at installationFan speed at the outlet | What it does to the holdThe pressure difference has to be created by the flow | What a household can still moveThe speed setting, at the cost of noise |
| Set at installationWhether the ceiling continues past the unit | What it does to the holdA stream can only follow a surface that is there | What a household can still moveWhat sits on the ceiling, never the ceiling itself |
| Set at installationSupply temperature against the room | What it does to the holdColder air is denser, so it lets go earlier | What a household can still moveThe setting, within a narrow band |
Where the stream lets go
Attachment ends somewhere in every room, and the location of that point is the whole question. Cold air is denser than the air around it, so gravity pulls on the stream the entire time it runs. The pressure difference at the ceiling holds it up against that pull. Separation happens where buoyancy finally wins.
ASHRAE handles the balance through the Archimedes number, the ratio that governs the trajectory of a non-isothermal jet. It weighs buoyant force against the momentum the air carried out of the outlet. The handbook gives a correlation for separation distance built from the outlet conditions themselves.
One practical consequence of that balance runs against instinct. Colder supply air separates sooner, so a lower setting can shorten reach instead of extending it. Nailor publishes a rule of thumb for the size of the shift. Horizontal throw moves by roughly one percent per degree Fahrenheit of difference between supply and room air, quoted against one fixed reference velocity.
Past the separation point a room is on its own. Price Industries describes what forms there. Where outlet velocity is insufficient or the outlet is poorly located, a stagnant layer of room air develops, and above that layer proper heat transfer stops happening. Temperature then stratifies across the space. That stagnant layer is the warm corner, and running the unit longer does not clear it.
This is where a capacity quote goes wrong, and it goes wrong quietly. A room whose stream separates early has a delivery problem visible on its own ceiling. More capacity lifts the flow somewhat and pushes the separation point somewhat further, at a price set as though the entire room were underserved. Ask which part of the space stays warm, and ask it before agreeing to anything larger.
Why the room can be right on paper and still split in two
A heat load calculation counts what the room gains. Floor area, glazing, orientation, occupancy and the equipment left switched on all feed into it, and none of them describe a surface the air has to travel along. The result is a correct capacity figure attached to a delivery assumption nobody wrote down.
So a flat can hold two true statements at once. The unit is sized properly for the space, and one corner of that space never gets served. Those statements only look contradictory while the room is imagined as a single body of air.
Ceilings that stop before the room does
Every published reach figure assumes a flat surface continuing in the direction the air is aimed. Singapore flats often break that assumption. Pipe runs, service risers and structural depth get boxed into bulkheads, and a bulkhead frequently lands on the same stretch of ceiling the stream needs.
A bulkhead partway along the run removes the surface mid-flight. The stream reaches the step, finds nothing within the distance the engineering data assumes, and behaves from there as free air with whatever momentum survived. That final stretch was already the slowest and coldest part of the run, which is why the loss shows up so clearly at the far wall.
False ceilings change the picture more quietly. Dropping a ceiling over one zone shortens the gap between unit and surface in that zone and lengthens it in the next. Where the carpentry arrives after the aircon, nobody re-measures the mounting height against the new soffit, and the fade that follows gets blamed on the machine.
Storage and cornices work from below with the same result. A run of shelving that reaches within a hand's width of the ceiling puts a wall across the stream at exactly the height it travels. Waist-high furniture sits harmlessly underneath and never touches it, which is why height matters far more than bulk here.
Format decides the starting condition and nothing later can undo it. A wall unit begins horizontal and near the ceiling, which is the arrangement the effect requires. A cassette sits flush in the ceiling and discharges outward along it on several sides, applying the same principle from the middle of the space. A unit aimed straight down has no surface alongside it and behaves as free air from the first inch.
| What sits in the path | What happens to the hold | How it reads from the floor |
|---|---|---|
| What sits in the pathA bulkhead boxing in pipes or a riser | What happens to the holdThe surface stops, so the stream separates at the step | How it reads from the floorCold under the unit, warm past the boxed section |
| What sits in the pathA false ceiling added after the aircon | What happens to the holdThe gap to the surface changed and went unchecked | How it reads from the floorOne zone fell behind after the renovation |
| What sits in the pathShelving or a cornice near ceiling height | What happens to the holdA wall stands across the stream at its travelling height | How it reads from the floorThe far end went warm when the carpentry went in |
| What sits in the pathA structural beam crossing the room | What happens to the holdThe run gets interrupted partway along | How it reads from the floorHalf a long room cools and the other half lags |
| What sits in the pathLouvres set steeply into the room | What happens to the holdThe discharge angle sits past the point that holds | How it reads from the floorA cold patch below the unit, warm elsewhere |
Decided at the wall, traded at the remote
Commercial design checks this on paper and domestic installation almost never does. Price Industries sets out the standard practice: plot the corrected throw onto the reflected ceiling plan, then check it for interference with obstructions, walls and other air jets. A flat gets a tape measure and a decision about pipe length.
One question at the quotation recovers most of the gap. Ask what the ceiling does between the proposed mounting point and the far end of the room. Anybody who has placed units in local flats answers immediately, since bulkheads and false ceilings are the two things they already work around daily. An answer that jumps to horsepower has skipped the surface entirely.
Draught-free comfort modes deserve reading as a trade instead of a feature. Spreading discharge across a large perforated face lowers velocity at every point on that face. Price Industries names the consequence: spreading a pattern increases entrainment, cuts the mass flow per unit area, and reduces throw. Comfort close to the unit gets bought with reach at the opposite end.
The trade is often the right one in a small bedroom, where nobody needs distance and somebody is sleeping under the outlet. What makes it a problem is meeting it by surprise. A household that switches the mode on, then finds the far corner warm a week later, has run into a designed behaviour and starts hunting for a fault instead.
Push back on any quote that reaches for gas, capacity or a chemical wash before somebody has looked up. Checking what the ceiling does costs one walk across the room. Where a corner used to cool and stopped, the ceiling is not the new variable and the machine deserves the attention. Where that corner has never worked since the day of handover, the surface was short from the beginning.
The note worth keeping from handover
Write down two things when the unit goes in. How far below the ceiling the outlet sits, and what stands on the ceiling between the unit and the furthest wall it has to serve. Both take seconds to record while somebody is standing there with a ladder.
Those two lines settle an argument years later that otherwise costs a diagnostic visit. A room that never covered its far end has a geometry answer and always did. A room that covered it and stopped has a machine answer, and one line of notes separates them without guesswork.
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.