Rated vs actual cooling capacity: the lab and the room
A rating is a measurement, taken once, under conditions a laboratory keeps still. It exists so machines can be ranked against each other, not so a bedroom can be predicted. Those two jobs are not the same, and the difference costs people money.
By Team Snowflake | Updated 5 Aug 2026
What a rating measures, and what it was built for
A rated cooling capacity is the result of a measurement, not a specification the machine is afterwards held to. A sample of the model goes into a test chamber. The air on both sides of it is brought to fixed values and kept there. Whatever heat the machine removes while those values hold becomes the number that gets printed.
Keeping both air conditions still is the whole reason the test exists. Two machines put through identical fixed conditions can be ranked honestly, because the only thing that differed between the runs was the equipment. Let every maker choose its own conditions and the published figures stop meaning anything against one another.
Comparability is genuine value rather than a marketing convenience. Each domestic model on the local market arrives with a figure produced by one shared method, so two boxes on a shelf can be ranked on output without trusting either badge. Nothing else printed on either box does that job.
The limitation lives inside the same sentence as the strength. A figure produced under conditions held still describes a machine working under conditions held still. A bedroom in a Singapore flat keeps nothing still. The number travels perfectly from one machine to another, and it travels badly into a home.
A test run also hands the machine every advantage available. Coils are clean, the charge is correct, air moves across both of them without obstruction, and the two halves sit close together. Nothing inside a chamber is dusty, boxed in, or stretched down a corridor ceiling. The published figure describes the machine on the best day it will ever have.
Where the rated figure still earns its keep
Ranking two models is what this number is for, and it does that job well. A model rated higher will out-cool a model rated lower in the same room, on the same afternoon, in the same state of repair. That ordering survives even though neither machine will reach its printed figure.
Checking somebody else's arithmetic is the second honest use. A sizing exercise produces a required output, a quote proposes a model, and the rated figure is what joins the two together. That check only works because the measurement was standardised. The BTU guide covers what the unit itself measures and why an hour is attached to it, and the cooling kW and power kW guide separates the capacity number from the electrical one printed beside it.
What sits between the chamber and the flat
Delivered output drops below the rating whenever a real installation differs from the test setup, and every real installation differs somewhere. The differences are not exotic. They are the ordinary conditions of an occupied flat, and each one takes a slice off the top of what the machine can do.
Outdoor air above the reference condition takes the largest slice in this climate. A condenser has to push heat into air warmer than the test assumed, so each unit of heat costs more effort to move. The ambient temperature guide sets out that relationship and the time-of-day pattern it produces, so this page treats it as established.
Moisture in the return air takes the second slice, and it takes it invisibly. Part of the rated total goes into pulling water out of the air rather than into lowering its temperature, and a thermometer cannot see the difference. The heat load guide works through the mechanism. The split between the two is not fixed, so one machine surrenders a different share of its rating in a damp room than in a dry one.
Airflow across the indoor coil decides how much heat each pass can collect. A loaded filter, a return grille pushed up against a wardrobe, a fan left on its quiet setting, or a duct longer than the fan was chosen for all cut the volume crossing the fins. Less air over the coil means less heat collected, whatever the compressor is doing behind it.
Dirt on a coil surface is insulation in the one place nobody wants it. Heat has to cross from air into metal at the indoor end, then out of metal into air at the outdoor end, and a film of dust, grease or biofilm slows both transfers. Fouling is also the only entry on this list that worsens on its own.
Pipe length and vertical rise cost a smaller amount, and they cost it permanently. Refrigerant travelling further gives up pressure along the way and picks up heat through imperfect lagging, which is why manufacturers publish correction figures against run length. A run sitting near the published limit is legal, workable, and quietly down on output for as long as the system stays in place.
| How the flat differs from the chamber | What the room gives up | Was it there from the first day |
|---|---|---|
| How the flat differs from the chamberIntake air above the reference condition | What the room gives upOutput slides as the ledge warms up | Was it there from the first dayYes, if the position was always hot |
| How the flat differs from the chamberHumid air arriving at the indoor coil | What the room gives upPart of the rating goes into drying | Was it there from the first dayYes, this is simply the local climate |
| How the flat differs from the chamberRestricted air path across the fins | What the room gives upLess heat collected on every pass | Was it there from the first daySometimes, since filters load up gradually |
| How the flat differs from the chamberDirt sitting on either coil surface | What the room gives upSlower transfer at both ends at once | Was it there from the first dayNo, this one builds up over time |
| How the flat differs from the chamberRefrigerant pipe near its published limit | What the room gives upA small standing loss that never lifts | Was it there from the first dayYes, fixed on the day of installation |
The slices stack, and the room feels the total
Read one at a time, none of these sounds like the difference between comfort and a complaint. Slightly less air over the fins. A ledge running warmer than the test assumed. A coil that has not been cleaned this cycle. Each one reads as tolerable, which is precisely why each one gets tolerated.
They do not take turns. A hot afternoon delivers warm intake air at the condenser, humid return air at the evaporator, and a room full of people and screens, all within the same hour. The conditions that hurt output are driven by the same weather, so they peak together instead of averaging one another out.
Each loss also applies to whatever the previous one left, not to the printed figure. The machine is not subtracting five separate deductions from its rating. It runs at whatever the conditions permit, and what reaches the room is what survived the entire chain. No single item on the list explains the result, which is why fixing them one at a time rarely closes the gap.
Here is where the arithmetic stops being academic. A system chosen to match a calculated room load, with that match made against rated output, has spent its whole margin on paper before anyone switched it on. Put it into a west-facing bedroom on a fouled coil in the middle of a hot spell and it is genuinely marginal. Nothing about it has failed.
The household reads marginal as faulty, and that reading is reasonable. A machine unable to hold the temperature it has been set to looks broken from the sofa. What follows is a call, an expectation that something will be found, and a supplier under quiet pressure to find it.
Treat with care any part quoted against a shortfall that has been present from commissioning. A component that never worked properly would have declared itself long before this. Ask which room load the capacity was chosen against, and what outdoor condition that sizing assumed. A supplier who can answer neither question is proposing a repair for a purchase decision.
Why physics gets blamed on the machine
A unit behaving exactly as physics dictates and a unit with a genuine fault produce the same sentence from the household: this thing is not cooling properly. The room offers no way to tell them apart, because the room only reports the outcome.
That is worth knowing before a warranty claim gets raised. A system delivering less than its plate figure in a hot, humid, occupied flat is not defective, and a manufacturer will say so. The gap belongs to the conditions and to the sizing decision, and neither of those is covered by anybody's warranty.
Buying to the minimum adequate rating
A sizing exercise returns a required output, and the obvious move is to buy the nearest model at or just above it. That move is where the margin disappears. The requirement was worked out against rated output, so matching it exactly buys a machine that is adequate under laboratory conditions and nowhere else.
The honest counter-argument is that buying bigger carries costs of its own. A machine holding too much capacity reaches the set temperature quickly, shuts down, and leaves the air damp behind it. The heat load guide explains why that happens. So the answer is not to buy the largest model that will fit the wall.
The useful reframing is that margin has to come from somewhere, and capacity is the most expensive place to buy it. A condenser lifted out of a boxed-in yard recovers output that no upsize can purchase back. A clear return path, clean fins and a pipe run kept well inside the published limit each hand back part of the rating at no monthly cost.
Spend on capacity only once the free margin has been collected. Buying a larger machine to cover a badly placed condenser pays for the same problem twice, at installation and then on every bill afterwards. The bigger machine still breathes the same hot air off the same ledge.
Where a real margin is warranted, ask for it in writing. A quote naming the room load it was sized against, and the outdoor condition it assumed, is a quote that can be argued with. A quote naming only a model cannot be. It matters most where the room was already difficult before anyone chose a model.
What a modulating system changes, and what it does not
An inverter alters the shape of this decision without removing it. A modulating compressor spends most of its running life below rated output, so headroom carries a smaller running penalty than it would on a fixed-speed machine. Capacity bought as insurance is cheaper to own than it used to be.
It changes nothing about the two limits that matter. At the bottom of its range a heavily oversized system still leaves a room cool and damp, and at the top it still cannot exceed what the conditions allow. Surplus capacity is not a substitute for a sensible condenser position, and no control strategy makes hot intake air cooler.
Reading the gap: designed in, or opened since
One question sorts most of these cases, and the household is the only party able to answer it. Has this room ever held the temperature it is set to through a hot afternoon. Not lately. Ever.
A room that never coped from the outset is a sizing, placement or installation conversation. That gap was designed in before the first switch-on, and nothing inside the machine will be found at fault, because nothing inside it has moved. Stripping and washing a unit in that situation buys a marginal improvement and leaves the cause exactly where it was.
A room that used to hold it and now does not is a condition conversation. Something has changed since the system was commissioned, and the shortlist is genuinely short: fouling on either coil, a charge that has drifted down, a fan that has lost speed, or a filter nobody has taken out. Working that list in order is the productive route.
The gradual middle is where these two get confused. Fouling accumulates rather than arriving, so a system that started out just adequate can slide into inadequate with no event to point at. Holding one hot day beside a comparable one shows whether the decline was a step or a slide. A step points at a component. A slide points at a surface.
The distinction decides who should be called and what should be quoted. A designed-in gap is answered by relocating a condenser, clearing a return path, or revisiting a capacity decision deliberately. A gap that opened is answered by finding what moved. Treating the first as the second produces a repair that fixes nothing, and treating the second as the first buys an upsize that hides a fault behind more machine.
Write the history down before anybody visits. When the system went in, which room it serves, whether the first hot spell was comfortable, and what has been cleaned or replaced since. A technician holding that is answering a narrow question. A technician without it is guessing at which of two very different conversations this is.
| What the history says | Where the gap came from | What gets examined first |
|---|---|---|
| What the history saysNever coped on a hot afternoon, right from the start | Where the gap came fromSizing, condenser position, or the pipe run | What gets examined firstThe room load the capacity was chosen against |
| What the history saysCoped once, stopped abruptly, with a clear before and after | Where the gap came fromSomething inside the system gave way | What gets examined firstElectrical and refrigerant-side readings under load |
| What the history saysCoped once, then faded away without any single event | Where the gap came fromSomething accumulating on the transfer surfaces | What gets examined firstCoil and filter condition at both ends |
| What the history saysEvery room copes except one particular room | Where the gap came fromThat room's own load, or its share of the air | What gets examined firstGrille, duct and what the room itself generates |
| What the history saysCopes until the flat fills up with people | Where the gap came fromThe load rose, while the capacity did not | What gets examined firstOccupancy, and what runs inside the room |
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.