Guide · heating sizing
In summer the difference is invisible. In winter it isn’t. An air conditioner’s heating output falls together with the outdoor temperature, and the catalogue “kW” never shows it. Using real curves we explain why sizing for heating “by eye” can leave you with a cold house at the worst possible moment — and what to calculate instead.
When a unit only has to cool, a sizing mistake is forgivable. Slightly oversized — a bit more electricity. Undersized — on a heatwave it won’t quite reach the set point, but no harm is done. The margin for error is wide, so sizing “by eye” usually gets away with it.
Heating is different. The same machine that cools in summer runs in reverse in winter — and here “by eye” can leave you with a cold house at the worst possible moment: at −15 °C. The reason is single and simple:
An air conditioner’s heating output falls as the outdoor temperature drops. The colder it is, the less it heats — and the catalogue “kW” doesn’t show it.
It does. Except that number is quoted for mild conditions — usually at an outdoor temperature of +7 °C. That is the “good-day” output. The problem is that a house needs heat most not at +7 °C, but at −10, −15, −20 °C — and then the unit delivers far less than the headline claims.
Look at the real curve of an actual unit. Call it Device A — a single-split of the 2.5 kW catalogue class. One column is the outdoor temperature, the other is how much heat it actually delivers:
| Outdoor temperature | Heating output (kW) | COP |
|---|---|---|
| −7 °C | 5,0 | 2,58 |
| −15 °C | 4,78 | 2,54 |
| −25 °C | 3,6 | 2,22 |
Source: national heating databook (Nordic market), p. 39. Low-temperature capacity table. Manufacturer anonymised. Design point Pdesign at −10 °C = 3.2 kW.
Down to −15 °C the unit holds output almost unchanged (5.0 → 4.78 kW) — excellent. But in harder cold it delivers less: only 3.6 kW at −25 °C. And it becomes less economical too — COP falls from 2.58 to 2.22, meaning you get less and less heat from each kilowatt-hour of electricity.
Now a second unit — Device B — of the same catalogue class (2.5 kW), but with different cold-weather behaviour:
| Outdoor temperature | Heating output (kW) | COP |
|---|---|---|
| −7 °C | 4,05 | 2,79 |
| −15 °C | 3,45 | 2,50 |
| −25 °C | 2,25 | 1,91 |
Source: national heating databook (Nordic market), p. 41. Low-temperature capacity table. Manufacturer anonymised. Design point Pdesign at −10 °C = 2.8 kW.
Notice: both units are the same catalogue class — 2.5 kW — yet at −7 °C they already differ in output (5.0 vs 4.05 kW), and at −25 °C greatly (3.6 vs 2.25 kW). Device A holds output deeper into the cold, but at mild cold it is slightly less economical (COP 2.58). Device B delivers less power in the cold, but at −7 °C it is more efficient (COP 2.79) — while in deep cold both its output and its efficiency collapse (2.25 kW, COP 1.91).
Which is “better”? Wrong question. The right one is the unit that fits your house and your location — and you can’t read that off the leaflet. In a −24 °C zone what matters is that A holds output; in a milder region B may be cheaper to run for most of the season.
There is a trap deeper than the output drop itself. Manufacturers calculate their “kW” by different methods. One states output without accounting for defrost (every so often the unit must melt off the ice and briefly stops heating). Another states the value already net of those losses. A third measures under yet other conditions.
The result: two “5 kW” figures from two catalogues do not mean the same thing. Placing them side by side — by eye — you compare things that aren’t comparable and draw the wrong conclusion. To compare fairly you must first bring every unit onto a common method and to the temperature that actually occurs in your region in winter (the design temperature).
Instead of a single leaflet number, sensible heating sizing looks at several things at once:
Each of these is a separate calculation on the unit’s real data — not something you can estimate by sight. And that is exactly why “by eye” heating sizing can come back to bite you only in January, when it is already too late to change.
You’ll find low-temperature output and COP values for specific units in the datklima knowledge base — every number carries a provenance tag. You can also check a heating selection with the assistant.
Because in heating mode an air conditioner is a heat pump: it moves heat from the outdoor air into your home. The colder the air, the less heat there is in it to “pump”, so heating output drops and the electricity used per unit of heat (COP) rises. The house needs heat most exactly when the unit delivers the least.
The leaflet number is quoted for mild conditions (usually +7 °C). On the real curve the same 2.5 kW-class unit delivers, at −15 °C, e.g. 4.78 kW or 3.45 kW — depending on the model — and even less at −25 °C. The exact value for a specific unit must be read from its heating curve, not from the headline.
No. The catalogue (nominal) output is measured in mild conditions and tells you how much the unit heats “on a good day”. In frost — that is, when it matters most — it delivers far less. Heating sizing depends on the real output at the local design temperature, not on the leaflet number.
Because manufacturers calculate their “kW” by different methods. One states output in stable conditions, without accounting for the defrost cycle (an optimistic value), another states it net of those losses. So two “5 kW” figures from two catalogues do not mean the same thing. To compare fairly you must bring every unit onto a common method and to the same temperature.
It is the lowest temperature realistically expected in a region in winter and against which heating is designed. Across most of Central Europe it is roughly −16 °C to −24 °C depending on the zone. Output is sized for this temperature, not for an “average winter”, so the house stays warm on the worst days too.
It is the outdoor temperature below which the unit alone stops being sufficient and an additional heat source (e.g. an electric heater) engages. Above it the heat pump/air conditioner heats on its own; below it, it needs support. A well-chosen bivalent point is a compromise between equipment cost and electricity cost.
You calculate the house’s heat loss at the local design temperature, read the unit’s real output at that very temperature (from its heating curve, not the headline), set the bivalent point, and in multi-split sets check whether the outdoor unit can carry all indoor units at once.
The tables above are anonymised excerpts from national heating databooks (Nordic market), given to illustrate the mechanism.