Cooling kW vs power kW: the two numbers on a spec sheet

A spec sheet gives cooling capacity and power input in the same unit, and readers routinely collapse them into one number. Cooling capacity is heat leaving the room. Power input is electricity entering the machine. A model listed at 2.6 kW of cooling does not pull 2.6 kW from the socket, and the gap between the two is wide.

Cooling capacity answers what the machine does to the room. It counts heat removed, measured against a fixed set of indoor and outdoor test conditions. This is the figure the trade means by size. It is what an aircon horsepower bracket stands in for, and it is the number a room load has to be matched against.

Power input answers what the machine costs to run. It counts the electrical energy drawn by the compressor, the two fans and the control board while the unit holds its rated output. Nothing else on the sheet feeds a bill estimate. Capacity does not, and the BTU/h figure does not either, because BTU/h is only capacity restated in another unit.

The confusion has one source. Kilowatts describe a rate of energy, and heat and electricity are both energy, so the unit really is the same. What differs is the direction and the medium. A kilowatt of heat leaving a bedroom and a kilowatt of electricity arriving at a wall socket are the same quantity pointed at different things.

Getting this wrong compounds quietly. Multiply a capacity figure by a tariff and the answer overstates the truth by a factor of three to five, and every conclusion drawn from it inherits the error. That single piece of arithmetic sits behind most of the alarm that arrives before anyone has read a meter.

An aircon does not manufacture cold. It relocates heat the room already contained, and shifting heat that exists takes far less electricity than conjuring a cold from nothing would.

Refrigerant does the carrying. Inside, it evaporates at a temperature below the room and soaks up heat from the air crossing the indoor coil. The compressor then squeezes it until it condenses above outdoor air temperature, and the fins on the condenser hand that heat off to the street. Electricity pays the compressor for the transport, not for the heat itself.

Output larger than input looks like something for nothing, and it is not. Add the heat removed to the electricity consumed and the outdoor unit rejects the sum of both. That is why the air off a condenser is hotter than anything the room ever held, and why a blocked outdoor unit punishes a system so fast.

The same hardware runs backwards in colder countries. Point the cycle the other way and it delivers heat indoors, still moving it rather than generating it. Nothing about the principle is exotic. A fridge does the identical job on a smaller box, which is why the back of a fridge is warm while the inside is cold.

The ratio has a ceiling, and physics sets it rather than marketing. A wider temperature gap costs more electricity for every unit of heat shifted. The same machine therefore buys its cooling at a worse rate on a hot afternoon than it does at night. Any claim of an output multiple far outside the range below is quoting a laboratory corner, not a living room in Singapore.

The relationship also explains a pattern homeowners report often. A system losing its ability to shed heat outdoors keeps drawing electricity while moving less of it, so the ratio collapses even though the input figure holds steady. The bill stays where it was, or climbs, while the room gets warmer. That combination points at heat rejection rather than at a machine that has stopped trying.

Divide cooling capacity by power input and the answer is the coefficient of performance. A unit shifting 3.5 kW of heat while drawing 1.0 kW of electricity has a COP of 3.5. That single value is the honest way to read a spec sheet, because it holds the two figures against each other instead of quoting whichever one flatters the model.

COP describes one measured moment rather than a season. Rated cooling capacity and rated power input are both taken at a defined indoor and outdoor temperature with the compressor held at full output. The ratio between them belongs to that condition. It compares two machines fairly because both were put through the same test.

Singapore publishes the pair for every registered household model, so none of this has to be taken on trust. The NEA register carries a rated cooling capacity and a full-load COP against each of 340 models. The lowest ratio anywhere in that set is 3.0. Not one registered machine draws anything near the electricity its cooling figure states.

The table below pairs the capacity bands homeowners actually shop in with the input figures sitting behind them. Read it as the shape of the relationship rather than a lookup for a specific model.

Two machines in the same band can separate sharply on input. The register holds a pair of single splits both rated near 3.5 kW of cooling, one drawing 0.70 kW at full output and the other 0.96 kW. Same heat out of the room, and one of them buys it with a quarter less electricity. Capacity on its own cannot see that difference. The ratio can.

A second efficiency figure exists for behaviour away from full output, and the register carries that one too. It answers a different question about the same machine. It does not change what either nameplate number means.

The rating label carries both figures, printed close enough together to be conflated at a glance. Cooling capacity usually comes first. Power input follows, and it may be printed as rated input, input power or power consumption depending on who made the machine. On a wall unit the sticker hides behind the front flap or runs along the lower edge of the casing.

A brochure states the same pair inside a specification table, with more columns and more room for error. Capacity appears as a rated value flanked by a lowest and a highest figure. Input appears the same way. Read across a single column. Pairing a rated capacity with a minimum input produces a ratio no machine on the register reaches.

A cost estimate uses power input and nothing else. Capacity in kW and the BTU/h figure are one statement in two units, and neither of them is electricity. If a calculator or a salesperson asks only for the BTU figure, it is treating capacity as a stand-in for input, which holds only while the efficiency behind it is being assumed rather than checked.

A third electrical figure on the same label catches people out. Rated current, printed in amps, describes what the circuit has to carry rather than what the meter records. Multiplying amps by voltage overstates the draw on any motor load, because current and voltage do not stay in step across the cycle. Take the kW input figure for cost, and leave the amp figure to whoever is sizing the breaker.

Where the label has gone, the model number settles it. The NEA register publishes capacity and full-load COP against the model, and dividing the first by the second returns the input figure even with no sticker left to read. Sun and repainting take the indoor label long before the machine itself gives up.

The outdoor unit carries its own label, and on a shared system that one is the honest total. Two bedrooms running together on a single condenser draw less than two separate systems would, which is a genuine advantage and a genuine complication for anyone trying to cost one room. The per-room answer is not printed anywhere, and it is not the head's capacity divided by anything.

The rated input figure is one point on a curve, and a modulating unit spends most of its life somewhere else on that curve. The compressor changes speed continuously to match what the room is losing, and input power tracks that speed minute by minute. The nameplate states what the machine draws while holding rated output, which is a condition it passes through on the way to comfortable and seldom returns to.

This is why multiplying the nameplate input by runtime overestimates. The multiplication assumes full output for the whole session. A correctly sized inverter in a settled room does the opposite, pulling hard until the set point arrives, then dropping to a fraction of rated input and holding there against whatever the walls and windows let back in.

The overestimate is not guaranteed, and that is the part worth knowing. An undersized machine, a room with heavy afternoon glass, or a coil fouled enough to blunt the heat transfer never lets the compressor back off. In those rooms the nameplate input is close to the sustained truth. A bill that matches the naive multiplication is a symptom worth investigating, not a coincidence.

The published range tells you how much modulation there is to work with. A specification table showing input as a rated value between a minimum and a maximum is describing how far the compressor can wind down. A wide span means the machine can idle genuinely low. A narrow one means it stays nearer rated whatever the room is doing, and the naive multiplication is closer to right on that unit than on a deeply modulating one.

Turning input power into a bill is arithmetic rather than judgement, and it needs runtime, room count and the tariff alongside the figure. The aircon running cost tool does that sum. This page settles which of the two nameplate numbers belongs in it. The label states factory condition, so servicing is what keeps a machine near its published ratio and neglect is what walks it away.

Ask for both figures before signing anything. A quote naming a capacity and no input has described what the machine does to a room and said nothing about what it costs to do it. Both numbers are printed on the same sheet, and the one that gets left off is never the flattering one.