The Carnot formula gives a ceiling, not a forecast. A unit rated at COP 4.2 at 20 °C ambient does not deliver 4.2 at -7 °C, and the fall is not a straight line. The winter output falls faster than the simple temperature ratio because three other losses stack on top of the Carnot limit.
This article separates the ceiling from the operating point and shows the stack. The reference is a 40 kW air-source water heater at 55 °C outlet.
The Carnot ceiling is not the operating COP
The ideal COP for a heat pump is the hot absolute temperature divided by the hot-cold difference. At 20 °C ambient and 55 °C water, the ideal is 328 / (328 – 293) = 9.4. A mature commercial unit reaches about 45% of that, or 4.2. At -7 °C ambient, the ideal is 328 / (328 – 266) = 5.3, and 45% gives 2.4, not 2.2. The exact fraction depends on the compressor and the exchanger, but the direction is fixed.
| Ambient | Ideal COP | Actual at 40–45% | kWh per metric ton* |
|---|---|---|---|
| 20 °C | 9.4 | 4.2 | 11.1 |
| 7 °C | 6.8 | 3.1 | 15.0 |
| 0 °C | 6.0 | 2.6 | 17.9 |
| -7 °C | 5.3 | 2.2 | 21.1 |
| -15 °C | 4.7 | 1.9 | 24.5 |
*One metric ton of water from 15 to 55 °C needs about 46.5 kWh of heat. The electricity is that heat divided by the actual COP.
Why the gap widens in winter
The formula already includes the temperature lift. The reason the real unit falls further is that the compressor and the coil do not behave like the ideal cycle. The suction specific volume rises, so the same swept volume moves less mass. The pressure ratio rises, so the volumetric efficiency falls. The two multiply.
At 3 °C evaporating, the specific volume of R410A is about 0.031 m³/kg. At -12 °C it is about 0.050 m³/kg, a factor of 1.6. The mass flow falls to 0.62. The volumetric efficiency falls from about 0.88 to 0.78, a factor of 0.886. The product is 0.55. A 40 kW unit becomes 22 kW before defrost, and 20–22 kW after a normal winter defrost cycle.
This is why a unit selected on the 7 °C nominal capacity can be 30–40% short on the coldest week. The nominal number is not wrong; it is a different point on the same curve.
Four losses the formula does not show
| Loss | Trigger | Effect |
|---|---|---|
| Defrost | 2–5 °C, RH >70% | 8–12% of winter heat |
| Cycling | Tank <40% of daily draw | 3–5% of run time |
| Scale | Hardness >200 mg/L | 6–9% COP |
| Pipe and tank loss | Thin insulation | 3–8% of heat |
Defrost is the largest and most visible. At 2–5 °C and high humidity, the coil can frost in 30–60 minutes. Each defrost stops heating for 5–10 minutes and reverses the cycle. A unit that defrosts every 40 minutes in a humid winter can lose 8–12% of its seasonal heat, and a unit that defrosts every 25 minutes can lose more.
Cycling is the second. A tank that is too small, or a control that starts the unit for every small draw, adds a low-efficiency start every time. If the tank is below 40% of the daily draw, the unit may start 12 times a day; each start has a 3–5 minute low-efficiency period. The loss is small per start, but it accumulates.
Scale is the third. A 0.5 mm layer on the condenser raises the condensing temperature 2–3 K and cuts COP by 6–9%. In a hard-water site this can be larger than defrost and cycling combined. Pipe and tank loss is the fourth; 20 mm insulation is a minimum, not a guarantee.
A 40 kW unit at -5 °C
At -5 °C ambient, the evaporating temperature is about -12 °C, the suction pressure about 0.53 MPa, the condensing pressure about 3.83 MPa, and the ratio about 7.2. The capacity factor is 0.55–0.62, so the unit delivers 22–25 kW. If the site needs 30 kW at that ambient, one unit is short.
The correct response is not to add electric heat without checking the load. First, verify the actual winter capacity with the manufacturer’s curve. Second, check the defrost setting and the coil condition. Third, check the tank and the recovery schedule. If the load is genuinely 30 kW, add a second unit or a low-temperature model with vapor injection; a 40 kW nominal unit cannot be tuned into a 30 kW winter unit.
Sizing rule
Select on the coldest week, not the annual average. Use the local design ambient, the leaving-water temperature and the defrost penalty. A unit selected at 7 °C with no winter correction will be oversized in summer and undersized in winter, which is the worst combination for both cost and comfort.
A practical rule: required nominal capacity = winter load / winter capacity factor. If the winter load is 30 kW and the factor is 0.58, the nominal is 51.7 kW. A 52 kW unit, or two smaller units, is the honest selection. A 40 kW unit is a summer unit.
When the limit is structural
Below -20 °C, the Carnot ceiling itself falls to about 4.4, and the actual COP to about 1.8. At that point the heat pump is barely better than electric resistance, and the economics may not support it unless there is a low-cost electricity tariff or a dual-source design. Vapor injection and a larger coil can extend the range, but they cannot repeal the temperature lift.
Above 80 °C delivery, the condensing temperature is outside the normal range and a single-stage unit is the wrong machine. A cascade or a high-temperature heat pump is needed; the Carnot formula still applies, but the hardware changes.
Boundary
This article covers the seasonal output of a single-stage air-source water heater. It does not cover ground-source or water-source units, whose source temperature is more stable. The method of stacking Carnot, mass flow, volumetric efficiency and site losses still applies.
Common questions
Why is the real COP lower than Carnot?
The Carnot cycle assumes isothermal heat transfer and isentropic compression. A real unit has 5–10 K approach, compressor entropy gain, pressure drop and defrost. A mature unit reaches about 40–45% of the Carnot value.
Does the COP fall in a straight line with ambient?
No. The temperature lift grows faster at low ambient, and the suction specific volume and volumetric efficiency add their own losses. The fall is steeper than the simple ratio suggests.
How much does defrost cost?
At 2–5 °C and high humidity, 8–12% of seasonal heat is a realistic range. A unit that defrosts every 25 minutes can lose more than one that defrosts every 60 minutes.
Why does a 40 kW unit give 22 kW at -5 °C?
The mass flow falls to about 0.62 and the volumetric efficiency to about 0.886; the product is 0.55. 40 kW × 0.55 is 22 kW, before the defrost time is deducted.
Should I add electric heat?
Only after checking the winter capacity, defrost, coil and tank. Electric heat can bridge a small deficit, but it is not a substitute for a correctly sized heat pump.
What ambient should I use for selection?
The local design ambient for the coldest week, not the annual average. Use the manufacturer’s capacity at that ambient and at the leaving-water temperature.
Why does capacity fall faster than Carnot?
Carnot assumes an ideal cycle. Real losses grow at low ambient: suction density falls, defrost cycles, oil viscosity rises and coil delta-T changes. The practical curve is steeper.
What else reduces winter output?
Defrost, snow blockage, low suction pressure, incorrect charge and a cold space. Fix the installation before blaming the refrigerant.
Further technical reading
For COP ratings, see How to Read a Heat Pump COP. For selection errors, see Heat Pump Selection Mistakes That Cost Money After Handover. For defrost, see Heat-Pump Defrost Logic. For the six quote inputs, see Six Inputs Before a Heat-Pump Quote.
Related reading
Heat-pump COP rating · Energy per metric ton of hot water · Heat pump vs gas boiler