Short answer: a heat pump that is too big for the house short cycles because, on most days of the heating season, the house needs less heat than the unit's lowest output. It cannot slow down any further, so it stops and starts. A unit twice the size the house needs spends most of the winter doing that, losing seasonal efficiency and adding compressor starts.
This article is about air-to-water heat pumps feeding radiant floors, the hydronic systems HeatAlgo works with. Most US threads about short cycling are about air-to-air units, ducted systems and mini-splits, and many also discuss cooling and humidity, which this article does not cover. The heating mechanism is the same in both: when the house needs less heat than the compressor's minimum, the unit cycles. One homeowner on Green Building Advisor asked it directly: "How much efficiency does a short cycling heat pump lose?" The honest answer depends on the two ratios below.
What "oversized" means
It is not about a big number on the label. In a European example from our source article, a unit delivering 16 kW (about 54,600 Btu/h) at design conditions in a building with a 15 kW (about 51,200 Btu/h) load is sized correctly. One delivering 8 kW (about 27,300 Btu/h) in a building with a 4 kW (about 13,600 Btu/h) load is oversized by a factor of two. These are illustrations, not sizes for any US house.
Two relationships matter:
- the unit's output against the house's design heat loss, which is the result of a room-by-room load calculation for that house in that climate;
- the unit's minimum output against what the house needs for most of the season, which gets discussed far less and decides how the unit actually runs.
Why a heat pump that is too big cannot just slow down
An inverter compressor modulates, but only within a range, and the bottom of that range varies widely between units. A minimum output figure means little without the operating point it was measured at. Data sheets usually give an output range at one rating point, for example A7/W35 (outdoor air at 7 °C (44.6 °F), supply water at 35 °C (95 °F)). Compare that range with what the house needs, not a percentage read off the model name.
At the same time, a house needs its full design load only in the coldest hour of the year. In the European example (68 °F (20 °C) indoors, a design temperature of -4 °F (-20 °C)), demand at 32 °F (0 °C) outside is roughly half the design load, and at 46.4 °F (8 °C) about a third. Most of the season is spent in that part-load range, not in a cold snap.
Put those two facts together and you have the problem. A correctly sized unit with a minimum of about 30 percent of its output can follow the house down to mild weather. Put the same kind of unit, twice too big, on the same house and its minimum sits at about 60 percent of the house's design load. In the source article's chart for that example, the oversized unit's minimum only meets demand at around 23 °F (-5 °C); above that, it can only cycle. The right-sized unit meets demand down to around 46.4 °F.
When the house needs less than the unit's minimum, the unit has no way to deliver less. It stops, waits and starts again. That is short cycling, and with an oversized unit it happens even when the piping is flawless. The bigger the unit, the higher its minimum sits against the house's demand, and the more of the season it cycles.
What short cycling costs
Seasonal efficiency. A data sheet's COP is measured in steady running. The seasonal figure on European air-to-water data sheets, SCOP, is not measured at all: it is calculated to EN 14825 for a reference building whose load matches the unit, with cycling entered as a fixed allowance meant for a well-matched unit. In a house where the unit cycles most of the season, the rated and the real seasonal figure are two different numbers.
Compressor wear. Manufacturers cap the number of starts per hour, because a start stresses the compressor differently from steady running: inrush current, oil migration and temperature swings. A unit starting several times an hour uses up that allowance far faster.
Comfort. Stop-start running swings the supply water temperature, and in a radiant floor with little mass the room temperature too. This is the part you notice without opening a bill.
Hot water. Heating domestic hot water interrupts space heating. In a unit already cycling, those interruptions add up.
Where the urge to upsize comes from
A habit from boilers. A gas boiler modulates across a wide range, and oversizing it costs little. "Take the bigger one for peace of mind" worked there. With a heat pump it does not.
Adding hot water on top of the heating load. Domestic hot water is an intermittent load shifted in time, not a constant one added to the house's demand. With a properly sized tank and priority control, adding it usually is not needed at all.
A rule of thumb instead of a calculation. A per-square-foot rule has a spread wider than the step between one unit size and the next. We measured it on our own module, for one European building: the full EN 12831-1 calculation gave 8.40 kW (28,660 Btu/h), while the rule-of-thumb estimate gave anything from 8.02 kW (27,360 Btu/h) to 23.52 kW (80,250 Btu/h), depending only on which insulation class the person estimating picked.
Fear of a call on the coldest night. The most honest reason, and probably the strongest. The answer to it is not extra capacity but a deliberate balance point.
The balance point: a smaller heat pump does not mean a cold house
The fear of going too small assumes the heat pump alone must cover the whole load in the coldest hour of the year. That is one option, not the only one. Below a chosen outdoor temperature, the balance point (European designers call this bivalent sizing), backup heat covers the difference. How many hours a year that is depends on the balance point and the location. What matters more is how much energy those hours carry: the backup covers only the gap between demand and the unit's output, not the whole demand, so over a year it is a small share. Both can be counted from climate data for the location instead of guessed.
The result runs against intuition: a smaller unit with a deliberate balance point spends most of the season inside its modulation range and achieves better seasonal efficiency than a larger unit that spends that same season cycling.
How to check the size you were quoted
- Get the design heat loss room by room for the outdoor design temperature where the house stands; for a US house that is the contractor's room-by-room load calculation. Size the heat pump on the whole house's load, not the sum of the room loads: ventilation is counted room by room with an allowance that cancels out across the house, so the room sum comes out higher. Use the room loads to size the radiant floor loops.
- Check the floors room by room and set the supply water temperature. One room whose floor cannot cover its load raises the supply water temperature for the whole system, and every degree higher lowers efficiency all winter. Without the room-by-room numbers you will not find that room, and without the temperature you cannot do step 3.
- Read the unit's real output from its data sheet at the design temperature and the supply water temperature from step 2, not the number in the model name. A unit sold as "12 kW" (about 41,000 Btu/h, a European model name) delivers considerably less at -4 °F outdoors with 131 °F (55 °C) supply water than at the conditions its name came from.
- Check the minimum output of the chosen unit, with the conditions its data sheet quotes it at, and compare it with what the house needs at around 41 °F (5 °C). In the European example, close to the middle of the heating season, demand there is about 40 percent of the design load. A minimum above that means cycling for most of the season.
- Choose the balance point deliberately and write it down with the decision. If backup heat is part of the plan, check that the electrical service can carry it.
- Treat hot water separately, as an intermittent load with priority control.
- Check the system water volume. Too little water causes cycling even with a correctly sized unit; manufacturers state a minimum, and radiant zones closed off in use can take the system below it.
Common mistakes
- Sizing from the number in the model name instead of the output at the design temperature and the needed supply water temperature.
- Adding hot water capacity straight onto the house's load.
- Ignoring the minimum output when comparing two units of similar nominal size.
- Sizing from a per-square-foot rule and ordering equipment on it.
- Sizing on the sum of the room loads instead of the whole house's load. It is the simplest arithmetic route to an oversized unit.
- A "just in case" margin added after the calculation. EN 12831-1 has no safety factor; its assumptions are already conservative: no internal or solar gains, every room at its design temperature at once, and a statistically extreme outdoor temperature. A percentage on top doubles that margin. The only addition the standard provides for is reheat power for intermittently heated systems, and that is calculated, not estimated.
- No record of the balance point, so at handover nobody can reconstruct why the unit was chosen.
Where HeatAlgo fits
HeatAlgo does not size a heat pump for a US house: it has no US climate data and no cooling. What it does is the radiant floor side of step 2: with each room's load from your contractor's calculation, HeatAlgo's radiant floor heating software works out the floor output for tubing in a slab or thin slab and the supply water temperature the floors need. That temperature is the number to read the heat pump's data sheet at. Any heat pump sizing figure is preliminary and is the contractor's call.
The engine behind the figures above also calculated HeatAlgo's reference house, a 102.2 m² (1,100 ft²) house calculated for a Warsaw design day of -20 °C (-4 °F), with every input published for checking. For the floor side, see radiant floor heating water temperature.
Check your radiant floor's water temperatureA HeatAlgo report follows EN 12831-1 and EN 1264, not ACCA Manual J, and is not accepted for permits or rebates.