Three quotes, three sizes

With the manufacturers’ declared data, an oversized inverter air-source heat pump uses almost the same electricity as the right size: in the worked example below, the 16 kW class used about 70 kWh a year more than the 10 kW class the sizing proposed, about 18 EUR. What the bigger unit changes is how often its compressor starts, its price on the quote and, if the quote is for two units in cascade, the standby power; undersizing is what costs energy, because the electric backup heater takes over, and the 4 kW class costs about 320 EUR a year more on the same house.

The rest of this page shows where those numbers come from, why the water in your heating system matters as much as the size on the label, and what the model leaves out.

The worked example

One house, six heat pump sizes, everything else the same.

• A detached two-storey house with 182 m² of heated floor and an unheated garage inside, built in 2025 to the German new-build standard, in Würzburg.
• The typical weather year the model builds for Würzburg.
• Underfloor heating, mechanical ventilation with heat recovery, no cooling, no battery, no electric car.
• 10 kWp of PV facing south at 30°, which produced 11,074 kWh in the year.
• A fixed tariff: 0.3214 EUR/kWh to import, 0.077 EUR/kWh paid for export, no fixed annual charge. Net cost is imports minus export payments.
• The wizard’s household defaults: four residents, 4,500 kWh of household electricity, 21 °C by day and 19 °C at night.
• An 8 kW electric backup heater in every run.

The sizing step proposed the 10 kW class. It found a peak heat loss of 6.2 kW at the design temperature of -11.5 °C, plus 0.55 kW for hot water, and the 10 kW class delivers 8.59 kW at -11.5 °C.

Every run delivered about 4,300 kWh of heat to the house. The SCOP here is the running efficiency, without standby, so a fixed 16 W does not drag down the figure of a house that needs little heat (see COP and SCOP). Its differences between sizes come from each unit’s published COP tables, not from cycling.

Two things stand out. Above the proposed size, the cost hardly moves while the starts climb: the 16 kW class starts its compressor about 40 % more often than the 10 kW class. Below it, the cost climbs fast, and the 4 kW class has a good SCOP and the highest bill, because its backup heater supplied 1,296 kWh at a COP of 1.

What happens below the minimum output

An inverter heat pump varies its compressor speed to match the heat the house asks for, down to a minimum. For the 10 kW class that minimum is 2.61 kW at 7 °C outdoors with 35 °C water; for the 16 kW class it is 3.3 kW. A well insulated house on a mild day needs less than either.

Below its minimum the unit cannot run continuously. It runs at minimum output, warms the water a little above the target, stops, waits while the house draws the water back down, and starts again. That is cycling, called short cycling when the gaps get short. A bigger unit has a higher minimum, so it spends more hours below it and starts more often.

Starts are what owners notice. A compressor start brings noise and vibration, and fewer starts mean less wear and a longer compressor life, as the heat pump maker iDM puts it in its note on on/off control .

Why the water volume matters

How long one cycle lasts depends on two things: how much water the heat pump heats, and how far the controller lets its temperature swing between start and stop (the control band). More water takes longer to warm and longer to cool, so each cycle is longer and there are fewer of them in an hour. Starts are most frequent when the house asks for about half the unit’s minimum output.

The model takes the water content of the heating system from a manufacturer’s planning table (Reflex Expertenkarte B31): about 20 litres per kW of design heat load for underfloor heating and 14.6 for low-temperature radiators. In the example the underfloor system holds about 130 litres. At the worst point, the 10 kW class can start about 4 times an hour, below the 7.5 that the controller’s minimum run and pause times allow.

Radiators hold less water per kW than an underfloor system, so on the same house the same unit starts more often on them. A buffer tank adds its own volume to the water the heat pump heats, and the model counts it. In a real floor the screed around the pipes stores heat as well, which the estimate does not count, so for underfloor heating it errs on the high side.

Standby

A heat pump draws a little power when it is not running. The EU product sheets of the units in the model declare 16 W in standby, 14 W with the thermostat off and 0 W for a crankcase heater. The model adds 16 W per unit in every hour the unit does not run. In the example that is about 120 kWh a year, about 27 EUR, and it is the same for every size, because every size declares the same 16 W. A quote for two units in cascade draws it twice.

Standby is in the heat pump electricity, in the cost and in the system figure of the results. It is not in the headline SCOP, which shows the efficiency while running. The label’s seasonal figure does include standby, worked out for a reference house, so it is not the same number.

What the label says about cycling

The European standard for seasonal efficiency, EN 14825, handles the loss from cycling with a degradation coefficient, Cdh. A Cdh of 1 means no loss; the standard falls back on 0.9 when a manufacturer has not measured it. The product sheets of the units in the model declare 1 for both their medium and low temperature ratings (Viessmann product sheet 6174272-05, page 4 , for the 10 and 13 kW sizes). So the model applies no energy penalty for cycling, and neither does the label.

Undersizing costs energy

The 4 kW class delivers 3.02 kW at -15 °C, less than half of this house’s 6.2 kW design loss. In every hour it falls short, the backup heater covers the rest at a COP of 1. That is why the bill rises step by step as the unit shrinks: 15 kWh of backup at 10 kW, 295 at 8, 652 at 6 and 1,296 at 4. The backup heater article explains when the element switches on in the model.

What the model does and does not do

• Heat pump energy comes from the manufacturer’s published data: capacity and COP by outdoor and water temperature, the minimum output, and no cycling penalty, because the product sheets declare none.
• Standby is counted in the energy and the cost.
• Compressor starts are an estimate saved with each run and never added to the energy. It uses a 1.0 K control band and a 3 minute minimum run and 5 minute minimum pause. Viessmann publishes none of these for the units in the model, so they are engineering estimates inside the typical ranges iDM gives: 0.5 to 1.5 K, 2 to 5 minutes and 3 to 10 minutes. Starts scale with the inverse of the band, so a 0.5 K band doubles them. The comparison between sizes holds better than the absolute count.
• The results page, the comparison and the PDF report have a Heat pump sizing block: the design heat load the size rests on, the size used and its output at the design temperature, the compressor starts and the standby.
• Purchase price, noise and compressor life are not modelled.
• The performance data comes from one manufacturer’s mid-range line.

Checking three quotes

• Ask each installer what heat load the size is based on, and the unit’s output at your design temperature. If the sizes differ, compare the heat loads first.
• Ask for the minimum output, not only the maximum. On a new or well insulated house it decides how often the unit will start.
• Ask how much water the unit will heat, and whether a buffer tank is planned with radiators.
• Simulate your own house. The wizard proposes a size from its heat loss; pick each size on your quotes at the Choose equipment step, run it, and compare the runs side by side for cost and backup heater electricity.