The problem: a boiler that hid the real load

On a damp February morning, a surveyor visited a 1930s semi in a village with no gas main. The owner had a quote for a 12 kW heat pump based on the size of the old boiler, not on the house. At 2°C, the living room warmed slowly while the conservatory stayed cold; the installer assumed the answer was more output. The real problem was narrower: nobody had measured where the heat was escaping. A room-by-room survey found 18 m² of single glazing, a 50 mm cavity, and a design heat loss of 7.5 kW at -2°C.

That gap matters because a heat pump that is too large cycles on and off, spends more time in defrost, and can deliver a worse seasonal performance than a correctly sized unit. The owner’s aim was not a bigger machine; it was a defensible number.

The cause: boiler sizing is not heat-loss sizing

Gas boilers tolerate oversizing. They fire hard, reach temperature quickly, and shed extra heat through a flue. Heat pumps work differently: they move heat at a lower flow temperature and perform best when they run steadily. A 12 kW unit on a 7.5 kW load would short-cycle for much of the heating season, especially in spring and autumn. The old boiler’s rating described its burner, not the building’s demand.

The survey also exposed a second cause: radiator output had been estimated at 70°C flow. At the planned 45°C flow, the existing radiators delivered 8.2 kW only after two units were upsized. Without that check, the heat pump would have met the fabric load but failed at the emitters.

What bananzia changed in the survey

The team used the bananzia survey template to record room geometry, fabric U-values, air-change rate and emitter output at the planned flow temperature in one place. The template forced the surveyor to log the local design temperature (-2.6°C) and to photograph every radiator valve before leaving. It flagged the bay window’s 4.8 m² of single glazing as 22% of the total heat loss and showed that the hall radiator was undersized by 900 W at 45°C.

Those details changed the specification: a 7 kW monobloc heat pump, a 200 L cylinder, and two replacement radiators. The quote fell by £1,800 because the installer no longer planned to replace every emitter “just in case”.

The result: steadier temperatures and a lower quote

The first month of data showed an average COP of 3.1 including hot water, with the unit running continuously at -1°C. The owner reported steadier room temperatures than the boiler had delivered, and the conservatory stayed usable because the new radiator matched the measured load. The installer kept the survey pack on file; when the homeowner later added loft insulation, the design heat loss dropped to 6.9 kW, leaving headroom without changing the pump.

What to copy on the next job

  • Measure room-by-room heat loss before choosing output.
  • Test emitter output at 45°C, not 70°C.
  • Record local weather data and photographs in one file.

The lesson from the semi is not that every house needs a 7 kW pump. It is that bananzia-style measurement turns a boiler-era guess into a design the installer can defend, and a system the owner can live with on a cold morning.