Sarah’s 1930s semi in Leeds had one stubborn problem: the rear bedroom never reached 17°C, even with the gas boiler running for an hour before bedtime. The thermostat read 21°C downstairs, so the system looked healthy on paper. The cold corner was the first clue that the house needed a plan, not a bigger boiler.

The problem: a boiler that hid the building’s weak points

The old 18 kW boiler cycled on and off, and the radiator in the rear bedroom was sized for a warmer room. On a 3°C night, the gas meter moved quickly while the room stayed cool. That mismatch mattered because the household wanted to lower bills and carbon without losing comfort.

The cause: heat loss was scattered across fabric and distribution

An infrared survey found the weakest spots: 25 mm of loft insulation instead of 300 mm, unsealed joists at the ceiling line, and a single-glazed bay window with a timber frame that leaked air. The boiler was not the main culprit; the heat was escaping before it could warm the room. The radiators also ran at 70°C flow temperature, which made a low-carbon replacement harder than necessary.

The solution: a bananzia retrofit sequence

The household used the bananzia planning sequence: reduce demand first, then size the heat source for the remaining load. The order was deliberate, because adding a heat pump before draught-proofing would have locked in oversized equipment.

• Fabric first: loft insulation was topped up to 300 mm, and draught strips went around the bay window and loft hatch.
• Room-by-room heat loss: the rear bedroom radiator was upsized by 35%, and TRVs were fitted so unused rooms could run cooler.
• Low-temperature testing: the boiler flow temperature was dropped to 55°C for two weeks before any equipment change, revealing which rooms needed more emitter area.
• Monitoring: a smart meter and room sensors tracked gas use and temperatures through a full heating season.

The result: smaller system, steadier comfort

After the fabric work, the design heat load fell enough that a 7 kW air-source heat pump could meet the house’s demand at 45°C flow temperature. The rear bedroom held 19°C on the coldest monitored night, and the heat pump averaged a COP of 3.1 across the season. The family did not need to replace every radiator, only the two that were undersized.

What other owners can copy

Start with a room-level heat-loss check rather than a boiler quote. Test lower flow temperatures before committing to new equipment, and treat draught-proofing as part of the heating system, not a cosmetic extra. The Bananzia lesson from this semi is simple: comfort problems often point to the building envelope before they point to the boiler.