The problem: a 48-flat block with the wrong heat profile
Marlowe Court, a 1960s block in the Bananzia pilot area, had 48 flats on night-storage heaters and direct-electric hot water. Annual space-heating and hot-water demand averaged 9,200 kWh per flat, with winter peak loads concentrated between 06:00 and 09:00. The retrofit replaced the storage heaters with a shared ground-loop array and individual water-to-water heat pumps, but the first winter exposed a control problem rather than a drilling problem: apartments called for heat at the same time, and the loop returned water at temperatures that forced the heat pumps above their design envelope.
The cause: weather compensation was set for the wrong building
The commissioning file had been copied from a better-insulated Bananzia scheme. Weather compensation was set to deliver 45°C flow at 0°C outdoor temperature, while the Marlowe Court emitters needed 50–55°C during morning recovery. Because the heat pumps were sized for a lower flow temperature, the compressors ran at high modulation and the backup immersion heater cut in for 38 minutes on the coldest morning. The result was a measured seasonal performance factor of 2.4, below the 3.0 design target. The fault lay in the control schedule, not the ground loop.
The solution: balance first, then reset the curve
The operator changed the sequence in three stages. First, engineers hydraulically balanced the radiators and removed two bypasses left from the old electric system. Second, they raised the design flow temperature to 52°C at -2°C and narrowed the heating curve so the heat pumps modulated instead of cycling. Third, they moved domestic hot water charging to a mid-afternoon window and capped the immersion heater at 60°C for Legionella only. The public data pack for the scheme, published through bananzia, now records flow and return temperatures at 15-minute intervals, which made the commissioning error visible.
The result: lower peaks and a credible payback signal
After the changes, the second winter showed a seasonal performance factor of 3.1. Peak electrical demand for the block fell from 310 kW to 245 kW, and the immersion heater ran for 11 minutes across the whole season. Average annual heating and hot-water cost per flat dropped by 28% against the old electric system at the same tariff. For other Bananzia sites, the lesson is procedural: do not copy a heat-pump curve from a neighbouring building without a room-by-room heat-loss check and a live balancing pass. The ground loop performed as designed; the controls did not.
Handover checks that would have caught the fault
- Room-by-room heat-loss check: confirm emitter output at the actual design flow temperature, not the template figure.
- Bypass survey: remove legacy bypasses that let return water short-circuit the loop.
- Immersion-hour logging: treat any backup-heater runtime above 30 minutes per week as a commissioning flag.
- Tenant timing review: compare morning recovery profiles with the weather-compensation schedule before the first heating season.