The starting point
Rose Cottage is a 177 m² four-bedroom property in rural Oxfordshire — a mixed construction of filled cavity brick and solid 610mm stone, the latter being both the building's most characterful feature and its most significant thermal challenge. A professional heat loss survey by Urban Plumbers using BS EN12831 (UK National Annex, MCS-compliant) established a design heat loss of 9.06 kW at the local 99th-percentile outdoor temperature of -2.9°C.
The house had been heated entirely by gas, consuming 15,135 kWh in 2025 — a proper whole-year baseline peaking at 3,400 kWh in January. Notably, at 88% boiler efficiency this represents approximately 13,924 kWh of useful heat delivered — about 20% less than the survey's calculated 17,605 kWh annual heating demand. This is entirely consistent with the HeatpumpMonitor.org finding that actual heat loss in most UK homes is lower than EN12831 calculations predict, especially where fabric improvements have been partially accounted for. The practical implication: the house needs a smaller heat pump than the survey's headline figure suggests.
Annual electricity consumption was 12,483 kWh, which appeared anomalously high until the Zappi EV charger (~4,700 kWh/year) and Eddi solar diverter were accounted for. Underlying household electricity demand was a more typical 6,500–7,000 kWh.
The dominant fabric weakness was unmistakable: external walls accounted for 3,801 W — 42% of total heat loss. The solid stone sections at U-value 1.20 W/m²K dwarfed every other element. The roof (756 W) and windows (359 W) were secondary issues. Three rooms stood out for their heat-loss density: the TV room at 82 W/m² (2,074 W total), the en suite at 92 W/m² (775 W total), and the bath/shower at 82 W/m². These three rooms alone account for over a third of whole-house heat loss.
The approach: layers, not a single fix
Rather than reaching for the largest heat pump that would fit the original heat loss calculation, the homeowner approached the problem as a system — addressing fabric, generation, and heat delivery simultaneously and in sequence. Each intervention informed the next.
The solar story unfolded in stages. The original system was a 4.6 kWp Viridian in-roof array with a Sonnen 9.53 hybrid inverter and 15 kWh battery — a clean, integrated installation with the panels replacing roof tiles rather than sitting on top. In May 2025, Phase 1 of an Enphase expansion added 4 × 510W Aiko N-type bifacial panels with IQ8HC microinverters on the garage (2.04 kWp). In October 2025, Phase 2 completed the array with a further 18 × 510W Aiko panels, bringing the total to 22 Enphase panels (11.22 kWp) plus the original 4.6 kWp Viridian — approximately 15.8 kWp combined. Export data tells the story clearly: March exports jumped from ~90 kWh (2025, small system) to ~550 kWh (2026, full array) — a 6× increase. April 2026 alone saw ~890 kWh exported despite the Sunamp, Zappi, and battery absorbing all they could.
The roof insulation and the main panel array were installed simultaneously — the logical sequence being insulation first, panels second. SuperFOIL is a British family business founded in 1997 under parent company Boulder Developments Ltd, headquartered in Lincoln, and the UK's leading manufacturer of multifoil insulation producing over a million square metres annually. The SF40BB is their premium breathable multifoil — a 3-in-1 product combining high-performance thermal insulation, a certified W1 watertight breathable membrane, and a radiant barrier in a single 75mm, 19-layer product made from 40% recycled materials. BBA-certified to BS EN ISO 12667 and ISO 8301. Fitted between and over the rafters, the SF40BB brought the rafter section U-value from ~0.60 to approximately 0.14–0.18 W/m²K (R-value 3.67 in pitched roof application). The joist section, already at 0.15 W/m²K with 250mm mineral wool, received SF60BB (R-value 4.42), pushing performance to approximately 0.08–0.10 W/m²K — close to Passivhaus standard. Then 18 × 510W Aiko N-type bifacial panels with Enphase IQ8HC microinverters were installed on the main roof. Combined with the Phase 1 garage array (4 panels, May 2025), the total reached 22 Aiko panels (11.22 kWp) plus the original 4.6 kWp Viridian — approximately 15.8 kWp combined. The loft temperature sensor captured the before/after thermal performance in real time: summer spikes to 48–50°C were eliminated entirely, and winter loft temperatures stabilised 8–10°C above pre-installation levels.
Rather than adding a single large air-to-water unit to handle the full 9 kW demand, the three highest-density-loss rooms were decarbonised first with a 6 kW split air-to-air unit. The TV room alone carried 2,074 W of heat loss at 82 W/m² — and the heat loss survey confirmed it was physically impossible to install adequate wet radiators there: a second radiator would require new 22mm primary pipework and cannot be piped in parallel with the first. The air-to-air solves this elegantly where the wet system cannot. It also acts as peak top-up for the whole house on the coldest days.
All windows replaced with Rawington Nord composite units — an aluminium-clad laminated timber system originally designed for the Scandinavian market, triple glazed to U-0.8 W/m²K with argon-filled units, warm edge spacer bar, and Low-E glass. The aluminium exterior cladding is maintenance-free, with the timber frame providing the thermal mass and the structural character appropriate to a stone cottage. Window heat loss reduced from 359 W (at the previous 0.80–1.10 W/m²K range) to approximately 250 W — a modest saving given the already-reasonable starting point, but the acoustic and comfort improvement on a rural property is considerable.
With the three highest-demand rooms now served by air-to-air, and fabric improvements reducing whole-house loss to ~8.4 kW, the wet system demand at design conditions is approximately 5.5–6 kW. A 7 kW air-to-water unit is correctly sized: delivering 5.5–6.5 kW at -2.9°C on a 40°C flow temperature, with the air-to-air providing top-up on the rare days it falls short. Twelve radiators are being replaced and sized to run at 37–42°C design flow temperature.
The loft sensor: real proof the insulation works
A temperature sensor installed in the loft before the SuperFOIL installation provided an unusually clean before-and-after dataset. The contrast was striking.
Loft temperature behaviour — before and after SF40BB installation (Oct 2025)
Pre-installation, the loft was a thermal amplifier — a heat battery in summer driving radiant load into upper bedrooms, and a cold void in winter increasing ceiling heat loss. Post-installation, the reflective aluminium layers eliminated solar gain spikes entirely and buffered winter temperatures above ambient. This is empirical performance data, not modelled.
Solar export: the expansion in numbers
The solar export data captures the staged installation more precisely than any other metric. The jump in March and April 2026 versus the same months in 2025 is almost entirely attributable to the full Aiko/Enphase array commissioned in October 2025.
| Month | 2025 export (kWh) | 2026 export (kWh) | Change | Context |
|---|---|---|---|---|
| January | ~5 | ~70 | +14× | Full array vs original only |
| February | ~25 | ~110 | +4× | Full array vs original only |
| March | ~90 | ~550 | +6× | First full spring with 15.8 kWp |
| April | ~180 | ~890 | +5× | Peak export month — battery, Sunamp, Zappi all saturated |
| May | ~250 (part-month Enphase Ph.1) | ~550 | +2× | Air-to-air reducing space heating draw |
| Jan–May total | ~550 | ~2,170 | +4× | |
| Full year 2025 | 2,580 | — | — | Partial year, original + Phase 1 from May |
The April 2026 export peak of ~890 kWh is the most important number in the solar dataset. Despite the Sunamp being charged daily, the 15 kWh battery being filled, and the EV charging on solar surplus, the system was still exporting at peak rate. This is exactly the load the planned air-to-water heat pump will absorb — running space heating on surplus solar in spring rather than exporting at low SEG rates. Every kWh shifted from export to heat pump self-consumption is worth 2–4× more at the avoided import rate.
The energy data: what the meters say
plus the first five months of 2026. This is unusually clean evidence of the transition in progress. A Sunamp 300e phase-change heat battery handles all domestic hot water — charged by surplus solar via the Myenergi Eddi from March through November, and by overnight cheap-rate electricity in winter. This means hot water has effectively been decoupled from gas entirely since the solar system was commissioned.| Period | Gas (kWh) | Electricity (kWh) | Notes |
|---|---|---|---|
| 2024 full year | 6,654 | 7,889 | Jan–Apr missing — meter data gap, unreliable |
| 2025 full year | 15,135 | 12,483 | True baseline. No heat pump. Gas includes hot water. |
| 2026 Jan–May | 3,976 | 4,960 | Air-to-air + solar running. Gas ~60% down vs 2025 equivalent. |
Before — January 2025
After — January 2026
The four-layer cascade: why it works
The strategic insight at the heart of this project is that a 7 kW air-to-water heat pump was always going to be undersized for a 9.06 kW house — but becomes correctly sized once the three highest-loss rooms are handled independently by the air-to-air unit. And a further layer already existed: a wood burner in the Living/Lounge, which the heat loss survey explicitly acknowledges as justification for accepting that room's radiator output at 89% of demand. The cascade has four tiers.
Removing the TV room (2,074 W), garden room, and one bedroom (approximately 750 W) from the wet system reduces the design demand to approximately 5.5–6 kW. A 7 kW air-to-water unit delivering 5.5–6.5 kW at -2.9°C on a 40°C flow design has a comfortable margin. On the coldest evenings, the wood burner supplements the lounge while the air-to-air handles the other demanding rooms. The wet system never needs to overprovide.
One room sits entirely outside this cascade: the en suite, at 92 W/m² and 775 W — the highest heat-loss density in the house — has electric underfloor heating already installed and was fully renovated, making wet system upgrades impossible. This is the one room running at COP 1.0 on electric resistance heat and is worth factoring into running cost expectations honestly.
This four-layer approach avoids the two failure modes most commonly seen in heat pump installations: undersizing (backup immersion heater running constantly) and oversizing (cycling on and off in mild weather). The HeatpumpMonitor.org dataset shows oversizing by 2–3× is one of the most consistent predictors of poor real-world SPF. The 7 kW unit here is sized to the actual wet system demand, not the whole-house survey figure.
The complete system
Installed and planned system components
Results so far
Air-to-air heat pump — real-world performance, Jan–May 2026 (Daikin app data)
SCOP calculated from Daikin app consumption and thermal output data. January–May 2026. The unit delivered 4.5 units of heat for every unit of electricity consumed — consistent with expected performance at typical UK winter temperatures of 0–10°C. The 2025 comparison (241 kWh consumed, 1,014 kWh output, implied COP 4.21) confirms the efficiency is stable across varying operating conditions.
The tariff: making the economics work
The energy tariff underpins everything. Rose Cottage runs on Intelligent Octopus Go — a time-of-use tariff with a night rate of 6.67p/kWh and a day rate of 27.75p/kWh, combined with Outgoing Octopus export at 15p/kWh. The Sonnen battery charges overnight at 6.67p and dispatches during the day, meaning the effective electricity cost for most consumption is closer to the night rate than the day rate. The January 2026 bill illustrates this precisely: of 1,437.7 kWh imported, 1,336.8 kWh (93%) was on the 6.67p night rate — a blended average of just 8.15p/kWh.
| Tariff element | Rate | Role in the system |
|---|---|---|
| Night rate (Octopus Go) | 6.67p/kWh | Battery charging, Sunamp top-up when solar insufficient, overnight EV charging |
| Day rate (Octopus Go) | 27.75p/kWh | Avoided almost entirely via battery dispatch and solar self-consumption |
| Export (Outgoing Octopus) | 15.00p/kWh | Spring/summer surplus — April 2026 ~£134 export credit alone |
| Standing charge | 41.74p/day | £152/year — eliminated entirely when gas disconnected (gas standing charge also removed) |
January 2026 net electricity cost: £136.57 imported minus £11.36 export credit = £125.21. The air-to-air heat pump consumed approximately 350 kWh in January, delivering ~1,400 kWh of heat at SCOP 4.50. At the effective blended rate of 8.15p/kWh, that heating cost approximately £29. The equivalent gas cost at 6.99p/kWh and 88% boiler efficiency would have been approximately £111 — an 74% saving on heating fuel cost, in the coldest month of the year.
What the data tells us about the remaining work
The 2025 gas data showed a summer floor of approximately 200 kWh/month from May through September — historically representing hot water only. That floor no longer exists. The Sunamp 300e charged by solar via the Eddi has eliminated gas hot water entirely from March through November, and the air-to-air heat pump has removed the remaining space heating gas load from the TV room, garden room, and bedroom. The 2026 gas data confirms this: from April onwards, gas consumption is effectively zero.
The only remaining gas demand is winter space heating for the wet radiator circuit. Once the air-to-water heat pump is commissioned, the gas supply will be disconnected entirely. This isn't just an efficiency decision — it removes the standing charge of approximately £10/month (£120/year) and eliminates any ongoing gas safety, servicing, and insurance obligations. Rose Cottage will be a fully electric home.
Projected annual energy cost post air-to-water, post gas disconnection: electricity import only, partially offset by solar export earnings, with self-consumption maximised through Homey load management. The boiler becomes redundant and will be decommissioned.
The wall problem that remains
The solid stone walls at U-value 1.20 W/m²K still account for the majority of the 3,801 W wall loss — approximately 45% of total heat loss. Everything else has been addressed or planned. Internal wall insulation on the worst-performing rooms would be the single highest-impact remaining intervention, potentially reducing whole-house demand by a further 1,500–2,000 W and removing the need for any top-up from the air-to-air on all but the most extreme days.
Testing the building: the boiler flow temperature experiment
One of the most useful things the owner did before committing to an air-to-water heat pump was to use the existing gas boiler as a test instrument. Once the air-to-air unit was installed and handling the TV room, garden room, and bedroom, the boiler's flow temperature was deliberately reduced from its default setting of 65°C down to 50°C — to simulate what the wet radiator circuit would feel like running at heat pump temperatures.
The result was instructive: at 50°C the house was still comfortable. The remaining rooms — dining room, living room, kitchen, bedrooms, bathrooms — maintained adequate temperatures even with the flow temperature 15°C lower than the boiler had previously been set to. The stone walls and existing radiators, it turned out, had significantly more output capacity than the boiler had been exploiting by running unnecessarily hot.
Further observation suggested 50°C was itself still higher than necessary. The fabric improvements — SF40BB and SF60BB in the roof, Rawington Nord triple glazing — had reduced the heat demand sufficiently that the remaining wet circuit rooms appeared to be well within heat pump territory: a design flow temperature of 40–45°C looked entirely achievable without any radiator changes at all.
This experiment is worth highlighting because it costs nothing and provides genuinely valuable evidence that no heat loss survey alone can give you. A survey tells you the calculated heat loss at design conditions. Running the boiler at reduced flow temperature tells you how the actual building — with its actual occupancy, actual fabric performance, and actual heat distribution — responds to lower temperatures in practice. The two don't always agree, and when they diverge it's usually the real-world data that's more useful for sizing a heat pump.
It also directly addressed one of the key risks in the original £27,000 quote: the concern that existing radiators would be inadequate at heat pump flow temperatures and would need replacing. The experiment suggested they wouldn't — which is why the planned air-to-water installation requires no new radiators and no pipe upgrades. The boiler, in its final months of operation, provided the evidence that made its replacement straightforward.
Why this approach: risk management over big-bang replacement
The conventional heat pump installation path in 2024–25 looked like this: accept a quote of around £27,000 for a full system — heat pump, cylinder, radiator replacements, pipework — surrender the gas boiler to claim the £7,500 Boiler Upgrade Scheme grant, and hope everything worked first time. In a stone cottage with a 9 kW heat loss and a TV room that the survey confirmed couldn't be adequately heated by wet radiators alone, that felt like an enormous single point of failure.
The owner's reasoning was straightforward: if a £27,000 system went wrong, or was misconfigured, or the radiators turned out to be undersized, you'd have no heating and no boiler to fall back on — having surrendered it to qualify for the grant. In mid-winter in rural Oxfordshire, that's not an inconvenience, it's a crisis. The BUS grant conditions effectively force an all-or-nothing commitment at the point of installation.
Instead, the approach taken was to decompose the problem and attack it in stages, each one derisk-ing the next:
The financial logic also stacks up differently when spread over time. Each stage delivered immediate, measurable returns before the next was committed to. Solar reduced bills from day one. The air-to-air cut gas consumption by 60% in its first heating season, verified by metered data. The fabric improvements benefit every heating system the house ever has. The £7,500 BUS grant is still coming — it's just being used to fund a clean, low-disruption final installation rather than a speculative whole-system replacement.
There is an honest caveat: this approach requires more time, more project management, and more technical curiosity than writing a single cheque. Not every homeowner has the appetite for it. But for those who do — particularly those with complex properties, high heat loss, or any doubt about a single-system installation — the staged, evidence-led approach produces a more resilient outcome and far more confidence in the final result.
Where this ends up
A fully electric home, by design not by accident
When the 7 kW air-to-water heat pump is installed and the gas supply is cut, Rose Cottage will be a fully electric home powered largely by its own generation. The journey from 15,135 kWh of gas in 2025 to zero gas will have taken roughly 18 months and a series of deliberate, reversible decisions — none of which required gambling the heating system on a single installation going right first time.
The numbers at that point will look approximately like this: ~15,000 kWh generated by solar annually, ~12,000–15,000 kWh consumed by the household including EV charging and heat pumps, electricity import costs offset substantially by export earnings and solar self-consumption, and a whole-house heating bill a fraction of the £1,503/year the gas system was costing. Carbon savings of 3.9 tonnes CO₂ per year against the gas baseline — and rising, as the grid continues to decarbonise.
What makes this case study useful for Decarbonarma readers isn't the technology — all of these products are available to anyone. It's the approach: treat decarbonisation as an engineering problem, not a purchasing decision. Get a heat loss survey. Understand your data. Fix the fabric. Solve the hard rooms directly. Let each step prove itself before committing to the next. And don't let a grant — however well-intentioned — push you into taking a risk you're not comfortable with.
Get the next one in your inbox
Clear, no-nonsense analysis of UK home decarbonisation — heat pumps, solar, EVs, batteries and the grid. Occasional emails, no spam, unsubscribe anytime.