A Year in Our Retrofit: Predicted vs Real
Notes on living in my own deep retrofit, a Victorian terrace in south east London, with a full year of measured energy data.
We moved back into A-Typical Terrace in October 2024, after a deep retrofit and small side extension. 2025 was our first full year in the house. Before the works it was a draughty, dark Victorian terrace with gas heating. After, it’s all-electric: the gas meter has gone.
We designed the project using a PHPP (Passive House Planning Package) model. Essentially it is a massive, very clever spreadsheet linked to the 3D model of the house. But an energy model tells you what a building should do. Living in it tells you what it actually does. This is an honest look at both: where the house beat the predictions, where it didn’t, and one discovery about terraced houses that I think many retrofit models get wrong.
The house in numbers
- Victorian mid-terrace, south east London
- Floor area: 101 m² treated floor area (112 m² gross internal)
- Wood-fibre internal and external insulation at the back, only internal insulation to the street
- Well-insulated floors and roofs
- Triple glazing
- 5kW air source heat pump for heating and hot water
- Mechanical ventilation with heat recovery (MVHR)
- 4.73kWp of solar panels across two roofs, with a 5kWh battery
- Airtightness: 4.94 m³/(h·m²) at 50 Pa (n50 = 4.7/h)
- EPC A, and the AECB CarbonLite ‘Step One’ retrofit standard
- Near EnerPHit standard, meets all ‘By Component method’ requirements except airtightness.
Predicted vs measured
The house was modelled in PHPP, the Passivhaus energy software. Here is what it predicted, against what we used in 2025.
| Electricity (kWh/year) | PHPP predicted | 2025 measured |
|---|---|---|
| Space heating | 2,112 | 709 |
| Hot water | 910 | 759 |
| Everything else | 1,616 | 2,714 |
| Total | 4,638 | 4,182 |
| Solar generated | 4,068 (installer estimate) | 4,084 |
Our total use works out at 37.3 kWh per m² per year (gross internal area). For comparison, a typical UK home uses well over 100 kWh/m²/year once gas is included.
Three things stand out. Heating used about a third of what was predicted. Hot water came in close to the model. And “everything else” (cooking, appliances, lighting, the MVHR and so on) used much more.
Heating: a third of the prediction The heat pump used 709 kWh of electricity for space heating in 2025*, against a predicted 2,112 kWh. That’s a surprising headline, and there are several reasons behind it.
*Figures reported by the heat pump app.
The first is the heat pump itself. We didn’t have the performance data needed to model our specific heat pump in PHPP, so we used a “standard heat pump” instead. We didn’t realise at the time that it assumes a COP (coefficient of performance) of around 2.5. Our heat pump currently reports a seasonal efficiency (SCOP) of 3.6, nearly one and a half times better.
But that’s not the whole story. PHPP predicted an annual space heating demand of 52 kWh per m², roughly 5,250 kWh of heat. At a COP of 2.5, that meant around 2,100 kWh of electricity. We actually used just 709 kWh, around a third of the prediction.
At the heat pump’s current efficiency of 3.6, 709 kWh of electricity is roughly 2,550 kWh of heat: about half the predicted demand, or around 25 kWh/m². That puts the house right on the EnerPHit certification criterion for heating demand, in practice. And that’s despite our airtightness being nearly five times EnerPHit’s stringent target of 1.0 air changes per hour. Even at the model’s lower efficiency, the house needed only about a third of the heat it was expected to.
So where did the rest of the gap come from? I think part of the answer lies in how airtightness is measured in a terraced house.
Party wall airtightness question
We lose heat through a building’s envelope in two main ways: by conduction through the walls, roof, floor and windows, and by infiltration, as air leaks in and out through gaps. To measure how leaky a house is, we use an airtightness test. A fan pressurises (and depressurises) the whole house and measures how much air escapes at a range of pressures. The industry standard is to report the result at 50 pascals, roughly the pressure of a 20 mph wind blowing directly against the house. From this, the energy model estimates how much heat is lost through leakage.
But the test can’t tell where the air goes. In a detached house it all goes outside, taking heat with it. In a terrace, some of it goes through the party walls into the neighbours’ homes.
We didn’t do airtightness work to our party walls. It would have been costly and meant taking apart more of the building than we otherwise needed to, for little benefit. So I suspect a good share of our measured leakage (4.94) is air passing between our house and our neighbours’: two heated homes at similar temperatures. Air moving between them carries very little heat away.
PHPP treats the test result as if every bit of leakage goes to the outside. For a terraced house, I think that overstates the real heat loss, and so overstates the heating the model predicts. Re-run with EnerPHit-level airtightness, PHPP predicts a heating demand of 36 kWh/m², down from 52. Our estimated result of around 25 is lower still, so the party walls can’t be the whole story. They may account for up to around 60% of the gap.
The rest probably comes from a mix of things:
- We probably keep the house a couple of degrees cooler than the 20°C PHPP assumes, and 2025 had a mild winter.
- We have an electric heated towel rail in the bathroom, which doesn’t appear in the heat pump’s figures at all. Its electricity is counted under “everything else,” but it is still heating the house.
- The rest of that extra “everything else” electricity works the same way, more quietly. Nearly all of it, from the fridge, cooking and devices, ends up as heat inside the house, and in winter that helps keep it warm.
- There is some uncertainty in the heat pump’s efficiency figure.
The practical lesson for anyone retrofitting a terrace is to focus airtightness effort where heat really escapes: the front and back walls, the roof, the ground floor, and around windows and doors.
Hot water: close to the model
Hot water used 759 kWh against a predicted 910. That is close, which makes sense: hot water use depends mostly on how many showers and baths a household takes, and much less on the building itself.
Everything else: the mystery
The surprise is the rest. Heating and hot water together came to 1,468 kWh, leaving 2,714 kWh for everything else: cooking, the fridge and freezer, washing, lighting, ventilation and all the small devices that are always on. The model expected around 1,600 kWh for those.
In a well-insulated house, this “unregulated” energy becomes the biggest part of the bill, and it’s the part building regulations and most models pay least attention to. To track down that unusually high energy use I’m now building a home energy monitoring system. I’ll report back in a follow-up.
Solar, battery and bills
Our 11 solar panels sit on two roofs: six facing west, and five facing almost due south. The installer estimated 4,068 kWh a year. In 2025 they generated 4,084 kWh - amazingly within half a percent of the prediction.
By sheer coincidence that is very close to the 4,182 kWh we used. So over the year, the house generated nearly as much electricity as it consumed. That doesn’t mean we were off-grid: we mainly import in winter and mainly export in summer. But the annual balance is close to zero.
But what does that mean for our bills? Well, that depends on the tariff you are on. Most tariffs pay you less for energy you export than what you pay to import the energy, but we’re on an unusual tariff. We’re on Octopus’s Intelligent Octopus Flux tariff, which pays the same rate for export as import. The catch is that you have to have a compatible battery, and that Octopus manages the battery, charging and discharging it to suit the grid. The other trick with Intelligent Flux is that it has a ‘normal’ and ‘peak’ (4-7pm) rate, each matching import and export.
The system intelligently fills up the battery either from solar or from the grid depending on the weather, and exports it at peak time. This provides the grid with energy when it needs it, and earns us a little money in the process. Because we exported nearly as much as we imported, and we also export for a profit, our monthly bill is just the standing charge. At 40p/day, we currently pay £15 a month, which is a bit more than we should - I plan to ask for a refund once winter is through, as the account is very healthily in credit!
Summer: the other side of the balance
PHPP doesn’t only count the heat a house loses. It also counts the heat it gains: sunshine through each window, depending on its orientation, size, glass and shading, plus heat from the people, cooking and appliances inside. In winter those gains are welcome. In summer they need somewhere to go. PHPP weighs them against how much heat the house can shed through ventilation and open windows, and estimates the “frequency of overheating”: the share of the year the inside goes above 25°C. For our house it predicted 0%.
Then came two record summers. 2025 brought four heatwaves and was the UK’s warmest summer on record, with London reaching 34.7°C on 1 July. 2026 broke that record again, with Kew Gardens reaching 38.1°C on 13 August.
In the hottest spells the house was warm, but I still slept fine. My new energy monitoring system will also record the indoor temperatures in some key rooms. We used night-time cooling and kept the west-facing curtains closed. The house stayed comfortable because the same insulation that keeps heat in the house during winter slows the heat getting in during the summer.
There’s a limit to what the model can see, though. PHPP treats the house as a single zone, an average across every room. It can’t pick out one room catching the low afternoon sun. Our first-floor west-facing windows noticeably heat up the room in the late afternoon, so before next summer we plan to add external shading to them. It has to be external: blinds inside the glass only stop the sun once its heat is already in the room.
What it’s like to live in
The numbers matter, but how the house feels matters more.
It’s comfortable, and it holds its warmth. In winter the house takes a very long time to cool down: days or weeks rather than hours.
It’s quiet. Triple glazing and good airtightness to the street mean we can barely hear cars passing. It’s a quiet road, but busy with people walking to the shops around the corner.
The ventilation is better than I expected. I had never lived in a house with MVHR before. Not having trickle vents is a big improvement for noise and cold draughts. The air is never stuffy, even in midwinter, and we keep the windows completely closed from October to April.
We’ve learnt to let it run itself. Rather than switching the heating off when we go out, we keep the house at a steady temperature all day. A heat pump in a well-insulated house is slower to react, but it is also much slower to cool down, so a steady setting works best.
A note on the data
These figures come from one year, and one year’s weather. The heating and hot water figures are the heat pump’s own readings of the electricity it used. Total consumption and solar generation come from the solar system and our smart meter. I don’t have the heat pump’s seasonal efficiency (SCOP) for 2025, though it currently reads 3.6. A more detailed monitoring system is now running, and I’ll update this article when it has a full year of data.
Summary, in a few words
- A deep retrofit of an ordinary terrace can perform very well. Our heating used a third of the PHPP-predicted electricity, and the house generates about as much as it uses over a year.
- In terraces, think carefully about party walls. Airtightness tests and models may overstate heat loss through them. Put the effort into the outside walls, roof and floor.
- Once heating is tamed, everything else dominates. My next mission is to track the energy used by appliances and always-on devices, and further lower our energy use to hopefully reach ‘zero bills’.
- Plan for summer, not just winter. Check where the afternoon sun hits, and shade it from the outside.
- Comfort is the real prize. Quiet, fresh, steady warmth matters more than any number or accredited certificate.
Thanks go to Patrick Osborne for building our PHPP model
Read more about the project here
If you’re thinking about retrofitting an older home, I’d be glad to talk it through. Get in touch
Contact
Not sure where to start? Book an Old House Health Check
︎Contact Form
︎ info@tparsons.co.uk
︎ Instagram
