The Heat a City Cannot See

This map is live. Open the city view to choose the scenario and read every ward yourself.

Last week we published a small calculator: what does the pipe between a heat pump and the house cost, one home at a time. The honest answer was modest, twenty-six pounds a year on the default settings, and one reader after another had the same reaction we did. That cannot be the whole story.

It is not. To see the rest of it, you need a whole city.

The envelope

We took the 107,336 homes of Nottingham that our stock model can price, every home compiled from the public record and simulated through the national calculation for a full year, hour by hour. Then we asked one question as a scenario envelope: if each of these homes had an external heat pump run of a stated length and insulation state, what would that pipe shed in a year, using each home's own simulated heating hours rather than anyone's assumption?

The pipe cases are explicit scenarios, stated in full in the method note. Nothing here claims to know any individual home's pipework. What the envelope shows is the size of the thing the official figures cannot show.

At ten metres of bare pipe per home, a long siting, the city's loss would be 142 GWh a year, nearly twelve percent of everything these homes spend on space heating. At five metres, the more typical siting, 71 GWh, six percent. Insulated to 25 mm throughout, the ten metre case falls from 142 GWh to 23. The swing between bare and done properly, across one city, is 119 GWh a year: ten percent of the city's entire space heating, roughly the space heating of ten thousand homes.

The map above is not uniform, and its unevenness is information. Hyson Green & Arboretum alone would shed 9.3 GWh a year in the bare scenario; the quietest ward a third of that. Density does some of that work, but so do heating hours: the median home in Castle ward heats for 2,732 hours a year, in Berridge for 4,834. A city is not one number, and neither is a street.

None of it appears in any compliance figure, for any home, anywhere. The national calculation models no pipework loss on emitter circuits at all; where pipework is in scope, the insulation is declared and never verified. A city can shed the heating of a town through a component no instrument of policy can see.

The correction we owed our own calculator

The calculator we published assumes a flat 1,800-hour heating season, the standard textbook figure. The city's records say otherwise. Across 107,336 simulated homes, the median heating year is 3,592 hours. One home in ten heats for more than 5,175.

The assumption fits about one home in forty: 2.6 percent of the city heats for fewer than 1,800 hours. The standard assumption understates the median home's pipework loss by half. Our own page carried that assumption, stated openly in its method note, and the city run has now measured its conservatism: the median ten metre bare run is not 629 kWh a year but 1,255, and for the tail that heats longest, nearer 1,808. We have updated nothing quietly. The page said what it assumed; the city said what is real; both statements stand, and the gap between them is exactly why a record kept per home beats a flat assumption.

Where this leaves you

If you look after housing stock, the question is not whether your homes lose heat this way. It is that nobody currently records whether they do. The envelope above is a scenario, deliberately so, and the record that produced it can read any real stock the same way, from public data, before anyone visits. The view for one home prices one pipe; the crossing queue shows one estate; and this essay's city run is now a view of its own: the same record at the scale of a city, live, ward by ward.

About the numbers

• Nottingham: 107,336 homes eligible for analysis, from 107,894 public energy certificates, each simulated through the national calculation, full year, hourly. Active heating hours are each home's simulated hours with nonzero space heat demand: P10 2,474, P50 3,592, P90 5,175.
• The pipe cases are a scenario envelope over real homes, with the pipe assumptions stated: run lengths of 5 m and 10 m; bare 28 mm copper, 13 mm elastomeric (k=0.037), 25 mm phenolic (k=0.026); water at 45°C flow against 5°C ambient; the published pipe loss formula of the national calculation's technical reference.
• Cohort annual space heat 1,195 GWh. Scenario cells: 5 m bare 71.2 GWh; 5 m at 13 mm 22.0; 5 m at 25 mm 11.5; 10 m bare 142.4; 10 m at 13 mm 44.0; 10 m at 25 mm 23.0. The swing from bare to 25 mm is 5.0% of cohort space heat at 5 m, 10.0% at 10 m.
• The run lengths are stated scenarios, not a survey. No national statistic of external run lengths exists; installer guidance keeps runs as short as practicable, so 5 m is the more typical siting and 10 m the long one. The formula is linear in length: city wide, about 14.2 GWh a year per metre of bare run.
• Ward boundaries are the ONS Open Geography wards (Open Government Licence); homes were joined to wards through the ONS Postcode Directory, May 2026 release, with all 107,336 homes matched. No ward reported holds fewer than ten homes.
• Every figure was recomputed twice by independent routes (272 independent checks, counts exact, heat within 0.001 kWh) and the formula implementation was checked against our published calculator to within 0.000001 kWh. Aggregates only; no figure for a single dwelling leaves the analysis.
• Illustrative scenario envelope, not a forecast, and not a claim about any individual home.