Heat Pump, EV or Neither: Sizing for the Consumption You'll Have in 2030

Most home solar sizing starts with the same step: export twelve months of smart meter data, add it up, and pick an array that covers a sensible share of it. It’s tidy, it’s defensible, and for a household about to add a heat pump or an EV it is wrong in a predictable direction. You’d be sizing for a house that won’t exist by 2030.

This post argues for a different method. Keep your measured data as the base, then graft modelled heat pump and EV load on top, and size to the result. If you want the groundwork on cleaning and reading your own meter export first, the pillar guide Sizing From Your Own Consumption Data covers it. Here we assume you already have a year of half-hourly readings and want to push them forward.

Why today’s total is the wrong input

Take a typical three-bed semi using 3,400 kWh a year of electricity (Ofgem’s medium typical domestic consumption value is 2,700 kWh, so this is a household with a few more gadgets and someone working from home). Gas heats the house and the water.

Now add the two things a lot of owners plan within five years:

That’s not a tweak. Demand more than doubles. An array sized at 3.5 kWp to match 3,400 kWh would be roughly 40% of what the house actually uses in 2030.

The shape matters as much as the total

A heat pump doesn’t just add kilowatt-hours. It adds them in the wrong months. Solar generation in most of the UK is heavily summer weighted: in southern England a 1 kWp south-facing array yields roughly 950 kWh a year, but December might give 20 kWh per kWp and June 130 kWh. Heat pump demand runs the opposite way.

A rough monthly split for that 3,500 kWh heat pump load looks like this:

Month   HP kWh   Solar (4 kWp, S, 35°)
Jan      590       100
Feb      480       170
Mar      400       300
Apr      240       410
May      120       500
Jun       60       520
Jul       40       510
Aug       40       450
Sep       90       350
Oct      260       220
Nov      490       110
Dec      580        70

(Solar figures are indicative for a 4 kWp array; run your own postcode through PVGIS for real values.) The overlap is poor. Your array will export in June and sit nearly idle in January, which is exactly when the heat pump is hungry. Sizing on annual totals hides this, and a battery doesn’t fix it, because a 10 kWh battery shifts a day, not a season.

The EV is kinder. Charging is flexible, so you can aim it at midday in summer or onto a cheap overnight tariff (Octopus Intelligent Go, for instance, at around 7p/kWh off-peak versus a standard rate nearer 25p). Treat EV load as steerable and heat pump load as stubborn.

Building the forward scenario with tools you already have

You don’t need a consultancy. A spreadsheet and three free sources will do it.

1. Your base profile. Export half-hourly data from your supplier app or via the Octopus API, or pull it from Glowmarkt/Bright, or from an IHD-linked dashboard like Loop or Hildebrand. Aim for 12 months. Aggregate to hourly in a pivot table.

2. A heat pump profile. Don’t guess a flat daily figure. Use heating degree days for your location (degreedays.net gives free monthly and daily data for your nearest station, base temperature 15.5°C). Take your gas use, subtract hot water (about 2,000 to 2,500 kWh a year for a family), and convert the remainder to heat delivered using boiler efficiency of 85%. Divide by a seasonal COP. Spread each day’s result across hours using a simple two-peak shape (morning 06:00 to 09:00 and evening 16:00 to 22:00). Open-source tooling helps here: the heatpumps and when2heat datasets publish ready-made hourly profiles you can scale.

3. An EV profile. Miles per year divided by miles/kWh, divided by 0.9 for charging losses. Allocate it to the window you’d really use. Run it twice: once plugged in on arrival at 18:00, once timed for 00:30 to 04:30.

4. Generation. PVGIS (free, from the EU JRC) returns hourly output for your roof’s tilt and azimuth. Pick the same year as your meter data where it’s offered, so weather and demand line up.

Then, in the sheet or in a Python notebook with pandas, sum base + heat pump + EV hour by hour, subtract the solar from each candidate array size, and total import and export.

A worked example

Here’s the three-bed semi from above, with a 2.4 kW heat pump planned for 2028 and an EV in 2027. Costs assumed: import 25p/kWh, heat pump on a time-of-use tariff averaging 17p, SEG export 15p. Installed cost about £1,300 per kWp at this size, ignoring scaffolding step changes.

Marginal cost and return, going from 3.5 to 5.0 kWp: about £1,950 more for roughly £300 a year more, a 6.5-year payback on the increment. Going from 5.0 to 6.5 kWp: £1,950 for about £275 a year, around 7 years. Neither is bad for a system with a 25-year life, and the whole-system payback sits around 6 to 7 years. Compare that with sizing on today’s data. The same model against the 3,400 kWh house tells you 3.5 kWp is plenty and that 5 kWp mostly exports at 15p. You’d buy the small system, and in 2028 you’d be paying again, to add panels on a roof that may not have another good run of unshaded space.

The extra-panel cost is the point. Adding modules to an existing install later means a second scaffold hire or tower (£400 to £700), possibly an inverter swap if the original is maxed out, and a second MCS-style paperwork cycle if you want SEG eligibility on the new capacity. Paying an extra £1,950 now avoids most of that.

Roof space and inverter headroom are the real constraints

Run the model first, then check the roof. Typical 430 W panels measure about 1.72 m by 1.13 m. Twelve of them (5.2 kWp) need roughly 23 m² of clear area. If you only have 8 panel positions on the south face, ask whether east-west splits are worth it. They lose roughly 15% of annual yield versus south but flatten the day, which suits a house with a heat pump running morning and evening.

On the inverter, choose a hybrid or a string inverter with spare input capacity, and a battery-ready design. A 5 kW inverter happily takes 6.5 kWp of panels, and clipping losses at that ratio are usually under 3% in the UK.

Be honest about the “Neither” case

If no heat pump or EV is in your five-year plan, the forward scenario reduces to your current data, and the pillar method applies unchanged. Say that to yourself plainly, though, because “maybe” is not “neither”. If you’re in the 2030 gas boiler replacement window, or your lease on a petrol car ends in 2027, the probability is high enough that the model should include them. Run both: a “no change” case and a “full electrification” case. Choose the array that performs acceptably in both, which is usually closer to the larger one.

Keep the model alive after install

Once panels are up, your smart meter and inverter app give you real data to check the forecast against. Log monthly import, export and generation in the same sheet (Home Assistant with the Octopus Energy integration can do this automatically), and compare to the prediction. When the heat pump arrives, swap the modelled profile for measured data and rerun the sizing question for the battery. Models are cheap to rebuild; scaffolding isn’t.

Size the array for the house you’ll be running in 2030, and let the 2024 meter data be only the starting point.