Heat Pump Size Calculator
Estimate your home's heat loss on the coldest design day and the heat pump size worth looking at, using the official EPC default values and the design temperatures MCS installers must use.
What Size Heat Pump Do I Need?
Describe your home and we'll estimate its heat loss on the coldest design day, using the same default values an EPC assessor uses and the design temperatures MCS installers must use.
It's on your EPC certificate.
Estimated peak heat loss
6.3kW
At -5.1C outside, 20C inside
Heat pump size to look at
7kW
Smallest common size at or above your heat loss
Heat loss per sq m
70W
Lower is easier for a heat pump
What a 6kW to 8kW air source heat pump actually costs
£12,164
median installed price, before the grant
£4,664
after the £7,500 Boiler Upgrade Scheme grant
Median of 8,054 real installations in that size band in 2025/26. Source: DESNZ Boiler Upgrade Scheme statistics, August 2026 (installations with a paid voucher, England and Wales). The grant applies in England and Wales.
Where the heat goes
What would shrink the heat pump
- Fill the cavity walls: -1.5kW
- Replace the windows with modern double glazing: -0.6kW
- Top the loft up to 270mm: -0.3kW
An estimate, not a design. MCS installers must size a heat pump to cover at least 100% of a room-by-room heat load calculation to BS EN 12831-1:2017 (MIS 3005-D Issue 3.0), and that survey replaces this number. Our defaults: U-values from BRE's RdSAP 10 specification (June 2026), window area from RdSAP 2012 Table S4, design outdoor temperatures from MIS 3005-D Table 2 column B, a whole-house 20C indoors, a rectangular plan with 2.5m storeys, no thermal bridging allowance, and air leakage from SAP 10.2 with a minimum of 0.5 air changes an hour. Solid floors use the design temperature rather than the annual average MCS specifies, so that part errs high. Flats and homes with extensions or room-in-roof conversions will be less accurate.
The Short Answer
Most UK homes need a smaller heat pump than people expect. On the estimates above, a typical 1930s three-bed semi in the Midlands loses about 6.3kW on the coldest design day with unfilled cavity walls, and about 4.5kW once the cavity is filled and the loft topped up. A modern detached new build can be under 4kW. That points to a 5kW to 7kW unit for most semis, not the 10kW to 12kW boxes that still get quoted.
Use the number as a sense check, not a design. An MCS installer has to do a room-by-room heat loss calculation before sizing anything, and if their figure is wildly different from this one, ask them to walk you through it.
How This Calculator Works
It does a simplified version of the same sum an installer's heat loss survey does: for every wall, window, roof and floor, it multiplies the area by how easily heat passes through it (the U-value) and by the temperature difference between inside and outside on a cold day. Then it adds the heat carried out by air leaking through the building.
- U-values come from BRE's RdSAP 10 specification (June 2026), the official default values an EPC assessor uses when they can't see inside a wall. For example, an unfilled 1930s cavity wall is 1.5, a filled one 0.7, and a solid brick wall 1.7.
- Design outdoor temperatures come from the MCS heat pump design standard, MIS 3005-D Issue 3.0, Table 2. These are the temperatures installers must design to, from -1.5C in Plymouth to -5.9C in Glasgow.
- Window area uses the RdSAP 2012 formula for a typical home of your age and size, because most people don't know the square metres of glass they have.
- Air leakage uses the SAP 10.2 allowances for masonry or timber construction, unsealed suspended floors and open chimneys (80 cubic metres an hour each), with a floor of 0.5 air changes an hour, the minimum the MCS heat load calculator uses.
What it can't know: your exact room sizes, how much of the house is extension, a room in the roof, thermal bridges, or whether the previous owner did something odd to a wall. That's why the number is a range-finder, and why the survey matters.
Typical Results by House Type
All for the Midlands design temperature (-5.1C), at 20C inside, using the defaults above:
| Home | Peak heat loss | Per sq m | Size to look at |
|---|---|---|---|
| 2015 detached new build, 120 sq m | 3.7kW | 31W | 5kW |
| 1960s bungalow, filled cavity, 270mm loft, 75 sq m | 4.5kW | 60W | 5kW |
| 1930s semi, cavity filled, 270mm loft, 90 sq m | 4.5kW | 50W | 5kW |
| Victorian mid-terrace, solid walls, 100mm loft, 85 sq m | 4.8kW | 56W | 5kW |
| The same terrace with one open chimney | 5.5kW | 64W | 6kW |
| 1930s semi, unfilled cavity, 100mm loft, 90 sq m | 6.3kW | 70W | 7kW |
| 1960s detached, unfilled cavity, 100mm loft, 120 sq m | 9.6kW | 80W | 10kW |
Two patterns are worth noticing. Mid-terraces do well because they only lose heat through the front and back. And the biggest single upgrade in that table is filling a cavity wall, which on its own takes about 1.5kW off the semi. If you have unfilled cavities, do that first: it's cheap, and it may mean a smaller, cheaper heat pump. Our guide to insulating before a heat pump covers the order to do things in.
Why Bigger Isn't Safer
The instinct is to add a margin "just in case". The evidence says the UK already does far too much of that.
A report for the Committee on Fuel Poverty, published by DESNZ in May 2026, compared real monitored systems. In the government's Electrification of Heat trial, heat pumps were rated at a median of 2.07 times the home's inferred peak heat loss, and 54% were more than twice the size they needed to be. In the enthusiast-run HeatpumpMonitor.org dataset the median was 1.42 times. The report notes the trial figures probably overstate it, because many of those homes didn't heat overnight, but the direction is clear.
Oversizing costs you twice. A bigger unit costs more to buy, and it cycles on and off more in mild weather, which tends to lower efficiency. In the same report, HeatpumpMonitor systems oversized by less than 1.5 times averaged a seasonal efficiency of 3.99, against 3.38 for those oversized 2 to 2.5 times. The report says "emerging evidence" suggests the best results come from heat pumps sized at roughly 1.3 to 1.5 times the calculated peak heat loss.
The flip side: an undersized pump leaves you cold on the worst days and leans on an electric immersion-style backup heater, which is expensive to run. MCS rules handle this by requiring the heat pump alone to cover at least 100% of the calculated heat loss at the design temperature, without counting any backup heater.
Reading a Heat Pump's Size Properly
The number in the model name is a label, not a promise. A heat pump's real output changes with the outdoor temperature and with how hot the water it's making is (the flow temperature). When you compare quotes, ask for the unit's output at your design temperature and your design flow temperature, from the manufacturer's data. That's the figure that has to beat your heat loss.
For context, Vaillant's aroTHERM plus comes in 3.5, 5, 7, 10 and 12kW sizes, and Mitsubishi's R290 Ecodan in 5, 6, 8.5, 10 and 12kW. Most homes fall comfortably inside those ranges. The Boiler Upgrade Scheme has its own ceiling of 45kWth for a single system (70kWth for a cascade), which only matters for very large properties.
What To Do With Your Number
- If it's under about 6kW: you're in the sweet spot for a single small unit. Be suspicious of any quote for 10kW or more without a heat loss report that explains why.
- If your upgrades list shows a big saving: do the cheap insulation first, then get the survey. Sizing a heat pump for walls you're about to fill means paying for capacity you'll never use.
- If it's over about 10kW: your home is either large or leaky. It can still work well, but radiator upgrades and fabric improvements will matter more, and it's worth getting at least two surveys.
Every MCS quote should come with the room-by-room heat losses written down: MIS 3005-D says the customer must be given every room's heat loss in W per sq m. Our guide on how to compare heat pump quotes shows what else a fair quote includes, and the heat pump savings calculator estimates what it would cost to run.