← All insights Insight

What estates teams should check before replacing gas boilers with heat pumps

Jonathan Aird · 3 June 2026

Most heat pump schemes that go wrong were decided badly rather than installed badly. The decisions that matter are made early, often before a consultant is appointed, and they are made from information that is easier to gather than most people expect.

What follows is the list we work through at feasibility stage. Every item has caught something real on a recent project.

1. What the building actually uses, not what the boiler is rated at

Installed boiler capacity is a poor guide to heat demand. Boilers get oversized at installation, then oversized again at replacement because nobody wants to be the person who fitted one too small. On a recent school project the installed boiler capacity was 130 kW against a calculated peak load of 34 kW. Sizing heat pumps from the nameplate would have almost quadrupled the plant, the electrical connection and the cost.

Twelve months of half-hourly gas data, matched against outside air temperature, gives you the real answer. It is usually available from the supplier for the asking.

2. Whether your peak is genuine demand or a warm-up surge

This one is easy to miss even when the data has been gathered. On a museum project, the highest half-hour of the year came out at 404 kW — close to what the earlier concept study had assumed. But every one of the fifteen highest readings fell at either 05:00 or 07:30. They were warm-up surges created by the way the plant was scheduled, not demand from the building. At the coldest design condition, the building's steady demand was around 120 kW.

Sizing on the peak would have bought capacity the building never uses, and paid for it in capital, plant space and electrical connection. Look at when your peaks occur before you treat them as load.

3. What temperature your emitters actually need

Most existing systems were designed around 80/60 °C flow and return. Heat pump efficiency falls steeply as flow temperature rises, so the achievable temperature is the single biggest determinant of whether the scheme works financially.

The question is not whether the whole building can run cooler — it is which specific rooms fail first. In most buildings a minority of spaces are the binding constraint, and replacing emitters in those rooms only is dramatically cheaper than a wholesale change. That assessment needs a room-by-room look at emitter output at the temperature you intend to run.

4. Whether the electricity supply can carry it

On most retrofits this is the real constraint, not the mechanical solution. Three things need establishing, and they are not the same thing:

• Your maximum demand. Measured, not estimated. Four weeks of incoming demand monitoring at two schools recently cost £2,200 and showed one site's actual maximum demand was 28 kVA — meaning the new 190 kVA connection was being driven entirely by the heat pumps rather than by the school.
• Your connection agreement. What the network operator has agreed to supply.
• What the incoming cable is actually rated at. These are not always the same. We have found a site with a 400 kVA connection agreement where the incoming cable appeared to be rated at less than half that.

5. What the network operator will charge, and when you ask

Raise the DNO enquiry before the funding decision, not after. On the same two-school project the quotation came back with two routes: the network operator carrying out the contestable works at £81,464, or an independent connections provider doing the same work at £10,213. That £69,000 choice only existed because the enquiry had been raised early enough to act on it.

Lead times matter as much as cost. Where reinforcement is required, the connection can dictate the entire programme.

6. What you are doing about hot water

Hot water is usually the hardest part of a heat pump design and it is routinely left until after the plant has been selected. It needs a higher temperature than space heating, which is where efficiency goes to die, and it has legionella requirements that are not negotiable.

Check what your existing cylinders and coils were selected for. One school we surveyed had a 500 litre cylinder with a 40 kW coil sized for 80 °C primary flow. A lower-temperature heat pump will not deliver the same recovery through that coil without a larger heat exchanger, more storage, or a separate high-temperature arrangement.

7. Where the plant goes, and what it sounds like

Heat pumps need space, they need air to move freely around them, and they are noisier than what they replace — considerably noisier in cooling mode than in heating.

On a recent museum scheme the plant sat at about 33 dB(A) at the nearest home in heating, comfortably inside the council's 39 dB(A) night-time limit, but 46 to 49 dB(A) in cooling, which is 7 to 10 dB over. That was found at design stage, which is the only sensible time to find it. Space is equally unforgiving: two units on that roof needed 6.8 m against the 6.0 m the existing enclosure offered, and three units would have needed 9.2 m and could not have been made to fit in any arrangement.

8. What it will cost to run, honestly

UK electricity currently costs roughly 4.3 times as much as gas per kWh. A heat pump needs a seasonal efficiency above about 4.3 simply to break even on running cost, and most retrofit schemes land between 2.5 and 3.5.

That does not mean the schemes are not worth doing. It means the case has to be built on plant condition, carbon obligation and whole-life cost — and the running-cost consequence has to appear in the paper that goes to the board, not emerge in year two. On one museum project the honest conclusion was that running cost would be broadly neutral, and the case rested on carbon, the condition of the existing plant and removing gas from a listed building. That is a perfectly good case. It is just not a saving.

What this costs to establish

Most of the list above is data gathering and a survey. Half-hourly consumption data is free. Four weeks of demand monitoring is a low four-figure sum. A DNO budget enquiry costs a few hundred pounds. An emitter survey is a day or two of an engineer's time.

Against a scheme costing hundreds of thousands or millions, and against the cost of discovering any of it after the plant is ordered, it is the cheapest part of the project.