Space heating is decided by the building. Hot water is decided by the household — and it is the half a low temperature heat source finds hardest, because there is a temperature below which hot water cannot go.
This article covers Module 5 of the PlumbMate low temperature heating course: sizing a cylinder, why coil surface area decides whether a heat pump can use it, the reheat calculation, legionella control, and the loading unit method. There is a 10-question mock test at the end.
The floor under the flow temperature
Everything in Module 4 pushed the flow temperature down. Hot water pushes back.
Stored hot water has to reach 60 °C for legionella control, and the water in the coil must be hotter than that to drive heat into it. A heat pump running space heating at 45 °C therefore has to lift to 55 °C or more for the cylinder charge — at a considerably poorer COP — or an immersion heater has to make up the difference.
This is why hot water is often the larger share of a heat pump's running cost in a well-insulated house. The space heating has been made efficient; the hot water cannot be.
Two consequences for design. First, size the cylinder generously, because a single well-timed charge on cheap-rate electricity is far better than three top-ups through the day. Second, charge it once, at a chosen time, rather than keeping it hot on demand.
Sizing the cylinder
Two approaches, and use both.
By occupancy. Around 45 litres of stored hot water per person at 60 °C is a workable domestic figure. A four-person household is 180 litres.
By demand. Work out what the household actually draws. A bath is around 80 litres of blended water at 40 °C; a shower at 9 l/min for 8 minutes is 72 litres blended. Blended water is not stored water, and the conversion is where the interesting part lives.
A cooler store has to be a bigger store
Nobody uses water at store temperature. It is blended down at the outlet, so what matters is how much blended water each litre of store can make:
Vblend = Vstore × (Tstore − Tcold) ÷ (Tblend − Tcold)
That 80 litre bath at 40 °C, from a 10 °C cold main: out of a 60 °C store it costs 48 litres of stored water. Out of a 50 °C store it costs 60. Same bath, a quarter more cylinder.
| Store held at | Blend per litre stored | Volume needed, against 60 °C |
|---|---|---|
| 60 °C — boiler | 1.67 | × 1.00 |
| 55 °C | 1.50 | × 1.11 |
| 50 °C — heat pump | 1.33 | × 1.25 |
| 45 °C | 1.17 | × 1.43 |
This is the step that gets skipped. Every per-person figure you will see quoted is at 60 °C, and a heat pump does not hold 60 — it holds 45 to 52 and visits 60 for the pasteurisation cycle. Size from the 60 °C table and install the cylinder it gives you, and the household runs out.
So on our worked house — four occupants, one bathroom with a bath and shower, one en-suite shower — the answer is 210 litres on a boiler and 250 to 300 litres on a heat pump. Same house, same people, same peak. The difference is one decision about storage temperature.
Two effects, not one, and they are easy to run together: the coil is sized for the reheat rate, the volume for the storage temperature. A heat pump cylinder is not a boiler cylinder with a bigger coil.
The energy in a cylinder
Q = m × c × ΔT
where m is mass in kg (litres, near enough, for water), c is 4.18 kJ/kgK, and ΔT is the temperature rise.
For our 210 litre cylinder from 10 to 60 °C:
Q = 210 × 4.18 × 50 = 43,890 kJ
Divide by 3,600 to convert to kilowatt-hours: 12.19 kWh. Allowing for the fact that a cylinder is never fully cold — a realistic charge from 15 °C is 10.97 kWh.
Reheat time
time (h) = energy (kWh) ÷ power (kW)
The same 10.97 kWh, four different heat sources:
| Heat source | Power to the coil | Reheat time |
|---|---|---|
| Gas boiler, good coil | 18 kW | 37 minutes |
| Heat pump, correct coil | 6 kW | 1 h 50 |
| Heat pump, old 1 m² coil | 2 kW | 5 h 29 |
| Immersion heater | 3 kW | 3 h 39 |
The third line is the one that matters. That is not a smaller heat pump — it is the same heat pump, throttled by the cylinder it was connected to. Which brings us to the coil.
One honest caveat on that table: it holds the energy constant at 10.97 kWh so the coil is the only thing changing, which is the point being made. In reality the heat pump gets its own cylinder — larger, and charged to a lower temperature. Our 265 litre store from 10 to 50 °C is 265 × 4.18 × 40 ÷ 3,600 = 12.3 kWh, about two hours through a 6 kW coil. The bigger volume and the smaller temperature rise very nearly cancel, which is why a correctly sized heat pump cylinder does not take dramatically longer to charge than the boiler one it replaced. It is the coil that does that, as the third line shows.
The pasteurisation cycle is a separate, much smaller input on top: taking those same 265 litres from 50 to 60 °C is 265 × 4.18 × 10 ÷ 3,600 = 3.1 kWh, which on an immersion at a COP of 1 is what the weekly cycle actually costs.
Coil surface area
Heat transfer through the coil follows:
Q = U × A × ΔTm
The A is coil surface area and the ΔTm is the mean difference between coil water and cylinder water. A conventional cylinder on an 80 °C boiler has a large ΔTm, so a modest coil is enough — often 1.0 to 1.5 m².
Put 50 °C water into that same coil and the ΔTm collapses. To recover the transfer rate, the area has to grow. A heat pump cylinder typically has about 3 m² — two to three times a conventional one. Follow the heat pump and cylinder makers (MIS 3005-D).
This is the single commonest and most expensive mistake in a heat pump retrofit: keeping the existing cylinder because it looks serviceable, and finding the reheat now takes five and a half hours. The heat pump is fine. The cylinder is the bottleneck, and the only fix is to change the cylinder — after the work is done, at the worst possible time.
Check the coil area at survey. If the plate does not state it, treat it as inadequate.
Legionella
Legionella pneumophila multiplies between roughly 20 and 45 °C and is killed above 60. That range is uncomfortably close to a low temperature system's operating band, which is exactly why the control has to be deliberate.
The control regime, from HSE guidance (HSG274 Part 2, which the L8 ACoP refers to) and BS 8558:
Store at 60 °C. Not 55, not “usually about 58”.
Distribute so that any outlet reaches 50 °C within one minute (55 °C in healthcare). The 1999 Water Regulations guidance (G18.2 and G18.4) is older: it asks for distribution at 55 °C and 50 °C within 30 seconds where practicable. BS 8558 (2015) and HSE’s HSG274 use 50 °C within a minute, which is what you work to.
Keep cold water below 20 °C — insulate cold pipes running through warm spaces, and keep them away from hot ones.
Avoid dead legs. A capped branch to a removed appliance is a reservoir at exactly the wrong temperature.
Held at one temperature, taken to another
Read literally, “store at 60 °C” would rule out every heat pump and every solar thermal system ever installed. It does not, because it describes the temperature the water has to reach, not a temperature the cylinder must sit at every hour of the day.
On a low temperature source the store is held at a temperature that suits the source and taken to 60 on a schedule. That is the design, not a workaround for a weak appliance:
| Source | Store held at | Raised to 60 °C by |
|---|---|---|
| Heat pump | 45–52 °C | Immersion, or the unit's own high-temperature mode, on a timer |
| Solar thermal | Whatever the resource gave you — 25 °C on a dull February day, 70+ in July | The auxiliary heat source, which has to guarantee the cycle regardless of the weather |
| Condensing boiler | 60 °C continuously | Nothing — it is already there |
Solar thermal is the clearest case. The store temperature is set by the sun, so it is never a reliable control measure on its own. The auxiliary is what makes the installation safe, and it has to be scheduled rather than left to whether the collector had a good day.
Two things make this a real control measure rather than a comfortable story. The cycle has to raise the whole store, not the top third above the immersion — which is what an anti-stratification pump is for, running during the cycle only. And it has to be scheduled: on cheap rate, part of the commissioned control strategy, not something the householder is expected to remember or free to switch off to save money.
How often? It is a risk assessment
Weekly is the usual domestic answer. But L8 and HSG274 are a risk-assessment framework, not a rulebook, and the assessment is supposed to reflect who actually uses the building — so an elderly, immunocompromised or respiratory-impaired occupant justifies a tighter regime than a household of thirty-year-olds.
Be careful which way you let that argument run. Going stricter is easy to justify. Going looser than the standard regime is not, in a house: there is no sampling, no monitoring, and nobody competent checking. A hospital can argue for a varied regime because it has a water safety group and monthly samples; a semi with a heat pump has none of that, and the occupants change without telling you. Weekly as the baseline, tighter where the household warrants it, and looser only on the back of a written assessment by somebody qualified to sign it.
And Part G
Storing at 60 °C and delivering at 60 °C are different things. Part G requires that bath hot water does not exceed 48 °C, which means a thermostatic mixing valve — a TMV2 valve (the domestic scheme) on the bath draw-off, set and recorded at commissioning.
Store hot for safety against legionella; blend down for safety against scalding. Both, not either.
Loading units
Pipe sizing for hot and cold water uses a different method from heating, because outlets are not all used at once. BS EN 806-3 defines one loading unit as a draw-off flow rate of 0.1 l/s, so a fitting's loading units are simply its design draw-off rate in tenths of a litre per second.
| Draw-off point | Draw-off flow rate | Loading units |
|---|---|---|
| Washbasin, handbasin, bidet, WC cistern | 0.1 l/s | 1 |
| Kitchen sink, washing machine, dishwasher, shower head | 0.2 l/s | 2 |
| Urinal flush valve | 0.3 l/s | 3 |
| Bath, domestic | 0.4 l/s | 4 |
| Garden or garage tap | 0.5 l/s | 5 |
The bath at 4 dominates, because it is the one fitting that draws hard and continuously.
Then the part that catches people out: there is no conversion to a flow rate. Add the loading units along the run and read the pipe size straight off a table chosen by material. The probability of simultaneous demand is already inside that table — which is exactly what makes it the simplified method.
Our semi's hot side comes to 10 loading units: a 4 LU bath, a 2 LU shower head, two 1 LU basins and a 2 LU sink. On copper that is 18 mm, and in Britain you fit 22 mm, because 18 mm is a European size and going up one is always allowed. Ten loading units is a litre a second of draw-off capacity being carried by a 16 mm bore — and that gap is the diversity the whole method exists to capture.
One warning worth more than any of the numbers. The older BS 6700 and BS 8558 tables also use something called a loading unit, on a completely different scale — a bath is 10 there, not 4 — with its own conversion to a design flow rate. Each set of values belongs to its own method. Take a figure from one and read it off the other's chart and the pipe comes out wrong.
Why oversized hot water pipe is a fault
On heating, an oversized pipe is inefficient. On hot water it is a defect, for three reasons.
Every litre sitting in the pipe is a litre that must be run off before hot water arrives. Oversize the run and you get a longer wait, wasted water, and a customer complaint.
Larger volume means slower velocity, more time in the 20 to 45 °C band, and a better environment for legionella.
And the standing loss from the pipe's own contents, cooling between draw-offs, is proportional to that volume.
Hot water pipework should be sized to the calculation and run as short as the building allows. “One size up to be safe” is the wrong instinct here.
Secondary circulation
Where a draw-off is too far from the cylinder to deliver hot water quickly, a secondary return with a small pump keeps the loop warm. It solves the wait, and it costs continuous heat loss from the whole loop — so insulate it properly, put it on a time clock, and only fit it where the run genuinely justifies it. Its return should come back at 50 °C or more (55 °C in healthcare). On a heat pump store held at 45 to 52 °C, avoid the loop if you can: shorten the run, or fit trace heating or a point-of-use heater at the far outlet. If you cannot, the risk assessment sets how often the store and the loop go to 60 °C. The loop sits below 50 °C all the time, so it will usually call for more frequent pasteurisation than weekly, with the loop pump running through the cycle.
On a heat pump system it deserves particular thought, because that continuous loss is being made up at a poor COP whenever the cylinder recharges.
📝 10-Question Mock Test
Click an option to see whether you got it right. Explanations appear instantly — no submitting at the end.
It has to reach 55 °C or more for the charge, at a considerably poorer COP. In a well-insulated house hot water is often the larger share of the running cost.
So 180 litres for four people — and more where the source is a heat pump, because one well-timed charge beats three top-ups through the day.
Blended water is not stored water. Treating the two as the same oversizes the cylinder substantially.
The 0.33 is the volumetric heat capacity of air, used for ventilation loss, and 3,600 is the number that converts kJ to kWh.
Energy divided by power. The same charge through a 2 kW coil takes five and a half hours — the same heat pump, throttled by the cylinder.
Two to three times a conventional cylinder's. With 50 °C water the mean temperature difference collapses, so the area has to grow to recover the transfer rate.
The commonest and most expensive mistake in a retrofit. The only fix is to change the cylinder — after the work is done, at the worst possible time.
Uncomfortably close to a low temperature system's operating band, which is why storage at 60 °C and a weekly pasteurisation cycle have to be designed in deliberately.
Store hot against legionella, blend down against scalding. Both — a TMV2 valve (the domestic scheme) on the bath draw-off, set and recorded at commissioning.
Part G sets the 48 °C ceiling at the bath. It says nothing about how quickly hot water arrives: that is guidance from BS 8558 and HSE, and the figure is 50 °C within one minute, not 30 seconds.
Adding individual flow rates instead produces pipework two sizes too big — which on hot water means a longer wait, wasted water and a better environment for legionella.
Hot water is where a low temperature system meets a temperature it cannot negotiate with. Size the cylinder generously, check the coil area before anything else, and charge it once a day on a schedule.