Choosing a Cylinder a Heat Pump Can Actually Heat: Coil Area Before Litres

The machine puts a ceiling on the coil before the coil is even measured

Whatever the coil could accept, it will never be handed more than the outdoor unit is making on the day, and that figure is not the one on the box. Nameplate capacity is quoted at a mild rating point — 7 °C air under BS EN 14511 in Europe, 47 °F in North America — and an air source machine loses output as the air it is drawing from gets colder. Take an 8 kW unit into a −3 °C design condition at a couple of percent per kelvin and there is about 6.4 kW left. That is the whole ceiling: space heating, hot water, everything.

Hot water mode lowers it again, and this part gets skipped. Charging a store to 48 °C means running a flow temperature above it, typically 50 to 55 °C, well beyond the 35 to 45 °C the space heating circuit was designed around — and capacity falls with rising flow temperature as well as with falling air temperature. The hot water figure at the design condition is therefore lower than the space heating figure at the same condition. It sits in the manufacturer's capacity table at the intersection of the two, it belongs in every calculation downstream, and there is no rule of thumb worth using in its place.

Then the consequence nobody costs. Domestic installations almost universally give hot water priority: while the diverter is over on the cylinder, the house is not being heated at all. On a condensing boiler that mattered little, because the cylinder was back up in twenty-five minutes. On a heat pump a ninety-minute charge is ninety minutes of a February morning with the emitters cooling, and the recovery afterwards is slow because everything about the machine is slow. This is what quietly settles the coil question: reheat duration is not a hot water inconvenience, it is a heating outage, and you shorten it with coil duty rather than with litres.

One clarification before the arithmetic, because the misreading runs the other way too. Priority operation means the hot water load is generally not added to the space heating peak when the machine is selected: MCS MIS 3005 and ACCA Manual S both size against the calculated design heat loss rather than the boiler that came out. Hot water does not make the machine bigger. It makes the time it spends away from the radiators longer, which is a different problem with a different fix.

The same coil, a quarter of the driving temperature

A cylinder coil is a heat exchanger and obeys the arithmetic of one: the heat it passes is its transfer coefficient times its surface area times the log mean temperature difference between the primary water inside it and the store outside. Two of those three are fixed by the object in the cupboard. Only the third changes when the boiler goes, and it changes by a factor that surprises anyone who has not written it down.

Work the two ends. A boiler running 80 °C flow and 60 °C return into a store averaging 40 °C during a reheat is working across differences of 40 K at the coil inlet and 20 K at its outlet, a log mean of 28.9 K. A heat pump running 50 °C flow and 45 °C return into the same store is working across 10 K and 5 K, a log mean of 7.2 K. Four to one, on the same coil, in the same cylinder, with the same water in it. A coil rated at 15 kW on the first will pass something under 4 kW on the second, and no amount of running the pump harder alters it, because the limitation is a temperature difference rather than a flow.

The table applies that ratio across the flow temperatures a changeover actually meets. Read it as a proportion rather than a rating — the real numbers belong to the specific cylinder — and read it as the optimistic version. The transfer coefficient is not genuinely constant: the store side of the coil works by natural convection, which weakens as the difference driving it falls, so the collapse at the bottom of the table is somewhat worse than proportional. Higher primary flow rates recover a little of it. Neither correction changes the conclusion by anything like the factor of four.

The last row explains the 44 °C store at seven in the morning. As the store approaches the flow temperature the difference driving the transfer approaches nothing, so the final few kelvin take a disproportionate share of the charge. A heat pump at 50 °C flow does not reach a 50 °C store; it asymptotes toward it and the controller gives up somewhere short. Hence hot water targets set around 48 °C rather than at the flow temperature, and a periodic high-temperature cycle run by the immersion element or by deliberately lifting the flow temperature, rather than by asking the coil to close a gap it cannot close.

So the number to obtain, from the manufacturer rather than from the cylinder, is the coil rating in kilowatts at a stated primary flow temperature and flow rate. Boiler-era literature rated coils at 80 °C primary because there was nothing else to rate them at. Cylinders sold for heat pumps publish the figure at 50 or 55 °C and carry coils of around 3 m² where the old one had 0.8 — precisely the factor of four the arithmetic demands, arrived at independently by an industry that had to solve this in production. Where the area is unknown and the maker no longer exists, that is itself the answer: an unquantifiable heat exchanger in the middle of a system you are paid to make work is a replacement, not a survivor.

Litres are the easy half of the question

Volume still has to be settled, and the household-size bracket is a reasonable place to start an argument even though it knows nothing about your store temperature or your heat source. Take its range, convert it out of US gallons, then do the thing that makes it useful on a heat pump job: turn the volume into the coil duty it implies. Energy to reheat a store is volume times 4.186 times the rise, divided by 3,600 for kilowatt-hours; divide that by the window you are willing to lose the heating for, and the answer is the kilowatts the coil has to pass.

The table does it for a full charge from a 10 °C cold feed to a 48 °C store — the worst case, with the vessel drawn right down and no standing loss allowed for. Set the results beside the derated machine output from two sections up and the binding constraint reveals itself. An 8 kW unit with 6.4 kW available cannot reheat 250 litres in an hour whatever coil is fitted, because the heat is not there to be transferred. Below about 180 litres the coil is the limitation; above it the machine is, and buying a bigger cylinder past that point simply lengthens the outage.

Which is the trade, stated plainly: litres buy the ability to serve a draw, coil duty buys the ability to be ready for the next one, and on a heat pump the second is scarce in a way it never was on a boiler. There is a third cost on the volume side that boiler-era habits ignore. A cylinder's standing loss is declared in watts on its energy label under the EU water heater and storage tank regulations, and it runs continuously: a label reading 65 W is 1.6 kWh a day into the airing cupboard against a household demand that might be 5.6 kWh. Oversizing the vessel is not free, and the legionella guide argues the store down as well, on residence-time grounds worth reading alongside this.

One thing this section deliberately does not do is convert litres into showers. That conversion depends on the hot fraction — how much stored water at 48 °C a blended 40 °C shower actually consumes, which is more than it consumed from a 60 °C store — and it is worked through properly in the shower guide, with the table for it. Take the number from there and bring it back.

Three or four minutes of reheat for every minute of shower

There is one number that makes the whole balance intuitive, and it comes from expressing a hot water draw the way an instantaneous appliance is rated: as a continuous kilowatt demand. A nine litre a minute shower delivered at 40 °C from an 8 °C winter main is 20 kW, and it makes no difference to that figure whether the heat arrives from a burner, an element or a store. Two of them running together is 40 kW. Against a coil passing 5.5 kW, the ratio is what it is.

Turn it into time and it becomes a design tool. Eight minutes of that shower removes 2.7 kWh from the store, and 2.7 kWh at 5.5 kW is twenty-nine minutes of reheat — roughly three and a half minutes of charging per minute of showering, with the heating off throughout. A 140 litre bath is worse: about 5.2 kWh in one draw, near enough an hour of recovery behind it. That ratio, rather than any rule about litres per person, tells you whether a household's actual morning fits inside the cylinder you propose to keep. Three people and twenty-five minutes between the first shower and the last is an event that must be served entirely from storage, because the coil contributes almost nothing inside the window.

The point of running it in kilowatts is that it makes the storage compulsory rather than optional, and that framing settles a question customers ask constantly on changeovers. A heat pump cannot heat water instantaneously — there is a factor of three or four between what a shower asks for and what the machine makes — so the cylinder is not a legacy component that a newer technology might eliminate. It is the buffer that makes the machine usable at all, and the coil is the rate at which the buffer refills. Whether an instantaneous appliance is the right answer somewhere in the house, and how the product families compare on flow and pressure, is a separate decision covered in the shower guide.

The secondary return that flattens what the coil just built

Houses with a long run to an en-suite acquire a secondary hot water return, sometimes as part of the same works. On a boiler it was a comfort item with a modest running cost. On a heat pump cylinder it is two problems at once, and the first is arithmetic. A modest domestic loop losing 300 W continuously is 7.2 kWh a day, against a household demand that might be 5.6 kWh — the loop costing more than the hot water it delivers, with the heat pump carrying all of it, including through mild afternoons when the machine would otherwise be off. Insulating the loop is not a finishing item on this kind of job, and timing the pump to the hours the household actually uses recovers most of the rest.

The second problem is stratification, and it is specific to the lower store temperature. The return enters part way up the cylinder and mixes as it arrives. On a 60 °C store that cost you some of a large margin. On a 48 °C store serving a 40 °C shower the entire usable band is about eight kelvin wide, and the coil — working by then at the bottom of that table two sections up — is slow to rebuild anything it loses. A loop running continuously into a heat pump cylinder can hold the vessel in a permanently half-charged state that looks like a fault and is not. Where a return is genuinely required, the cylinder wants a dedicated return connection at the right height rather than a tee into the cold feed, and the loop wants a control strategy rather than a switched live.

Two things about that loop are settled elsewhere: the temperature regime the return has to meet and the monitoring that proves it, inside the legionella guide's written scheme, and the loop's own design — developed length, balancing valves, index circuit — which belongs to the recirculation guide.