Sizing a heat pump: heat loss, design temperatures and why it matters

The single requirement that most distinguishes an MCS heat pump installation from an unregulated one is this: the heat pump must be sized from a calculation of how much heat the building loses, worked out room by room, in accordance with BS EN 12831, and not from the size of the boiler being replaced, the floor area or a rule of thumb.

The reason is that the boiler was almost certainly the wrong size too, and nobody noticed. A boiler that is twice as big as the house needs simply runs for half as long, at a modest cost in efficiency. A heat pump that is twice as big as the house needs cycles on and off all winter, wears its compressor out, runs badly, and costs more than it should. Oversizing is the most common failing in heat pump installations, and the heat loss calculation is the discipline that prevents it.

Heat leaves a house two ways

All of it comes down to two mechanisms, and a heat loss calculation is simply an accounting exercise across both.

Through the surfaces. Heat conducts through walls, windows, roofs and floors. How fast depends on three things multiplied together: the area of the surface, how readily it passes heat, and how large the temperature difference across it is.

The middle term is the U-value, measured in watts per square metre per kelvin. A single glazed sash window is around 4.8. A modern triple glazed unit is around 0.8. A solid Victorian brick wall is around 2.1; the same wall with 100 mm of internal insulation is around 0.3. These numbers are the whole story of why insulation works.

With the air. Air leaks out of a house and cold air leaks in to replace it, and that cold air has to be warmed. The calculation is the volume of the room, multiplied by how many times an hour its air is replaced, multiplied by the temperature difference, multiplied by 0.33 watt-hours per cubic metre per kelvin, which is what it takes to warm air.

In a leaky old house, ventilation can be a third of the total. In a well sealed modern one it is much less, though it never reaches zero because people need to breathe.

The temperatures the standard fixes

A temperature difference needs two temperatures, and MIS 3005-D specifies both instead of leaving them to judgement. This is what makes two calculations on the same house comparable.

Inside

Table 1 of the standard sets a minimum internal design temperature for each kind of room, taken from CIBSE Guide A:

These are minimums, not targets to negotiate downwards. A bathroom designed at 18 degrees will be a cold bathroom, and the customer will find that out in January instead of on the quotation. A householder who likes the house warmer than the table can ask for a higher figure, and should, because a system designed for 21 degrees will struggle to hold 23.

Outside

Table 2 gives the outside design temperature: not the coldest it has ever been, but a cold day the system should be able to cope with. It is given for eight locations, in two columns. Column A is the temperature exceeded for 99 per cent of the hours in a year. Column B is the colder 99.6 per cent figure.

The choice between columns is the contractor's, based on the location. If column B is used, no further uplift for intermittent heating is required, because the colder figure has already built in the margin.

Floors, which are the exception

A solid floor sitting on the ground does not lose heat to the outside air. It loses heat to the ground, and the ground under a house is slow, thermally massive, and never gets as cold as a January night. MIS 3005-D therefore requires that a solid ground floor be calculated against the local annual average air temperature instead, which in Britain runs from about 8.5 °C in north-east Scotland to 11.3 °C in the Thames valley.

The difference is not small. A living room in London is designed against 22.7 degrees of temperature difference through its walls, but only 9.7 through its floor. Calculating a solid floor against the outside design temperature more than doubles its apparent loss, and inflates the whole house with it.

A suspended timber floor over a ventilated void is different again: the void is close to outside air, so the design temperature applies.

Why room by room

A whole-house figure is enough to choose the heat pump. It is not enough to design the system, and this is the point people miss.

Each room's heat loss determines the size of the emitter in that room. A house needs its heat distributed in the proportions in which it is lost, and those proportions are not the proportions of floor area. A north-facing bedroom over an unheated garage, with two external walls and a large window, may lose twice what a room of the same size loses in the middle of the house.

Then there is the flow temperature. A house runs at whatever flow temperature its worst-served room demands. One room with a radiator too small for its heat loss forces the entire system to run hotter, all winter, for every room. That single radiator can cost several hundred pounds a year, and only a room-by-room calculation will find it.

Choosing the machine

MIS 3005-D requires the heat pump to provide at least 100 per cent of the calculated heat load at the design flow temperature, at the outside design temperature, without help from any immersion heater built into it. That last clause matters: a machine whose published output includes a 3 kW electric element is not a machine of that size.

The output figure has to be read at the right conditions. A heat pump advertised as 8 kW is generally 8 kW at 7 °C outside and 35 °C flow. The same unit at −3 °C outside and 50 °C flow might deliver 5.5. The MCS product database publishes the output at a range of conditions for exactly this reason, and it is the figure at your design conditions that has to cover your heat loss.

The standard also asks that an air source system should be able to hold the internal design temperatures across repeated defrost cycles, and that a supplementary electric heater, where one is fitted, must be designed not to operate above the design temperature. It is a backstop for weather beyond the design case, not a part of the heating system.

The 55 degree rule

Two clauses in the standard are worth knowing about as a customer.

High temperature heat pumps should be avoided unless the application genuinely needs a flow temperature above 55 °C. They exist, they work, and they let a house keep its radiators, but they run at a lower efficiency and cost more to run.

And where a design proposes a flow temperature above 55 °C, the contractor must also produce an alternative design at 55 °C or below, and explain the difference in efficiency and energy consumption, so the customer can choose. If a quotation proposes a hot system without offering the cooler alternative alongside it, the standard has not been followed.

Reading a calculation you have been given

A few checks catch most of the errors.

• Is there a row for every room, or just a total? A total on its own does not show that the method was followed.
• Are the internal temperatures at least the tabulated ones, room by room?
• Is the outside design temperature roughly right for the location, and stated?
• Are solid ground floors calculated against the annual mean, not the design temperature?
• Do the U-values describe the house that exists? An assumed cavity fill that was never installed is a common source of a heat loss that is far too low.
• Are window areas taken out of the wall areas, or counted twice?
• Does the air change rate reflect the building? A 2015 house at 1.5 air changes an hour is being flattered; a 1900 one at 0.5 is being fantasised about.
• Does the total, divided by floor area, come out somewhere sensible? A well insulated house lands around 40 watts a square metre; a poor one comfortably over 100. A figure of 15 or 200 needs explaining.