A design heat load calculation is the first number in an entire heating project. Everything downstream inherits its error: the pump capacity, the supply temperature, underfloor pipe spacing, buffer volume, the client's heating bill. Get this number wrong and the rest of the work is faultless arithmetic on bad input.
This article covers what the calculation is, what specifically depends on it, and why a rule of thumb is not enough - whatever software you use.
What it is
The design heat load is the power a building needs during the coldest hour of the year to hold its target indoor temperature. The method is PN-EN 12831-1:2017, and it requires working room by room rather than applying a single figure to the whole envelope.
Three streams make up one room's load:
What a room's load is made of
- Transmission through building elements
- Heat escaping through walls, windows, roof and floor. U-values from PN-EN ISO 6946, ground heat transfer from PN-EN ISO 13370.
- Ventilation
- The heat needed to warm incoming air. Depends on ventilation type, building air-tightness and any heat recovery.
- Thermal bridges
- The junctions where elements meet: corners, lintels, balconies. Their share grows in a well-insulated building.
Heating-up power
An allowance for systems that run intermittently, so the building can recover its temperature after a setback.
The output is a figure in watts against each room, plus a total for the building. Those are not the same number, which matters more than it sounds.
What depends on it
This is the real answer to "why do I need this".
Heat source capacity. The obvious use, and the most expensive to get wrong. A pump sized against an inflated load short-cycles at part load, which costs both the seasonal efficiency figure you are quoting and compressor life. Sized against an understated load, it cannot keep up in a cold spell. One class of equipment either way is usually a five-figure difference in the quote.
Radiators and supply temperature. This use gets lost behind the discussion of source capacity, and it matters just as much. Knowing a single room's load and the output of the radiator installed in it, you can work out the supply temperature that radiator needs. One room with an undersized radiator raises the required temperature for the entire system - and every extra degree costs the pump's COP all winter. Without a per-room split you never find that bottleneck.
Underfloor heating. Underfloor loops have a hard ceiling: floor surface temperature. To check whether a loop can carry a given room at all, you need that room's load, not the building's.
Buffers, cylinders and hydraulics. Buffer volume and flow rates follow from the power the system has to move.
The client's bill and tariff. Annual energy use is estimated from the design load. That is what decides whether a heat pump pays for the client, and on which tariff.
Documentation and grants. Subsidy programmes and audits expect a calculation performed to the standard, with the method and assumptions stated - not a figure worked out by hand.
Why a W/m² estimate is not enough
The common shortcut is a coefficient: so many watts per square metre, depending on the building's "class". The trouble is that the spread of that coefficient is wider than the difference between a good and a bad equipment choice.
We measured this on our own module. For one specific building, where a full calculation to PN-EN 12831-1 gives 8.40 kW, estimating the same building from a coefficient produces anything from 8.02 to 23.52 kW, depending purely on which insulation class the person estimating picks. That is close to a threefold spread, and the choice of class is subjective.
One building, two methods
- Coefficient estimate
- 8,02 - 23,52 kW
- Calculated to PN-EN 12831-1
- 8,4 kW
At room level - where radiators get chosen - it looks worse. Mean absolute error per room was 17%, worst case 37%, and 4 of 14 rooms changed verdict: from "the existing radiator is adequate" to "too small", or the reverse. Four rooms where the estimate leads to a different purchase than the calculation does.
An estimate is fine for a first conversation with a client. It is not fine as the basis for ordering equipment.
Design temperature: one number, the whole chain
The calculation uses the design outdoor temperature for the building's climate zone. In Poland the range is wide:
Design outdoor temperature
- Szczecin-16 °C36 K
- Gdańsk, Wrocław, Poznań-18 °C38 K
- Warsaw, Kraków, Łódź, Katowice-20 °C40 K
At 20 °C indoors, the difference between -16 °C and -20 °C is 36 K against 40 K, or roughly 11% of the load. Swapping Szczecin for Warsaw can move the selection by a whole equipment class.
More important still, the same temperature has to hold across the whole chain. If the calculation used -16 °C and the equipment selection assumed -20 °C, two documents for one building state different assumptions, and somebody will ask about it at handover. Manufacturers publish heat pump output at specific outdoor temperatures, so the assumed temperature also changes how much the unit actually delivers.
A note on the Polish context
Poland adopted EN 12831-1:2017 by recognition, without publishing a national Annex NA. In practice the default values in the standard's informative Annex B become the de facto Polish defaults, unless the designer deliberately sets their own. It is worth remembering when comparing output from two different programs: if they assume different defaults, they produce different results, even though both "follow the standard".
What a usable calculation contains
Whatever tool produced it, a result fit to hand on includes:
- A per-room breakdown, not just a building total.
- Losses split into transmission, ventilation and thermal bridges, so you can point at what dominates.
- The design temperature used, and where it came from - climate table or a manual value.
- U-values for each element, with the layers they were derived from.
- Ventilation assumptions - type, air change rate, heat recovery.
- The method and standard, stated explicitly, so the result can be checked.
The last point is the one most often skipped, and it is the one that decides whether the document holds up in an audit.
Common mistakes
- Calculating only the building total. You lose the information needed to size radiators and underfloor loops - the usual root cause of a system that "does not heat one room".
- Taking the design temperature from the wrong place. Usually the software's default rather than the building's location.
- Ignoring thermal bridges. Their share grows in a well-insulated building, precisely because transmission through the elements shrinks.
- Confusing design heat load with annual energy use. Two different quantities; neither substitutes for the other.
- Rescaling a finished result. If the load was calculated to the standard, it should not then be "corrected" by a coefficient or by fuel consumption history.
How we do it at HeatAlgo
HeatAlgo's heat load calculator works room by room to PN-EN 12831-1:2017, with PN-EN ISO 6946 for building elements and PN-EN ISO 13370 for ground heat transfer, and shows the loss breakdown against every room.
Since July 2026 the heat pump selection module reads that calculation live rather than working from an imported copy: a change in the calculation shows up in sizing immediately, and the pump is sized at the same design temperature the building was calculated at. The underfloor heating module uses the results the same way. We wrote up that change in the release note.
If you would rather commission the calculation than run it yourself, we also offer individual calculations to order. More guides live in the help center.
Try free for 14 days