Building Energy Modelling, Explained

Building energy modelling is a calculation, not a measurement. It takes a description of a home, its walls, roof, windows, heating system and controls, and estimates how much energy that home would need under a set of standard conditions. The standard conditions matter as much as the building description, because they replace real occupants with a typical pattern of heating, hot water use and room temperatures.

That is why two homes that look alike on paper can behave differently in real life, and why a model result should be read as a comparison tool rather than a bill forecast. A model is most useful when it lets you compare options on the same basis, for example whether wall insulation or a heating upgrade changes the calculated result more, and least useful when it is quoted as a guaranteed saving for one household.

In practice a modeller builds a simplified version of the home. Rooms are grouped into zones, heat loss through each surface is estimated from its construction and area, and gains from sun, people and appliances are added in a standardised way. Heating and hot water systems are then represented by their type and efficiency, and controls are represented by how they are assumed to operate. Every simplification is a judgement, and different modellers can reasonably make different ones.

The output is usually an energy demand figure, a rating or a band, plus a breakdown that shows where calculated heat goes. Reading the breakdown is more informative than staring at the headline number. If most calculated loss is through the roof in one model and through ventilation in another, the two models disagree about the home, and that disagreement is worth resolving before money is spent.

Before you act on a model, ask for its inputs in writing: the construction assumed for each wall and floor, the window types and areas, the heating system and controls, and the occupancy and temperature assumptions used. Check those inputs against what you can see in the home, such as the actual boiler type, the presence of a room thermostat, and whether loft insulation is visible. A single wrong input, like a solid wall recorded as a filled cavity wall, can move the result more than a modest retrofit measure would.

Also ask which method was used and why it fits the question. A compliance calculation, a retrofit options appraisal and a detailed design simulation are different tools with different levels of detail, and a result from one should not be presented as if it came from another. If the modeller cannot explain the method in plain language, treat that as a reason to pause rather than a reason to trust the headline figure.

Modelling rules are not the same everywhere, and this site keeps the EU and UK systems separate for that reason. In the UK, the Standard Assessment Procedure, known as SAP, underpins EPC ratings for homes, with its own conventions for standard occupancy and heating patterns. In EU member states, national calculation methods implement the Energy Performance of Buildings Directive, so the same physical home could receive different ratings under different national methods.

The practical consequence is that ratings and model outputs should not be compared across borders as if they were the same measurement. Within one country, a model is a reasonable way to compare options or track a change, provided the method stays constant. Across countries, only the underlying physical description, wall areas, insulation levels, system types, travels; the rating does not.

A common mistake is to treat a modelled saving as cash in hand. A model that says a measure cuts calculated demand by a fifth does not promise a fifth off a real bill, because real bills reflect real weather, real occupancy, real thermostat settings and real system maintenance. Households that heat more rooms, or keep a home warmer after insulation, will see smaller bill changes, and that is normal behaviour rather than a failure of the measure.

Another mistake is to commission a detailed model before the basic facts are known. If wall construction is uncertain, if the heating system age is unknown, or if draughts and damp are unexamined, a sophisticated simulation will simply carry those unknowns forward with more decimal places. Fix the evidence first, then choose the level of modelling the decision actually needs.

Keep the model report, the input schedule and any certificate that came from it together, and note the date and the method version. When a later assessor or installer proposes work, ask them to show how their proposal changes the same inputs, rather than presenting a new headline number with no trail back to the old one. That comparison is the real value of a model.

Questions worth asking include: which inputs drive this result most, what happens if one key input is wrong, and what would have to change for a different measure to look better. A good modeller welcomes those questions, because they show where the model is robust and where it is fragile.