Heat loss calculations are a bit of a faff. They involve lots of measuring and lots of assumptions. The most notable of which are the U-values of walls and windows. This is because not all walls are created equal and there is significant variance in the fill, density, and construction of walls in some older properties.

There are many people and plumbers that have noted that U-value tables are inaccurate and can lead to over-sizing heat-pumps. Alternatively, a modern building may not be as insulated as expected and therefore lead to under-sizing of heat-pumps.

Heat-loss calculations are, therefore, somewhat unscientific and a lot of thought goes into trying to make them more precise without thinking of the obvious “why not just take a measurement?”

The Electric Radiator Method (Idea 1 – now superseded)

1. Keep the normal heating on for a few days constantly – to warm the structure of the building.
2. Monitor external temperature of the building.
3. Put an electric radiator with monitoring equipment in every room of the house. These should measure room temperature and the power usage at regular intervals.
4. Turn off normal heating and turn on electric radiators.
5. Set the radiators to maintain the room temperature.
6. Perform a time-series analysis of the heat-load of the property.

This method would provide the exact heat loss of a property at a given temperature, as electric radiators are 100% efficient since all they do is produce heat.

However, after some thinking I thought this was a bit faffy and assuming a house has similar walls and windows throughout, all you need to do is measure the output of a single radiator in one room.

Simplified Single-Room Measurement Approach

This approach can be further refined into a practical DIY method:

1. Choose a representative room – ideally one with external walls and average window area.
2. Set up an electric radiator – with energy monitoring and set to the correct temperature.
3. Install temperature sensors inside and outside the room and externally.
4. Seal the room as much as possible to minimise air exchange with other spaces.
5. Turn off any other radiators in the room.
6. Run a test with the radiator maintaining a constant temperature (e.g., 21°C).
7. Record data at regular intervals:
   • Indoor temperature in the room
   • Outdoor temperature
   • Power consumption

It is essential to maintain the temperature of the rooms around the room to ensure minimal heat transfer to the test room. I didn’t do that for this test, but do as I say not as I do.

Now that you have the data of the energy used during the time of operation, is it possible to estimate the room’s external area U-value. You can check this against data-sheet U-values to see whether one is much higher or lower.

You can then do a full heat-loss calculation and proof the theoretical calculations against the tested values.

Proof of Concept – My test

Measured Heat Loss vs External Temperature

Based on the single-room measurement approach, I collected data to model the relationship between external temperature and heat loss. My room required 0.6 kWh of heat energy per hour to maintain an internal temperature of 18.5°C when the external temperature was 7°C.

Heat Loss Coefficient Results

From this baseline measurement, I calculated a heat loss coefficient of 0.0522 kWh/h/°C for the room. This coefficient enabled me to estimate heat requirements across various external temperatures. While maintaining a room temperature of 18.5 degrees, in reality I’m just modelling the delta so think of the temperature difference rather than focusing on the External Temps.

Extrapolated Heat Loss at Different External Temperatures

Limitations of This Experiment

This is a rough experiment, rather than a conclusive result. As I wasn’t able to heat the rest of the house, and the walls weren’t up to temperature as the room had been 10 degrees C when I entered it. One of the walls was a thick chimney breast which was probably absorbing a lot of heat. Additionally, I only have a single data point which isn’t really an experiment, but it is interesting none the less – for me anyway.

Despite these limitations, I think this approach might offer people who are unsure about their properties heat-loss a foundation for heating system design that extends beyond purely theoretical calculations, while requiring minimal equipment and setup time compared to whole-house testing methods. Further refinement of the measurement approach could yield even more accurate results.

I should have compared my measurements to predicted values from doing a heat-loss survey, but I had other tasks to get on with so didn’t have time to perform measurements. I will go back and measure in future though!