Three ways to size a central heating system

Space heating is one of the largest contributors to personal carbon dioxide emissions. Pretty much all this carbon dioxide comes from burning natural gas in domestic boilers. If we are to reduce emissions to zero, the only way that this can be achieved is to replace the burning of gas with electricity-based heating which can ultimately be sourced from zero-carbon generators like wind and nuclear.

Realistically, this means the installation of heat pumps. But to begin to design a heat pump installation, you need to know how big it should be and do know that you need to know how much heat is needed to keep a house warm.

This process is called sizing a central heating system.

There's many ways that this central heating sizing can be done, ranging from rules of thumb to calculation from first principles.

I'll delve into calculations from first principles in future. For this article, I'm going to explore and compare three different ways to size a central heating system:

  • Using historical usage
  • Using measured heat loss and degree days
  • Using Energy Performance Certificate (EPC) estimates

Let's look at each in turn.

Using historical usage

Although just using historical gas usage is easiest, it's also one of the more error-prone methods.

It doesn't take into account the different temperature outside the house each year. It also changes over time as I make improvements in energy efficiency to the house.

That being said, I can see what it looks like very easily. I take the data I have on gas consumption for the full years spanning 2015 to 2022 in this house and I subtract from that my estimate for gas used for heating water and for cooking.

This leaves a graph that looks like this:

Although usage is decreasing over time, the average for 2021 and 2022 is 9,320 kWh/year.

Heating estimate method 1: Historical usage

Heating requirement as estimated using historical readings is about 9,320 kWh/year.

Using measured heat loss and degree days

A degree day is a measure of heating or cooling. Properly calibrated for the building, it should be possible to use degree days to work out exact heating or cooling requirements for a building based upon outside air temperatures at the building's location.

Degree days are calculated with reference to a base temperature. This base temperature is chosen to represent the outside air temperature at which the house remains at a comfortable temperature indoors. Then, the outside temperature is compared to the base temperature over many days and the difference between the base temperature and the outside temperature is multiplied by the number of days that difference was maintained.

For example, in the UK the base temperature is conventionally 15.5°C. This is lower than normal room temperature to account for activities in the house that warm it up without turning the heating on (e.g. electricity use, people moving about etc.).

If the outside air temperature was 14.5°C for 1 day, that represents 1 degree day of heating required.

If the outside air temperature was 0°C for 2 days, that represents 31 degree days of heating required (15.5°C multiplied by 2 days).

It's perfectly possible to more thoroughly estimate the base temperature that matches the thermal properties of a building 1 but this is beyond the complexity I need for a quick analysis.

For me to be able to use degree days as a way of sizing my heating system I need to calibrate degree days for this house. Given that degree days represent a difference in temperature maintained over a time period, this calibration will end up looking like an amount of energy needed to heat the house by one degree day. Effectively this is kWh/degree day.

Calibrating degree days for this house

Calibrating this house requires a period with accurate meter readings, a consistent internal house temperature, and a precise records of degree days.

I'm going to do this calculation for the period 1st January to 30th March 2023 - a total of 89 days.

Over this time, I burned 5,595 kWh of gas, of which around 88% is for heating, which amounts to about 4,835 kWh for heating.

To get degree days for this period I used the excellent calculator at degreedays.net2 which uses my local weather station as well as the time period to tell me that over this 89 day period, there were 864.5 heating degree days.

This means my house loses heat at a rate of 5.6 kWh/degree day i.e. a heat loss coefficient of 233 W/K3 (5.6 kWh / 24 h).

I can combine this calibrated value with historical degree days readings for my area to predict what my house usage "should" have been:

YearDegree days recordedTheoretical heating for the year (kWh)Actual heating for the year (kWh)Ratio

This means that my degree day estimates are actually slightly higher than what I recorded in 2021 and 2022. However, the similarity in the two ratios is a positive sign, suggesting that may be off by a consistent factor.

In fact, this difference probably reflects the fact that I was heating the house more or less continuously during the 89-day test period, where as I would normally switch the heating off overnight.

Heating estimate method 2: Degree days

Heating requirement as estimated using degree days is around 11,300 kWh/year.

Using Energy Performance Certificate (EPC) estimates

The easiest way of all may be to simply look up the EPC certificate for your house on the government database4. EPCs were introduced in 2007 so if you've purchased your house since then it's likely there's a certificate available. Certificates are also required to let properties and for certain energy grants so it's worth looking in almost all cases.


My certificate has the energy rating, as shown above. It also provides an estimate of how much energy is needed to heat my house, which is included toward the end. In my case, this estimate is 10,592 kWh/year for heating.

The EPC also includes an estimate of 2,833 kWh/year for hot water which is quite a long way from the true value I've calculated of 1,050 kWh/year. I believe that this reflects relatively low water usage in our house though, rather than being a poor estimate of how much energy is needed to heat water per se. We typically use about 120 L/day of water in our 2-person household. Our local water supplier, Thames Water, reports that the average usage is 145 L/person/day5 which means we're only using water at 41% of average rates. If I scale up my hot water energy usage from 41% to 100%, it would be 2,538 kWh/year – far closer to the EPC estimate of 2,833 kWh/year.

Heating estimate method 3: Energy Performance Certificate

Heating requirement as estimated by the Energy Performance Certificate (EPC) is around 10,600 kWh/year.

Summary and next steps

The three different methods I've used to estimate heating have produced slightly varying results. Here's the summary:

MethodGas usage for heating (kWh/year)Gas usage for heating (kWh/m²/year)
Historical usage9,32085
Degree days11,300100
Energy performance certificate10,60096

The final column gives values of kWh/m²/year which allows me to correct for the size of my house (a larger house will always need more heating than a smaller one). These normalized values can be directly compared to those of other buildings and to national and international documents, like building regulations.

Going forward, I'll be comparing these values to best practice for new buildings and for retrofitting existing buildings. I'll also be comparing these to values that I prepare using a "first-principles"-type calculation.

Posts in this Series

comments powered by Disqus