In order to create a building level model for estimating non-domestic operational energy consumption, we referred to DECs which are the only source of publicly available meter data we have for non-domestic buildings. For non-domestic buildings, we are interested particularly in their energy use intensity (EUI) in kWh/m2. Intuitively this makes sense as a storage facility with a large floor area will have very different heating demands to an office of a similar size. However, we very quickly realised several limitations in the dataset.
Within each Display Energy Certificate, there are very few building attributes that could be used to correlate with energy use intensities. These include:
- Building category
- Floor area
- Heating fuel type
- Ventilation mode (e.g. heating with natural ventilation)
- Whether the building has air conditioning and its estimated/surveyed rating
The information available was mostly ‘categorical’ and limited in variations within each sub-category. For example, the use of natural gas, electricity, district heating, oil etc. in ‘fuel type’ does not provide any practical ordering in terms of consumption. Potentially we can consider the efficiency of each fuel type, but that is highly dependent on the heating systems. The dataset also showed that the majority of buildings are heated by gas, so the sample sizes in other fuel types are very small, and unlikely to cover the range of combinations in building types.
As a simplified* approach to obtaining overall estimates of non-domestic gas and electricity consumptions in London, we extracted the median energy use intensity for each non-domestic building usage from DECs and applied it to the total corresponding floor area in London. This uses a similar approach to the recently released CIBSE benchmarking tool (beta version).
*We also carried out some other exploratory analysis such as matching EPCs with DECs and using borough level consumption statistics. Please get in touch if you are interested in finding out more!
We found the final consumption figures we estimated using the median energy use intensity for each non-domestic building were significantly underestimated for both gas and electricity consumption when compared to the BEIS sub-national statistics. We believe that this is likely due to two issues:
- Errors in our analysis:
- The total floor area has been underestimated.
- Inaccuracies in the median benchmarks extracted from DECs
- BEIS non-domestic consumption statistics consists of energy use from both buildings operations and other industrial processes.
We are fairly confident of our floor area estimates. We collated data from 3 different government sources including data from the VOA, DECs and EPCs. Our final estimate of 86 million m2 of non-domestic floor space in London is close to an estimate we found from an independent separate analysis which concluded the total floor space to be 84 million m2(Association for the Conversation of Energy 2016).
The median energy use intensity extracted from DECs are not representative of every building type. We know from our floor area analysis [link] that DECs provide very little or no coverage for office, retail and industrial floor spaces compared the whole non-domestic stock in London.
The median energy intensity benchmarks extracted from DEC are higher than values quoted in BEES study (BEIS). Buildings that have DECs are perhaps more incentivised to be energy efficient, which would suggest why the figures we estimated are lower than the BEIS reported consumption. However, when we compared the figures to other benchmarks such as from the Building Energy Efficiency Survey (BEES). We found that the estimated energy intensity values from DECs were in fact higher than other benchmarks. For example, DEC data suggests a median thermal energy use of 105 kWh/m2 for Offices, whereas the BEES benchmark for offices was 89 kWh/m2.
So finally, we started to ask how much of the non-domestic meter consumption reported by BEIS was actually attributed to building energy use.
BEIS uses a fairly crude assessment to differentiate between energy consumption from domestic and non-domestic meters. If a meter reading shows an annual gas consumption of 73,200kWh or higher, BEIS classifies this meter a non-domestic user.
Without better labelling of meters, misclassifying small non-domestic buildings as domestic, and large domestic consumers (e.g. communal heating) as non-domestic is almost inevitable. To put this in perspective, the Xoserve estimates about half a million small non-domestic users have been misclassified as domestic consumers for gas meters.
The story is different with regards to electricity meters, where there are Profile Classes that enable the BEIS team to more accurately differentiate between domestic and non-domestic consumers.
The magnitude in discrepancy between our estimate and the BEIS statistics is simply huge. To put this into prospective, our calculations have not explicitly taken into account energy consumed by TFL, who can be considered as one of the largest consumers in London. According to figures in their energy purchasing paper in 2017, TFL consumers about 1.65 TWh of electricity, and 0.8 TWh of gas. This leaves us with at least 10 TWh of discrepancy between the reported statistics and our estimated consumption from buildings.
This would seem like a huge figure to reconcile from accumulations of errors, so the only logical conclusion would be that the discrepancies are contributions from industrial processes, waste plants, data centres etc.
One thing is for certain, non-domestic energy use needs to be much lower to meet climate targets. The city of Toronto’s “Zero Emissions Buildings Framework” have set energy performance targets for a range of building archetypes (e.g. office, retail, mixed use, high rise etc.) in their efforts to reduce buildings emissions. When we compare the median energy uses extracted from DECs to Toronto’s phased targets, it highlights the massive amounts of work we need to do in improving energy performance in our non-domestic buildings.
The more we found out, the less we knew.
The basis of our analysis was focused around consolidating the BEIS non-domestic consumption statistics with the EUIs we have derived estimates from through our analysis of VOA statistics, DECs and EPCs.
At the start of the journey trying to estimate how much energy non-domestic buildings used in London, we thought we could take the energy consumption of different building types, run analysis on the energy consumption for these building types, and derive an Energy Intensity Use (kWh/m2) for each building type. The idea being that we could then marry the total energy consumption of non-domestic buildings, with the area of different building types to paint a picture of energy use in London. The reality is that we are nowhere near this yet, and the level of granular detail required to draw clear conclusions is not publicly available.
We have been referring to the BEIS non-domestic consumption statistics as our gold standard reference, and so have a lot of building professionals when doing strategic planning work to help local boroughs develop zero carbon pathways and the like. Despite the limitations of the BEIS statistics and without better data, this is clearly the best set of data we have to understand how much energy non-domestic building use in London.
Our analysis presents us with a very big problem insomuch that there is a huge unknown as to what proportion of the non-domestic consumption are from operational energy in buildings, and which is from other processes considered as building energy use. Our conclusion has perhaps left us with more questions than answers, how much energy are non-domestic buildings really consuming?