I work at Energy Sparks where we are analysing the energy data for over a thousand schools to provide them with individual insights to help them reduce carbon emissions, save money on their energy bills and teach children about tackling climate change.
We also use this data to extract insights for other schools not using Energy Sparks. With a bit of insight into the data, it’s easy to identify some quick wins. For example how reducing baseload during holidays can have a big impact on costs.
Behind the scenes we’re juggling a fair amount of data. And part of my learning over the past few years has been in understanding how it all fits together.
Schools are quite complex.
They are often older buildings, but may have a mixture of more modern temporary and permanent buildings.
They have multiple buildings on a site, and may be split across sites. This means they may have multiple central heating and hot water systems. And these may be very different systems.
They have playing fields and some may have flood lights. Some schools even have swimming pools. Many schools offer use of school buildings for community use or commercial lettings in the evenings, weekends and during holidays.
All of this has impact on their energy consumption.
One common characteristic is that schools usually have more than one meter monitoring their energy consumption. Setting aside solar metering, you might expect there to be a single gas and electricity by some schools have many different meters.
Here’s a visualisation showing number of meters by floor area using some data from earlier this year:
As you can see some schools have a lot of meters. Sub-metering provides more fine-grained information about energy consumption for specific buildings or areas of the school.
This creates a challenge when attempting to synthesise a consistent time series across the data, because:
- The data for individual meters may not be available at the same time, which creates challenges in providing an up to date view of consumption
- Some meters may be faulty or may be unable to send data reliably, so data might be patchy for other reasons
- Meters might be of different types, e.g. AMR vs SMETS meters. Or half-hourly vs non-half-hourly. This may have impacts on how quickly consumption data can be accessed and how it is interpreted.
- Some meters may only be metering small parts of the school (e.g. a single out-building) or just specific infrastructure (gas taps in laboratories, flood lights on a pitch) so the expected patterns of usage might be very different
- New meters might be installed, e.g as part of upgrades or changes to the building fabric, so they won’t all have a consistent archive of data
- Meters might be installed as replacements for another existing meter, so the time series for one meter needs to be stitched together with another as they are monitoring the same area of the school
- The infrastructure being monitored by a meter might change so the consumption and patterns of consumption might vary over time
In Energy Sparks we have evolved a set of rules for how we aggregate data from individual meters, and the ability to add configuration that will help inform how the data is interpreted. This helps to create a more consistent summary of a school’s overall energy use, whilst still allowing some drill-down into individual time series.
The same will be true for any larger non-domestic energy user. The data will be more complicated than in a domestic setting.
Lost Boy 