22 October 2021

Energy Managers turn to machine learning to optimise energy consumption

Six in 10 managers responsible for energy performance in buildings believe that artificial intelligence (AI) could transform the way they interpret data to improve efficiency.

A snapshot poll of participants attending an Energy Assets’ webinar also found that less than one in 10 currently have capacity to review consumption data more than once a week.

The poll reflected the challenge for energy managers in analysing the huge volume of data now available through half-hourly automated meter reads. This underlines that manual interpretation of big data simply cannot identify the fingerprints of energy waste hiding in plain sight in a timely way without an army of analysts – and that simply is not practical.

One of the most powerful tools emerging in the armoury of energy managers in industrial and commercial (I&C) sectors  is the application of machine learning, informed by AI, to enhance energy performance based on half-hourly data from automated meter reading (AMR) systems.

With machine learning, it is possible to interrogate years’ worth of historic half-hourly data in seconds. Taking this as an absolute reference point, the AI system can be used to spot tell-tale signs of energy waste unique to each building through pattern recognition – flagging up equipment running needlessly, heating controls incorrectly set for example. From this, a checklist of priority actions can be created to drive out waste and reduce carbon output.

This innovative approach has been adopted by The Energy Consortium (TEC), a contracting authority owned by its members which delivers a wide range of services in energy procurement, data reporting, risk management and cost reduction on a not-for-profit basis for its predominantly university sector membership. TEC, which currently risk manages 11TWh of gas and power across 10,500 meters, is partnering with Energy Assets AMR DNA energy data service, powered by kWIQly, to apply machine learning across a number of HE campuses.

Pinning down energy waste and improvement opportunities over an estate of complex, multi-faceted buildings, requires rock-solid benchmarks to compare like-with-like. It then becomes possible for the AI driven system to progressively learn what best performance for each building looks like.

Energy managers are well used to monitoring performance through multi-utility data portals, but without a data validated benchmark, managers won’t know that a building is performing poorly, even if event exception alarms are integrated. And even when an issue is flagged, finding the cause can be like finding a needle in a haystack, whereas AI can also provide a diagnosis.

With AMR DNA, analysis of consumption data linked to a set of variables, such as weather information and comparative building performance, enables the  system to spot patterns outside the expected norm. Once learned, the AI analyses half hourly data overnight and provides a daily checklist of potential problems for investigation.

For example, in the case of one school, AMR DNA flagged that the heating system was operating from 4am and that the boiler was firing at the end of the school day. Investigation revealed that opening external doors at home time was activating the thermostat – a problem that AMR DNA analysis revealed was prevalent in 30% of schools in the portfolio. An algorithm was written to address the issue.

The system also places simplicity at its core. For example, for a supermarket chain an out-of-hours turndown load report is colour coded to quickly show whether they are beating efficiency goals (blue) or failing (red). Energy managers can then investigate and either take action or identify reasons for the consumption change (a new in-store bakery for example). If it is the latter, an AI informed system will ‘learn’ this new profile for future reports.

In short, AI does the heavy lifting for energy managers when it comes to making sense of data, freeing up time to enable skilled professionals to get on with managing energy rather than looking at data, which in turn opens up more opportunities for efficiency gains and carbon reduction.

Taking TEC as an example, the application of machine learning has enabled its members to achieve significant improvements in energy efficiency.

A study of the full TEC portfolio showed that an annual saving potential of £6,000,000 could be achieved if all buildings that do not turn consumption down to 50% overnight were to do so. Obviously in the case of TEC there are a number of buildings that are not able to do this, however the software allows the addition of markers to support necessary filtering.

It’s becoming increasingly clear that energy managers in I&C settings have a critical role to play in greening Britain’s economy. The good news is that the enabling technologies that can manage energy efficiency in buildings more effectively already exist. There are signs, too, that the financing and investment tools needed to drive these changes are gaining traction at energy generation source and from network to meter.

This matters because the Government aim is to reduce carbon emissions in the UK by 2030 by at least 68% compared to 1990 levels. Indeed, the goal is to achieve a 78% reduction by 2035, but achieving this will require the widespread adoption of big data analytics and digital control systems to effect a transformation of building energy efficiency, particularly in the I&C sector.  

Want to know more?

Go to AMR DNA – Energy Assets and you can:

1. Watch our webinar,
2. Listen to our Prodcast,
3. View literature and a video guide, or
4. Book a demo to see how AMR DNA can support your business.

This article first appeared in Energy in Buildings & Industry – October 2021.