Hammerson-supported tech start-up brings artificial intelligence to Bullring shopping centre

1 Aug 2019

​Artificial intelligence (AI), or machine-learning, is rapidly moving from research labs to real world application. Bringing AI to something as old school as a shopping centre feels like a challenge but, it seems there is fertile ground here. At Bullring in Birmingham we have been working with Grid Edge, a tech start up spun out from Aston University, to apply machine learning (or AI) to the building’s climate control systems to optimise our energy demand on a day-ahead basis.

Grid Edge approached us in 2016 to support them in trialing their very successful R&D project in a live environment. Following an extensive period of data gathering and monitoring, this has produced a platform that allows us to actively manage the Bullring’s grid energy demand for heating and cooling on a day-ahead basis whilst allowing for changes in footfall and weather.

The Idea

The ultimate goal is to make buildings like Bullring, energy assets, enabling them to become active participants in the energy system rather than simply passive consumers. This would allow, for example, peak load management to take pressure off grid supply. This could have significant benefits in reducing reliance on fossil fuels during periods of peak energy demand. And of course it can save money.

Put simply, Grid Edge’s predictive AI technology is designed to give the Bullring Estate team foresight of their energy profile and the flexibility to shift the asset’s energy demand. We can take pressure off the grid whilst using the building’s innate ability to hold its temperature to ensure visitors experience a comfortable environment.

This sounds very much like the behaviour of a battery, not a building, but that is effectively what the Bullring looks like to the grid; a giant thermal battery that just happens to be home to shops and restaurants. Unlike a conventional battery that stores energy within its chemical cells, the Bullring stores or releases energy from the thermal inertia of the fabric of the building itself. It can then shape and shift the building’s HVAC loads to respond to the cost and carbon volatility of the grid, but crucially making sure to always stay within desired comfort (space temperature) limits.

No alt text provided for this image

The building maintains a relatively stable temperature, experiencing peaks and troughs at particular times. By using this thermal inertia to prepare the building for those peaks and troughs, we take out the need for ramping up cooling or heating, smoothing the demand across the day, even when marketing events or seasonal trading peaks put pressure on the systems.

Why is that new?

Reducing demand at peak times has of course long been possible with effective use of set points, timers and a building management system. This is important but is relatively limited in its application. Systems are either on or off, for relatively long periods of time and they are normally set seasonally so the same programme can run for some months – often with the same errors.

The value-add of the Grid Edge approach is that by combining our knowledge of up-coming events with the systems knowledge of footfall patterns, weather events and how the building performs under specific circumstances, the team can set an optimised energy strategy for the building for the day ahead that takes into account the characteristics of that particular day. This could include shutting systems down in areas of the building we know will get low footfall whilst boosting it in an area we are driving footfall towards through a marketing event.

Effectively Grid Edge’s artificial intelligence technology provides the team with a data-driven ‘crystal ball’ giving zone-by-zone predictions that enable them to create a day-ahead dynamic energy profile.

This foresight- and flexibility-based approach to energy management is important because, as we know, the cost and carbon intensity of the energy we consume is not fixed – in fact, it changes and fluctuates all the time depending upon where we are sourcing our power supplies from and the overall demand placed upon the network. At the same time, the Bullring’s energy profile is also subject to similar intra-day fluctuations, as dynamic patterns of footfall, the ever-changing British weather and the general ‘ebbs and flows’ of a busy shopping hub all intertwine to shape and influence the energy demands placed upon building. In this volatile environment, smart energy management is no longer just a question of trying to manage how much energy you use, but rather it is increasingly a question of trying to manage when you use it.

No alt text provided for this image

The Bullring is a really interesting site for the Grid-Edge AI approach as it is a data-rich environment that incurs a time-based footfall and utility pattern (busy and quiet times). This lends itself to applications for machine-learning based prediction and optimisation. Moreover, it is also a landmark building that sits at the heart of the city centre in Birmingham, playing a significant role as a major actor in the local energy system – a role that can be leveraged to deliver greater environmental and economic benefit through optimisation of its energy profile.

So what did Grid Edge actually do?

Gride Edge integrated their AI system, Edge2X, with the Bullring’s existing BMS and controls system. Starting the trial required the building of a predictive data model, so logging equipment was installed over 3G for BMS points, and regular footfall data uploaded from a third party (ShopperTrak) to become inputs to the model. This allowed Grid Edge to build a predictive building performance model of Bullring based on 2 years of data.

With the data logging in place, three core predictive models were developed that underpin the strategy and optimisation features Grid Edge have now released (and are continuing to release) to the site: predictive future occupancy, predictive future internal space temperature, and predictive future HVAC load. Building on top of these core predictions, Grid Edge provides a growing suite of analytics and optimisation tools to shape and inform a pro-active approach to energy management.

The system provides the ability to be sufficiently flexible to unlock demand management on our terms, which can be monetised. This was really critical – Grid Edge’s approach has always been that the building operator should set the terms by which their energy assets are operated, so that flexibility services – where assets are dialed up or down to address imbalance in the power grid - are transparently and holistically viewed as part of the wider site strategy. The system allows the Bullring site team to predict the potential impact on comfort, carbon and cost of dialing HVAC assets up or down to release flexibility to the grid - you can see what the impact would be in space temperature deviation of turning the air conditioning down or off and whether this is acceptable or desirable from a customer experience point of view.

The electrical flexibility that can be made available from the energy assets found in buildings like the Bullring is becoming increasingly important and valuable in the UK's energy transition in terms of keeping costs low for consumers and crucially allowing our power supplies to be decarbonised. But to make best use of the flexibility that is available in buildings, the energy industry needs to find a new and better way of engaging and rewarding empowered consumers like Hammerson, who rightly expect to have transparency, trust and confidence that their ability to provide flexibility services will not come at the expense of their own onsite comfort, carbon and cost management priorities. Grid Edge believes that its optimisation and forecast technology can play a vital part in building this trust and transparency.

The Outcome

Bullring’s HVAC assets have now been trialed and can be optimised to respond to energy price fluctuations– i.e. to the fluctuations in price of a supply tariff, and the ‘pass through’ DuOS charges. The trial showed that for the six-week period where we were actively cooling the site, the Bullring could optimise its HVAC loads to respond to these price fluctuations and create savings of approximately £23,000 over the test period. This is a really exciting development as it demonstrates the savings that can be made through really active management of energy demand – both financial savings for the asset and reductions in pressure on the grid.

No alt text provided for this image

The work with the Bullring site team is continuing and Hammerson have now agreed to support GridEdge to extend their work to Grand Central in Birmingham. Being over a major railway station this site has very different energy challenges to the Bullring but the ultimate aim is the same – for our assets to become active participants in the energy market.

Tom Anderson, CEO of Grid Edge is keen to acknowledge the important role Hammerson and the Bullring team have played in bringing this technology to this stage of its development: ”I would emphasise that the early support and faith that Hammerson placed in us was 100% critical to our company getting started. Hammerson has a genuinely ambitious vision for energy and carbon management and took a risk on us when they could have easily said “no”. That was a pivotal moment for us, and following the opportunity to work at the Bullring, Jim, Dan and I left the University to set up Grid Edge as a spin out. At this point we had also just started working with Berkeley Housing Group and Birmingham International Airport.”

From Hammerson’s point of view, the project forms part of our journey towards our Net Positive Carbon target. Having such an ambitious target has really helped focus the business on supporting projects like this which have the potential to fundamentally change our approach to energy demand. The speed and flexibility of response this gives us has to be the way forward for major energy consuming assets. We are continuing our work with Grid Edge in Birmingham and, as the results emerge, will be looking at the potential their system offers for other assets across the business.