London’s Biggest Office Owner Uses Its HQ As An AI Guinea Pig
The world of science usually frowns on doctors experimenting on themselves. But when it comes to real estate’s battle to cut its carbon emissions, one big REIT is using itself as a test bed to see what artificial intelligence can do to beat the climate crisis.
Landsec, London’s biggest office owner and the UK’s second-largest REIT, is putting the use of AI under the microscope to see if it can be used to help the building use less energy while saving carbon and money.
If the experiment at its 50K SF headquarters in Victoria Street, London, proves a success, Landsec is considering rolling it out across its portfolio, which totals £10.2B, just over £6B of which is in central London.
“The beauty of AI is that everything is automated [and] based on internal and external factors, and that means we can take decisions and undertake actions far more efficiently,” Landsec Senior Energy Manager Andy Mazzucchelli told Bisnow.
“A person would have to analyse the data, spot an opportunity, discuss that internally, and then someone has to actually go and implement that decision. With AI, that is all autonomous.”
Landsec has committed to spending £135M over the next seven years to reduce the operational carbon emissions of its portfolio — emissions that are generated by the day-to-day operations of its buildings. It will also fund the move to clean energy sources.
The investment is being made partly because Landsec has signed up to the Science-Based Target Initiative to keep its carbon emissions in line with limiting global heating to 1.5 degrees Celsius. It was also influenced by mounting data showing that greener office buildings command higher rents and sell for higher capital values as tenants and investors look to comply with increasingly stringent global sustainability regulations.
Part of Landsec's £135M was spent on replacing hardware that uses a lot of hydrocarbons, including gas boilers with air-source heat pumps.
But AI deployed alongside sensors, as well as greener building management and HVAC systems, has the potential to reduce energy consumption and carbon emissions by significantly more than when the hardware is deployed on its own, Mazzucchelli said.
The company started working with Canadian company Brainbox AI on how AI could be used in its portfolio about a year ago. The trial of the AI system at its 100 Victoria St. HQ began just over three months ago.
A device attached to the BMS of a building collects data on the building’s energy use from the various systems in operation, employing sensors to collect data on factors like room temperature and occupation.
Brainbox AI then uses a cloud-based algorithm to analyse both this data and external factors such as the weather.
The AI has a learning phase of two to three months. During that time, it figures out when people begin using the building, when they leave, which areas are occupied at what times of day, and how all this affects things like temperature.
After this learning phase, it can work out the optimal time for heating and cooling and air flow systems to be turned on, ensuring different areas of a building are at a comfortable temperature, but systems are not in use when the building is not being occupied.
It is still early days for the trial, which could run for up to a year, but Mazzucchelli said the initial findings had been positive.
“We need to take more time to understand the impact, but we’ve seen some promising savings,” he said.
A key signal the system is doing its job was that there were fewer points in the day when the temperature of an area was changing significantly compared to a preset baseline, meaning the system had set the right temperature in the first place, Mazzucchelli said.
As with many technology providers, Brainbox AI provides its service on a subscription basis, with the price based on the size of the building. That may not sound important, but such subscription-based models could be vital in ensuring the real estate sector decarbonises as a whole rather than just larger players.
While big firms like Landsec have the capital to invest in the systems and people needed to cut emissions, smaller companies that might own one or two smaller buildings do not. The price of technology needs to be low enough or tailored to the size of the user to be deployed widely. Renting rather than having to pay upfront makes it easier.
Landsec experimenting on itself could lead to wider use of AI across its portfolio at a moment when the potential of such machine learning systems is being widely explored in myriad forms.
And the company is also thinking about how to help tenants across its portfolio understand their carbon emissions. That is crucial for real estate decarbonisation given tenants use about 50% of a building's energy, per an Urban Land Institute report.
That includes producing energy audits on behalf of tenants and providing advice on how to reduce energy usage, Mazzucchelli said. Landsec is also thinking about introducing league tables of tenant emissions, to gamify carbon emissions and incentivise them to use less energy.
“It’s important this isn’t a one-off, that you get back in touch to see how tenants are progressing, to find out the challenges and barriers to implementing plans, and to show the impact that having energy data can have.”