Jordan O’Brien, contributing editor for Electrical Review, discovers how AI and machine learning is creating a new breed of smart buildings.
The world is going nuts over self-driving cars and their ability to make your life easier, but the reality is, we’re still a long way off from buying cars that do away with human drivers altogether. That’s why we should be concentrating on self-driving buildings instead, a technology that is very much already with us.
Sure, it sounds nuts when I call it a ‘self-driving building’, because the building is of course stationary, and not ‘driving’ anywhere. However, what I mean by that phrase is a building that uses AI and automation to adapt on the fly without needing any user input. It’s building automation of steroids.
Companies such as Schneider Electric, silicon valley start-up, Verdigris Technologies, and German automation firm, Dabbel, are already rolling out solutions that leverage AI, and are using the Internet of Things to better inform a central building automation system and to teach it when to act without any user input.
Electrical Review spoke to Emilie Hung, solutions architect at Verdigris, to get a better understanding on what a ‘self-driving building’ could be.
What are the advantages of adaptive automation over traditional building automation?
“Traditional building automation systems are not intelligent. They are based on simple feedback loops and can be binary. The controls are programmed based on a few static inputs, design conditions, and often are not re-programmed after a building is commissioned. In this scenario, building performance can drift undetected. Moreover, if control actions are associated to a single variable then some controls may work against each other (i.e. heating and cooling at the same time). It also requires knowledgeable personnel to re-calculate the control parameters and re-program a system.
“Adaptive automation is intelligent and predicated on periodic analysis and optimisation. Algorithms learn the behavior of the building, accommodating multiple inputs such as historical building consumption, weather, occupancy, predicted forecasts. In this scenario, the models can identify which parameters are strongly and loosely coupled, combining more inputs than possible with human computation, and optimizes building electricity systems without human intervention.”
What are the cost differences between adaptive and traditional automation, including any cost savings?
“Currently adaptive automation integrates into existing building control systems and is an additive feature. Setup requires the ability to connect and digitally send signals to the building and equipment controls.
“Savings from adaptive automation can be though demand management or energy efficiency. We estimate annual savings from adaptive automation to range between $10,000 – $25,000/yr (USD) for a 100,000 sq ft building, that translates to approximately $0.11/sq ft – $0.25/sq ft (USD) in savings.”
How is adaptive automation easier for a facilities manager to use vs standard building automation?
“Building automation as a whole is under-the-hood technology. Standard building automation is a static model with preset control sequences whereas adaptive automation uses artificial intelligence, is dynamic and as the name suggests, adaptive to changing conditions. The difference is in the experience of comfort and cost and energy savings from building operations.”
The IoT is crucial in creating self-driving buildings, as the central automation system makes decisions based on the sensor data that is being fed into it. That means if the temperature sensor detects that the room has dropped below an optimal level, the system will then consult other factors, such as an occupancy sensor, to decide whether to turn the heating on. If it’s coming up to 5pm and the last person is about to leave, then it may decide to leave the heating off.
To reach our goal of carbon reduction then using modern building automation systems using AI and machine learning is an absolute must. While it’s important to reduce our carbon footprint, these systems also come with the upside of potentially saving a building owner a significant amount of money.
One example of how the system can help building owners can be found at the Grand Hyatt San Francisco. Since the hotel switched to a smarter building automation system, one that uses AI and machine learning, it has found its energy use to be greatly reduced. In fact, the new system has consistently saved the hotel over 20% of the costs of the controlled load, resulting in an average monthly ROI of 41% and earnback of the initial investment in less than six months.