Many new buildings are constructed to maximize the efficiency of HVAC systems and to minimize their environmental impact. But what about older buildings? How can they work to achieve efficiency and environmental goals? One answer involves the use of artificial intelligence (AI).
“There are a lot of older buildings, and existing buildings are rarely demolished,” says Jean-Simon Venne, Co-Founder and Chief Technology Officer of Brainbox AI. “There are so many of them and they consume a lot of energy and have a big impact on the number of emissions.”
Older buildings are reactive rather than proactive
HVAC consumes around 50 to 60 percent of the energy used in a building, second only to lighting, whose impact is diminishing with the transition to lower-energy LED lighting. Although older buildings have control systems, they are reactive rather than proactive.
Temperature schedules are set according to times and a set of fixed rules. Any changes to the set schedule require a technician who knows the control type and the programming language.
Using AI to predict analytics of buildings
Brainbox AI provides self-adapting AI to proactively optimize the energy consumption and GHG emissions Brainbox AI provides self-adapting artificial intelligence technology to proactively optimize the energy consumption and greenhouse gas emissions of buildings. The technology connects with existing control systems and extracts information that artificial intelligence (AI) then uses to “learn” how a building reacts to temperature fluctuations.
Extrapolating that information, the system can provide predictive analytics; in effect, a glimpse into the future. It’s an approach that quickly provides savings of up to 25% in energy usage and 40% in emissions, even in existing older buildings.
Zone-based energy savings
An example of how Brainbox AI operates in older buildings can be seen in the Holiday Inn Longueuil in Quebec, Canada, which has over 10,000 square feet of meeting space and a six-story atrium. They sought to reduce their energy consumption and optimize the HVAC systems in the common areas.
BrainBox AI calculated energy savings by following the demand of each zone based on the weather and then normalizing the energy consumption based on heating degree days (HDD) and cooling degree days (CDD). Over five months, BrainBox AI was able to achieve an average savings of 34%.
The Brainbox system uses machine learning and “deep learning,” which employs neural networks to mimic the activity of the human brain. Brainbox exposes the networks to data so that they can “learn.” This differentiates Brainbox AI from other companies that use technology dashboards to provide insight into a building’s operation but do not provide autonomous operation.
Cost-savings while improving environmental concerns
In AI and HVAC, spending a fraction of energy on smaller corrections avoids having to spend more energy
Venne compares the approach to guiding a space shuttle. If it begins to go off course, a series of small corrections (such as booster rockets), can guide it back to the target path. In the case of AI and HVAC, spending a fraction of energy on smaller corrections avoids having to spend more energy on larger corrections. A computer analyses all the possible moves, their expected impact and chooses the best path among the possible outcomes.
Both energy use and environmental concerns are driving building owners to the new approach. They may be willing to pay a cost to lower emissions, but in this case, using less energy provides cost savings as well as improving environmental impacts.
“Now we can both save money and save the planet,” says Venne. “In the last six years, we have become able to do a lot of things at a low cost [using AI], things we were only dreaming of a few years ago. Now we are only limited by our imagination.”
Greener energy
Another element to minimize environmental impact is to use more energy that has been “responsibly” sourced (such as solar or nuclear power plants) and less energy that comes from high-polluting coal plants, for example. At some times of day, the available kilowatt-hours can be very “dirty;” other times they are greener.
Brainbox AI is optimizing its use of greener energy through a partnership with WattTime, a non-profit that offers technology solutions to achieve emissions reductions. In effect, they are incorporating more environmental variables into the AI engine that drives a building’s systems.
WattTime is a U.S.-based entity, and the Brainbox partnership is expanding usage to Canada, Europe, Australia, and other parts of the world.
Effectiveness of AI in HVAC
AI engine can detect problems like a defective sensor or a valve stuck open, which might have gone unnoticed
The AI system also helps to maintain an HVAC system. Connecting to a system, the AI engine can detect such problems as a defective sensor or a valve stuck open, which might have gone unnoticed by an operator or maintenance company.
The company can provide a “defective list” to a building’s owner. Some will make the repairs immediately, others may not. The effectiveness of the AI system is limited by the equipment that is in place. “We do what we can with what we can,” says Venne.
Using self-adapting technology
There are about 700 different control platforms in the HVAC market, although a smaller group of about 17 controls about 80% of systems. Brainbox AI interfaces with the 17 main building systems, which include popular protocols such as BACnet, Tridium, Modbus, and Johnson Controls.
“It’s important to understand that each building is unique in how it is configured and operates,” says Venne. “There are no copy-and-paste solutions. We need technology that self-adapts. The neural network adapts to buildings and retrains it to evolve without human intervention.”
Brainbox AI launched in May 2019 on the commercial side, after a four-year development period. The system has been deployed in 100 million square feet of building space in 15 countries and 70 cities globally, reaching all time zones. There is a large installed base in Australia, New Zealand, the Middle East, Canada, and the United States. The company is expanding rapidly and has 100 employees.
