Dive Brief:

• Smart HVAC room controllers equipped with Schneider Electric’s proprietary “edge AI” model reduced energy consumption relative to room controllers without AI by 5%, on average, the company says in a March 17 report.
• In field trials at four Canadian facilities from mid-November through late January, the controllers reduced energy consumption by as much as 15% “under specific operating conditions” while successfully maintaining temperature regulation and comfort compliance more than 85% of the time, Schneider says.
• “This is a real application of AI in a use case that actually affects the built environment,” Peter O’Neill, vice president of BMS controls and devices in Schneider’s digital buildings division, told Facilities Dive.

Dive Insight:

Static control strategies and lack of real-time adaptability mean traditional HVAC systems struggle to optimize building energy efficiency and occupant comfort, Schneider says in the report.

Earlier this year, Schneider debuted a room controller that it says can reduce energy consumption by up to 35% when integrating HVAC, lighting and blinds operation. The new controller offered an opportunity to “enhance features that would have lived in the cloud and bring them down into the room,” O’Neill said.

Each room controller is equipped with AI that processes historical temperature data, occupancy patterns and environmental conditions in real time to predict the ideal temperature path for the room and dynamically adjust climate control settings, Schneider says. The model measures the results and learns over time, refining its thermal models and optimizing temperature predictions.

The extent of the model’s apparent energy savings surprised the Schneider team, O’Neill said. Optimized stop and start times, fewer temperature set point changes and room-level rather than floor- or building-level controls all contributed, he said.

“In that room-by-room granularity, there’s more of an opportunity to save,” he said.

The field trials also revealed a surprising level of user compliance. Very few users chose to turn the AI feature off despite the controller making it easy to do so, O’Neill said.

“That would tend to indicate that people are not noticing an issue with the environment they’re in, which we were most pleased to see,” he said.

Schneider ran the field trials at three of its offices in Montreal and Edmonton, and a fourth non-Schneider, “early adopter” location in Ottawa. One of the Montreal locations deployed 67 controllers across open office spaces, closed offices, conference rooms, a cafeteria and a range of other environments, the company says.

Schneider designed the new room controller to integrate into standard thermostat wiring for ease of installation, O’Neill said. And while the field trial locations were required to use Schneider’s EcoStruxure Building Operation building management system, the AI-enabled HVAC room controllers are “BMS-agnostic,” he noted.

“We don’t want to say that, if you don’t have a Schneider BMS, you can’t use this product,” he said, noting that it’s the “first device of this type with AI on the edge.” 

With field testing complete, Schneider is rolling out its AI-enabled HVAC room controller offering to commercial clients. One early adopter is a major hospital in Saudi Arabia, O’Neill said.

As commercial deployments continue, Schneider aims to roll out additional indoor air quality features and eventually quantify the product’s impact on facilities’ carbon intensity, O’Neill said.