BuildingIQ, a provider of energy management software headquartered in California, has developed technology platform that optimizes heating, ventilation, and air conditioning (HVAC) systems in commercial buildings for maximum efficiency. Using machine learning, the algorithm-based system provides “round the clock” energy management optimization based on a thermal energy model, which the system creates using both internal and external building data, including metrics related to occupant comfort. BuildingIQ’s solutions are in part based on technology originally developed by the Energy Division of the Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia’s national lab.
Why There Is a Need
The average commercial building wastes 30% of the energy it uses. Even simple changes, such as adjusting temperature settings and ensuring that lights are off when no one is around, can produce substantial savings. Traditionally, optimizing an HVAC system required technicians to walk through a building with a clipboard in hand, generating an extensive report on energy usage and strategies to control the HVAC system. This approach was generally very time- and labor-intensive, and often a very expensive way to save money. Even with the report, building operators were still required to take the technicians’ recommendations and act on them. Working with the current facilities management team, BuildingIQ is not only able to identify HVAC control recommendations, it can also remotely implement them and improve efficiency over time.
After the software is installed, BuildingIQ creates an energy profile for the building based on variables such as occupant comfort, weather forecasts, energy tariffs, building characteristics, and historical and real-time meter data. Once these guide rails are identified, the platform can make incremental temperature and pressure changes to a facility through the building management system (BMS) to optimize energy consumption and energy costs, while maintaining occupant comfort. As artificial intelligence (AI) continues to progress, the company expects that the next stage of innovation will focus on optimizing other sources of energy consumption such as chilled water systems and lighting, optimizing mixed use of renewables, integrating cogeneration and energy storage systems, and achieving operational efficiencies through automated diagnosis of equipment faults.
How It Works
Building automation is an example of a distributed control system (DCS). In a DCS, control elements are distributed throughout the system, in contrast to a centralized system, which uses a single controller. A DCS lends itself to “narrow AI,” which is AI focused on a single task. A building controlled by any building automation system (BAS) is often referred to as an intelligent building or a smart building. It is important to note that BuildingIQ was not designed to replace either the existing BAS or the human facilities management staff – on the contrary, when layered on top of the existing BAS, it is an example of computer-aided DSC control. To achieve maximum efficiency, BuildingIQ’s AI-based system produces a new optimization profile for the BMS every 4 hours and can include changes to BMS settings as often as every two minutes – enabling the building to respond in real time to changes in weather, occupancy, energy costs, and even demand response (DR) signals from the utility. Even so, the system cannot positively identify every anomaly, and thus requires the insights and expertise of both the existing facility staff and BuildingIQ’s building team to troubleshoot and fix certain problems, such as the heating system not containing any water, which is not a problem AI can solve.
Not all smart buildings use AI, but AI applied to an existing BMS does enable smart control systems to learn about human habits and preferences without being redesigned, or worse, replaced at great cost. This makes it affordable to personalize building automation systems en masse and leverage scale to identify best practices across buildings.
This technology is not a panacea. AI algorithms will identify patterns that are statistically valid, but conceptually false. At their statistical heart, AI analyses will produce false positives and false negatives and make mistakes no human would ever make. Relying on these conclusions too heavily may result in false conclusions. In addition, humans can hack and game AI systems, a problem often referred to as “split incentives” wherein tenants will be tempted to optimize the system for their own comfort and not the interests of the building owner or the environment. To overcome this, BuildingIQ has implemented an expert team of operators, building engineers, and data scientists, who monitor each building’s performance and work with the onsite building operators to detect and fix anomalies.
The Path Forward
Since this is such a strong use case for AI, it is hard not to imagine BAS vendors such as Honeywell, United Technologies, Schneider Electric, Siemens, Legrand, and Johnson Controls will not leap on the bandwagon. Schneider Electric is in fact an investor in BuildingIQ though Aster Capital. The drive toward AI in building automation may also include companies making HVAC controls such as Azbil, Building Robotics, Delta Controls, Distech Controls, KMC Controls, Reliable Controls, and Trane.
Pioneers like BuildingIQ are part of the reason we feel the Building Automation industry is well positioned to leverage AI capabilities. In Tractica’s Artificial Intelligence for Enterprise Applications report, we forecast that spending on AI software in the building automation industry will grow to $252.4 million by 2025.