First, how do we define AI? In commercial buildings AI is often thought of as applying analytic engines on gathered information then use those models, algorithms, and rules to make better decisions on how the systems are operated. Most of us aren’t programmers and we use the term AI with the intention of adding sophistication and analysis that might not exist in the detail that might be implied.
In many ways we are using integrated building analytic tools in ways that are advanced controls sequences, evaluations and continuous building system optimization. AI can then aim to advance beyond this simplified use to do predictive controls, which looks at the operating characteristics of the equipment and systems and then models a prediction of the ideal operating points should be for the given conditions. While this logic might constantly hunt for the perfect point, the AI also learns how the systems react to prevent runaway of the equipment, systems, and conditions before being applied beyond the model. In many models the AI learns a method of ‘trimming’ instead of applying wild swings, which allows the system to gradually bring ideal operating points to the equipment and optimize the system.
This approach for AI is more computationally intense but has the potential to improve efficiency and performance dramatically as well as begin to sense and predict failures before they occur. When applied to existing facilities, AI could vastly improve on an outdated building management or automation infrastructure. In many ways the use of AI is the next phase in building controls and is likely to disrupt the industry as much as DDC did to pneumatics years ago. Engineers and designers may soon be developing algorithms and boundary logistics instead of a sequence of operations.
As a reminder, AI is not a replacement for a human operator. It is another tool for us to utilize, such as software tools to predict loads, cooling calculations, commissioning programs and more, albeit in a more comprehensive manner. There seems to have always been an ethical concern when applying AI, whether to the building industry or others. With Asimov’s laws the data center will continue to serve the people (carbon-based life forms) first, through safety and security, and the machines (silicon-based life forms) second.
In the coming decade it is very likely that the firms that design, build and operate data centers will make investments in learning and applying AI. Most want to be on the forefront of the industry and see a future with AI as a complimentary technology to aid project delivery. As more datasets emerge with higher granularity and quality we will see AI as a more trusted means of marching toward efficiency and eventually being able to point out weaknesses and points to improve how the data center, and perhaps campus, can reach its pinnacle of performance – every day of the year.