In Search of a Brain for Buildings ...it’s only a matter of time before someone creates the self-learning software brain that the self-driving building requires.

Brad White, P.Eng, MASc Principal, SES Consulting Inc.

I’ve been spending a lot of time recently reflecting on the Internet of Things (IOT) and puzzling over exactly what the implications are for the building operations industry. The main theme of most definitions of IOT is connected/integrated devices that share information. Connecting IOT devices actually strikes me as one of the least disruptive aspects of IOT. In the building controls world we’ve been connecting and integrating devices that generate lots of data for a long time. Sure, it’s getting a lot easier to connect things and sensors are increasingly diverse and inexpensive, but this aspect is not really transformational. So if it’s not the devices themselves, what is it about IOT that is transformational?

Buildings that Teach Themselves

To start to answer this question, it’s worth considering a couple of IOT applications. On the home automation front, the IOT poster child is Google’s Nest thermostat. The Nest does a lot of things well, but one of its key differentiators from the programmable thermostats that it replaces is that it’s not programmable. You teach the Nest your schedule and its algorithm’s figure out the rest. If we consider a slightly different  technology for the commercial building market, Building Robotics’ Comfy, we see again a very user friendly interface paired with machine learning algorithms to optimize control. Both of these technologies showcase some of the hallmarks of good IOT solutions: self-learning ability that allows for an enhanced user experience (ie no programming) while providing an impressive ability to optimize performance based on a whole lot of real world data.   
 
This all finally clicked for me, when I came across this quote in a recent Automated Buildings article:

I think it's been taken as a given that once we have connectivity, then the clever programming needed to take advantage of that data will follow. What I don’t believe is widely appreciated is that a future of connected IOT devices means a fundamental change in how we approach building controls. IOT has meant the introduction of self-learning software and algorithms in a significant way. These are now starting to creep in around the edges of our controls systems to deal with particularly complex aspects of building operation, like Comfy does. We also see an early version of this in areas like optimal start programs, but these are still rather primitive compared to what’s coming.

If you need more convincing, consider that we can now, or in the near future, provide the building automation system (BAS) with access to data on current occupancy, expected occupancy, weather, current building conditions, occupant comfort preferences, and utility pricing or demand response forecasts. All of these pieces of information could be used, for example, to set your equipment schedules more intelligently. How then do you do that exactly? If you’re scratching your head trying to think of what that sequence of operations might look like, then you and I are in the same boat. We could probably put something together that works ok, at least part of the time. But it would be extremely complex and you could spend a long time tweaking variables and always have more room for improvement. The static code you’ve created would also have to be changed anytime something changed in the building that threw off this delicate balance. This isn’t a particularly elegant solution, especially when you consider starting to do this in every program.

A more elegant solution to this, and other complex automation problems, is already out there in the application of machine learning. Just look at the Google Driverless Car. No one is sitting down programming the car’s every response to every conceivable scenario, but rather teaching it. Incidentally, the car drives itself far better than any human could. The solution for buildings then becomes apparent, a self-driving building. Given that the technology exists and there is going to be a clear demand for its application to the building industry, it’s only a matter of time before someone creates the self-learning software brain that the self-driving building requires. 

Adopting a Customer Service Mentality

Aside from smarter software, the other thing that successful IOT solutions do is engage with their users. Given that solutions like dummy thermostats are still widely employed, this is another area that the building industry is in need of a big leap forward on. To quote Canada’s new Prime Minister, “it’s 2015”.

Building operations, at its heart, is an exercise in customer service, keeping the occupants happy. More and more, businesses and organizations rely on social media to interact with their customers. Customers love this for good reason, their feedback is instantly public, and they get responses. How many buildings do you know that have a Twitter feed for their operations department? Voicemails, emails, and online service tickets still abound. More often than not, the communication only goes one way, with issues disappearing into the ether, at least from the perspective of the complainant. Great operations staff will be responsive and will make communication a priority, but that’s often seen as going above and beyond and not a basic requirement of good customer service.  Technology on its own doesn’t solve this problem, however, if an operations group is prepared to take customer service seriously, solutions like Twitter or purpose built applications like Crowd Comfort can provide the tools to have meaningful engagement.

A Vision of the (not-too-distant) Future

A building that combines sophisticated self-learning software brain with a user friendly platform for engaging and communicating with occupants will be in a position to reap the full benefits of the IOT revolution. Consider how the following scenario would play out today compared with our building of the future.  

[an error occurred while processing this directive] Let’s imagine a typical office building on a typical winter day, maybe somewhere in the Northeast. During the startup routine, a pump serving a reheat loop in one wing of a building fails to start. Today, the building operators might find out there’s a problem when they arrive at work and see a BAS alarm, or perhaps they don’t find out until the first cold complaints start to come in. In the best case, they may have gotten an email generated automatically at the time of the failure and logged on remotely from home to confirm the issue and put an urgent call into a service tech to go look at the faulty pump. Meanwhile, building occupants start showing up for work as usual, only to find their space is freezing. Will the heat be out for an hour, or all day?  Occupants start getting frustrated as the folks who would normally respond to complaints are too busy trying to fix the problem.

In our building of the future, things start off similarly, the pump breaks on startup, then they start to diverge. At the same time that the operations team is automatically notified, the building “brain” considers some information: there’s broken equipment, part of the building is failing to warm up, it’s going to be a cold day. Based on this information, a text message is automatically sent out to occupants who work in the affected area, advising them of the problem, apologizing and helpfully suggesting that they consider working from home, or at least dressing extra warmly if they must come in. Once the scope of the problem is known, the affected occupants are sent a follow-up message from the operations team with an indication of when the problem will be resolved and their space will be back to normal.

In both cases, the problem didn’t get resolved any faster, the main difference is actually the speed and openness with which information is relayed to the affected people. This example highlighted building operations, but this same scenario could be repeated with IT or security. Which building would you want to work in? Which building is more likely to retain its tenants? Thinking though this scenario helped clarify for me the power and promise behind the IOT craze. Connected devices aren’t really the point. The real value is in using information from an ever increasing number of sources to operate buildings smarter and enabling faster and more transparent communications to build better relationships between occupants and support personnel.