Tempering Hot Summer Days with the Cool Hand of Big Data

Machine learning is making buildings more comfortable and energy-efficient
Buildings use a lot of energy for heating and cooling. Pixabay

Share

We all know the feeling of walking into an air-conditioned office building on a hot summer’s day. It’s 94º outside. It’s 64º inside. Suddenly, you need a knit sweater just to sit at your desk. (I know someone who used to bring a space heater to work — in August.)

Aside from being uncomfortable, this is horrifically inefficient. The low hum of the air conditioner is the sound of coal being burned and money being spent to make you miserable at work. BuildingIQ is trying to change that. Their weapon of choice? Machine learning.

The company has developed cloud-based software that gathers information about a building and uses it to build a thermal model. That model can predict how much or how little energy is needed to keep the building’s occupants comfortable and adjust the temperature accordingly.

The model analyzes a number of variables — indoor temperature and pressure, electricity consumption, the weather forecast, the price of electricity — and tailors heating and air conditioning for maximum energy efficiency.

Building overview from BuildingIQ
Building overview from BuildingIQ BuildingIQ

It might ramp up air conditioning early in the morning when power is cheap to pre-cool the building, cutting consumption during peak hours. Or, it may see a cold front approaching and turn down air conditioning to compensate. The software uses machine learning to constantly update the model.

“That thermal model continues to learn from itself,” said Michael Nark, President and CEO of BuildingIQ. “We model a building in downtown New York City and it has a glass facade that is south-facing, so it’s going to heat up all the time. Six months later, the plot next door has another building that’s built on it that’s 6, 8, 10, 12, 15 stories tall that blocks the facade. That changes the thermal characteristics of the building.” With machine learning, the model is updated to reflect the changing conditions.

Machine learning can be found everywhere. Amazon, Spotify and Siri all use it. It’s how Netflix learns that you like comedies with strong female leads. The more data, the better. Thanks to a recent acquisition, BuildingIQ will soon take human preferences into account.

Changes to heating and air conditioning are always bound by the limits of human comfort, and those limits vary from person to person. Designers assume people like the temperature in the high 60s, but some may prefer it warmer or colder. Others may be indifferent.

Nark imagines a time when office workers could use their phones to tell the building when they are too hot or too cold. It wouldn’t work like a local thermostat. Rather, this data would be integrated into the thermal model. Imagine a retirement home where the occupants are always too cold. If the system knows this, it will ramp down the air conditioning in the summer to keep grandma and grandpa nice and cozy.

How well does it work? BuildingIQ says it regularly produces energy savings of between 10 and 25 percent on heating and air conditioning. HVAC comprise 50 or 60 percent of a building’s total energy use. Unfortunately, only one in ten buildings are outfitted to take advantage of the system, but BuildingIQ has developed software to serve older buildings with more rudimentary management systems.

The software will help limit carbon pollution by making it convenient to use less energy — slashing energy bills and improving comfort all while reducing our reliance on fossil fuels. It won’t be long before you can leave your space heater at home.

Jeremy Deaton writes for Nexus Media, a syndicated newswire covering climate, energy, politics, art and culture. You can follow him at @deaton_jeremy .

 
The best Black Friday deals including a jackery generator, airpods, a TV arranged on a plain background.

PopSci's Guide to Cyber Monday

The best Cyber Monday sales, deals, and everything else you need to know. Our team spends hundreds of collective hours searching and evaluating every deal we can find online, focusing on well-made and reviewed products for prices that make sense.