The term “artificial intelligence” (AI) often conjures images of upright androids delivering mail or pontificating on Isaac Asimov’s Three Laws of Robotics – and with mechanical men like Honda’s ASIMO already learning how to walk, run and cut a rug, that robotic future doesn’t seem far off. But AI is more than dancing robots.
If we want to get technical, artificial (or computational) intelligence refers to any system that acts intelligently. As some experts define it, an intelligent agent or system is one that “is flexible to changing environments and changing goals,” can learn from experience and “makes appropriate choices given perceptual limitations and finite computation.” With this definition, something as simple as a thermostat can be said to have artificial intelligence, as its behavior is determined by its environment.
Perhaps the most famous, non-fictional AI systems are the two developed by IBM, Deep Blue and Watson, which have outsmarted humans at chess and Jeopardy! respectively. But, just as AI is more than dancing robots, it’s also more than competitive gaming.
Energy Waste in Today’s Commercial Buildings
The next generation of energy encompasses more than simply cleaner, renewable sources of power like solar, wind and biofuels. It also means optimizing the energy that is used by our homes, offices and machines.
According to the U.S. Energy Information Administration, commercial buildings account for almost 20 percent of U.S. energy consumption and emit 12 percent of the country’s greenhouse gases. That’s a major energy investment, and MIT has committed long hours to analyzing just how much of it goes to waste. The Natural Resources Defense Council has gone so far as to give this waste a name: “operational stray.” The term refers to building equipment that “strays” from optimum settings, such as bugs in software code that trigger automated systems or thermostats that don’t reset to their usual schedules after some impromptu fiddling.
Architects like Planet Expert Scott Rodwin have developed green building techniques that utilize superior insulation, geothermal heating and cooling, and sustainable materials to conserve energy and reduce emissions. These net zero, LEED-certified methods do wonders on the brick and mortar front, but what about the systems that make commercial buildings livable? What about the HVAC side of the equation (how the industry collectively refers to heating, ventilation and air conditioning)?
That’s where a really good AI comes in.
How to Boost a Building’s IQ
Recognized by Bloomberg as a New Energy Pioneer in 2013, BuildingIQ represents the next generation of smart building tools. The Australia and California-based company has developed an algorithmic energy management platform that optimizes a given HVAC system for maximum efficiency. The company’s work at the Australian Technology Park (ATP) was recognized as the “Best Commercial Energy Efficiency Project for 2014” in Australia, and the same technology will soon be implemented in U.S. government buildings as part of the GSA’s Green Proving Ground Program.
How BuildingIQ does what it does is complex. It is an algorithmic, machine-learning AI that provides 24-hour optimized energy management based on thermal energy occupancy. What it does is much easier to explain. It saves energy and money.
As of this writing, BuildingIQ has installed its AI in over 15 million square feet throughout North America and Australia and plans to expand into Europe and Southeast Asia in the near future.
To understand both what and how the company does what it does, I contacted Peter Dickinson, BuildingIQ’s Chief Technology Officer, whom oversees the company’s product management team. Dickinson has worked in the energy management and commercial property sector for 15 years.
Before this kind of energy-saving platform was introduced, said Dickinson, efficiency was primarily tied to (and inhibited by) human operators.
“When I was introduced to this technology in 2009,” he said, “I’d been around the industry for about 10 years, and at that point I’d been doing energy monitoring and fault detection.”
Traditionally, he explained, “optimizing” a given HVAC system involved engineers walking up and down a building with a clipboard in hand and generating a 100-page report on energy use and control strategies. This approach was “very manual, very time- and materials-heavy, very customized for [each] building, and generally very expensive.” Building operators were still required to take the engineer’s recommendations and act on them.
“The circle was still a little bit open,” said Dickinson.
This is where artificial intelligence usurps the traditional way of doing things. The BuildingIQ platform not only acts on the full spectrum of engineer-level recommendations, it implements and updates them in real-time.
After the software is installed, BuildingIQ creates a power profile of the building based on an operator’s set of ideal temperatures. The AI then initiates what Dickinson calls a “sort of hill-climbing algorithm” that tests tens of thousands of different iterations, moving those set points around the building to get the power loads, the total cost, and doing so until it finds the combination that represents the lowest cost way to navigate the next 24 hours.
“My mind was basically blown,” said Dickinson, recalling the first time he saw the AI in action.
“Suddenly, we’re moving beyond what 99.9 percent of the market is focused on, which is: ‘How do we get the absolute efficiency at this exact point in time?’ We’re now changing that conversation to go: ‘Well, what about the best combination of decisions over the next 24 hours?’
“We’ve gone from being very blinkered and looking at right now to being able to look over the horizon where it might make sense to actually ramp the loads right up while power is cheap, or before a demand-response event, or before you cause a demand charge event. It might make sense to shift load earlier or hold the load off till later and make all these very complex decisions that are almost impossible for a human operator to make.”
Dickinson is quick to add, however, that the AI is not intended to replace a human operator but rather to complement it. “A lot of what I do is working towards the best combination of ‘what are humans really good at?’” he said. “How do we complement algorithms with the power of the human brain and the human eyeball?”
Once plugged in, the software can take a 10-year-old building and present its multiple temperatures, humidity and ventilation systems in an easy-to-read interface. Dickinson mentions that they have even managed to optimize a 20-year-old system, though not without overcoming some challenges.
Earning the Industry’s Trust
After five years, Dickinson says the industry can still be reluctant to adopt this paradigm shift.
“We’re still in a very conservative industry,” he said, “even in San Francisco where most companies are willing to give anything a try. When it comes to the tool-belt guys in the basement, they’re still super-conservative and they want to know who you are and they want to put you through your paces.”
However, many companies are now on the lookout for green innovations. Installing LED lights and switching to more sustainable practices represent the “low-hanging fruit” of the clean energy transition. BuildingIQ, said Dickinson, is like an introduction to the next tier. “These companies that have been talking green for some time,” said Dickinson, “are running out of steam a little bit, because they’ve done so many innovative things. They’re like, ‘Oh man, what’s next?’”
There are two things that have helped the company turn around reluctant parties, he said. One is the relative ease of its installation. “We’ve been able to get our installation down from the industry standard of two-to-three weeks,” said Dickinson. “Our current record for an installation with a couple pieces already in place is about 20 minutes.”
The second is BuildingIQ’s guarantee of no upfront costs. “The system will return more savings than it costs to run,” said Dickinson, “and if it doesn’t we’ll run the system for free until it’s paid for itself.”
That message has been a powerful one, he added, because “so many of the other solutions that are out there…suggest you might get five to ten percent savings, but they don’t actually guarantee it. The fact that we’re putting our money where our mouth is has been getting a lot of traction in the marketplace.”
BuildingIQ has reported customer savings between 10-25 percent in HVAC energy costs.
In contrast to the Congressional gridlock over the future of energy, Dickinson said the federal government has been eager to explore green innovations.
“The GSA [U.S. General Services Administration] has been really proactive over the last decade or more in trying to both improve their own performance and keep up with the industry, between the DOE [Department of Energy] and the GSA, they’ve taken on this real leadership role of giving newer technologies a chance to get started in the market.”
Dickinson was enthusiastic about the BuildingIQ’s relationship with Green Proving Ground, the GSA program that has contracted with the company.
“Green Proving Ground gets you into the federal government system,” said Dickinson, “it tries you out, and unlike so many companies and programs where you just get piloted to death, the GSA has a very strong mandate that’s based on levels of success. They want to help you go from one to many very quickly. So that’s exciting for us.”
Countering the Homeland’s Bi-Polar Energy Attitude
BuildingIQ’s computational intelligence offers a fascinating snapshot into the future of the energy industry, but its home country seems to vacillate between that future and the indomitable past. One day the University of New South Wales is setting solar conversion records, the next day the government is repealing its carbon tax.
I asked Dickinson if this bi-polar relationship with energy has affected the industry’s innovators.
Dickinson acknowledged that the recent change in government has become more conservative, and that with so much coal in the ground there are plenty of vested interests who want to take advantage of it. “At the same time,” he said, “the thing that a lot of folks don’t quite pick up on —and a lot of people are only just starting to get angry about in Australia— is the excessive amount of infrastructure building that’s gone on.
“We basically have a gold-plated energy infrastructure, because after the privatization a lot of the poles and wires companies have had basically an open checkbook with a guaranteed return on investment to go ahead and build whatever they want. So they emphasized capacity problems, everybody began installing domestic air-conditioning and they built and built and built.
“But at the same time, companies and local and state governments were doing all of this work —a lot of real estate, a lot of big property trusts in Australia at the top 10, top 20 of the sustainability index— the industry itself has gone, ‘Well, we’re going to do this energy efficiency stuff because it makes sense.’ And so the net result of that is this gold-plated infrastructure but a net reduction in a lot of the energy demand and network loads.”
What the country is left with is a dichotomous energy profile.
“We’ve got this weird situation,” said Dickinson, “where our energy performance market in the building is at the pointy end of worldwide performance because we’ve been doing sub-metering, rebate programs, for a good 20 years.”
As the company and its client base continue to grow, Dickinson says BuildingIQ hopes to expand into the European and Asian markets. And as the industry itself continues to progress, he adds, the next stage of innovation will focus on systems integration, or getting multiple AI controllers to coordinate with each other in a single building.
“Systems integration in the last decade or two has been about, let’s get the video system, the access control system, the fire alarm, HVAC, lighting, all [wired] back to a single operator work station and make all of the data available in a single database. And that’s been good and that’s got a lot of value. But what we’re seeing now, if we look ahead, none of those previous solutions were really that terribly intelligent. They weren’t making decisions for themselves; they didn’t need to coordinate between each other too much.
“If you take just a simple example of BuildingIQ being in the building with an intelligent battery pack, suddenly you have two very intelligent systems in the building, and you want to ensure that they’re coordinating. And that’s very different from previous systems’ integration approach with a single operator workstation and a single database. Now you need some way to ensure that decision-making is happening in some kind of coordinated fashion. That’s going to be the AI coordination platform.”