Earlier this week, we covered an EPRI study that indicates how energy disaggregation technologies can yield very different levels of accuracy on parsing out energy use at individual household system and appliance levels, depending on how they go about getting the whole-home energy data that they use as their source.
One of those findings is that systems that use hardware to analyze household power can deliver more accurate and detailed disaggregation, compared to software-based systems, which take various types of available whole-home energy data and subject it to analysis in the cloud.
But it’s important to remember that EPRI’s research represents only one of many scenarios for how energy disaggregation technology can be deployed in the home. Perhaps bringing more granular whole-home energy data into the picture could help those software platforms get more accurate – and, maybe getting the people living in the home to play a role could help even more.
At least that’s how Peter Porteous, CEO of Blue Line Innovations, sees it. Blue Line has connected about 300,000 homes with its meter data collection sensors and home energy monitoring tech, mostly in Canada’s Ontario province, which has mandated smart meters and time-of-use pricing for all its residents.
Over the past year or so, Blue Line has been working with Bidgely and PlotWatt, two companies that provide cloud-based software for energy disaggregation, to help those customers get more insight into that day-to-day energy use, from big and easier-to-predict cooling and heating loads, down to individual appliances and consumer electronics.
That’s a lot more detail than the software-based systems were capable of delivering in EPRI’s test. But then, Blue Line’s sensors can provide much more granular data, down to every 25 milliseconds, than one would presume EPRI’s test was delivering -- most smart meters deliver data to utilities in hourly or fifteen-minute increments, for example. And more frequent data for Bidgely and PlotWatt’s algorithms yields greater likelihood of accurate, pinpointed results.
“It takes about two weeks before the customer would see usage down to appliance level,” Porteous noted, because the algorithms need enough data to get a sense of how each individual home operates. But once the system has been up and running for awhile, Bidgely and PlotWatt can break out how many dollars per day in energy bills are accounted for by air conditioning, water heating, refrigerators, washers and dryers, and even lights and plugged-in loads.
Accuracy, Credibility, and Customer Buy-In
But before Bidgely and PlotWatt start promising this kind of dollars-per-day accuracy, they have to take caution to make sure they’re accounting for another kind of unpredictable behavior -- consumer acceptance.
“I think credibility trumps accuracy” when it comes to detailed home disaggregation data, Porteous said. “What I mean by that is, if this is your home, and you’re connected, and all of a sudden the data comes back and says, “You’re using $10 a month on a hot tub,” and you say, “I don’t have a hot tub,” well, you’re never going to look at the data again.”
That fundamental understanding of human nature is why Bidgely or PlotWatt take a “very risk-averse, conservative approach to revealing usage,” he said. Customers are asked to supply their feedback as the systems start to supply big household load data after their two-week learning cycle, he said. They’re also asked to help the systems identify loads that don’t track precisely to the libraries of data on appliance and system energy usage characteristics that both companies use, he said.
To do that, the web interfaces create what’s known as a “learning category” for loads it hasn’t identified, he said. “We can see waveforms, we’re pretty sure we can associate those back to appliances, but we’re not yet confident,” he said. “We may even come to you with a question, saying, ‘Will you confirm with us that you have, say, three huge aquariums?’ There will be steps to ensure credibility before those are revealed.”
But once the system has gone through that process of identifying and confirming individual loads, “We’d say there’s a 95 percent degree of confidence in the number revealed,” he said.
That’s not just valuable for the homeowners, he noted. “The response from the utilities, when they see now that they can have access to this data, it really is game-changing,” he said. “That’s their main frustration -- they have no idea what their customers are doing inside the home, and spend gobs of research trying to figure that out.”
It's clear that utilities will have to navigate delicate issues of customer data privacy and security to be able to get customers to participate in this process, or to tap the more frequent data that can come from in-home sensors, rather than fifteen-minute or hourly smart meter readings.
That means that “understanding the human-behavior, human-science part of it is as important, if not more important” than the technical challenges involved in energy disaggregation, Porteous said. “The companies that will be successful will be the ones that have both.”