Why waste watts?

At home and at restaurants, PlotWatt identifies energy-eating appliances and practices

ChallengeGather, clean, organize and analyze data from numerous sources to guide decisions about energy management.
SolutionPlotWatt uses JMP® software to identify opportunities for optimizing electricity use and refine their machine learning algorithms.
ResultsWith the information PlotWatt provides in daily reports, fast-food restaurant managers have lowered energy costs and improved customer satisfaction.

Using electricity is like shopping at a grocery store in which the items have no price tags. You reach the checkout line with no idea of how much you’ve spent on what. So says Daniel Arneman, Director of Energy Analytics at Durham, NC-based PlotWatt.

PlotWatt’s mission is to change that dynamic, to provide an itemized accounting of how homes and businesses are using, and misusing, energy each day.

PlotWatt’s website quotes Yogi Berra: “You can observe a lot by just watching.” But the practiced eye will see much more. PlotWatt’s vigilance is studied, thanks in large measure to JMP software. JMP offers insights that translate into money-saving recommendations.

In the fast-food industry, for example, PlotWatt analyses are saving franchisees thousands of dollars per year just by revealing wasteful practices that would otherwise remain hidden from managers’ sight.

Arneman says that most of us assume that the things we see around us are the things that cost us. “People ask me, ‘How much energy does my blender use?’ ‘How much does my microwave use? If I did just a little less microwaving, would I save money?’

“The answer is that your microwave is on 30 seconds to two minutes a day, and it really isn’t a significant contributor; it’s really almost invisible in the span of the energy usage in your day.”

Similarly, businesses are in the dark about energy misuse. The actual culprits are lurking out of view. PlotWatt nabs them.

Each with its own ‘wattprint’

Back when PlotWatt founder Luke Fishback worked at Lockheed Martin (yes, he was a rocket scientist), he began questioning why his energy bill was so high. He took what Arneman describes as a “high-tech/low-tech approach” to his investigation.

Fishback set up a digital camera to photograph his energy meter every minute or so, and entered the data into spreadsheets. His friend John Cunningham – at the time completing a PhD in machine learning at Stanford and now PlotWatt’s chief scientist – then wrote an algorithm for “energy disaggregation,” thereby planting the seed from which PlotWatt sprung.

Energy disaggregation begins with the premise that each appliance uses energy in a unique way, each with its own “wattprint.” PlotWatt's algorithms sleuth out these wattprints to determine which appliances are the real energy-consumption culprits.

“We tend to think our lights are using a lot of energy,” Arneman says. “But now that many people are switching to compact fluorescents and LEDs, lighting becomes less and less an energy consumer. But that DVR the cable company gave you is actually consuming quite a bit of energy.”

Heat pumps, septic pumps – these too are common wasters.

PlotWatt also tracks always-on energy. “Sometimes people have very high always-on energy use and we can target shutting those items off and see the results,” Arneman says. “The savings really add up.”

Invisible consumption

Arneman joined PlotWatt in 2011, bringing JMP with him. A dedicated JMP user in his previous work at the University of North Carolina-Chapel Hill, Arneman had helped the university save millions of dollars in energy costs. He came to appreciate the ability of JMP to easily gather, clean and organize large quantities of data from disparate sources, enabling him to ask the right questions of that data and get reliable answers.

At PlotWatt, Arneman uses JMP in his work with homeowners, but it’s with restaurants that he’s really able to demonstrate the value of statistical discovery.

Fast-food restaurants consume up to five times the energy of the average commercial business, and utility bills represent a significant portion of their operating costs. Even so, electricity consumption and associated costs have been “largely invisible to them,” Arneman says. “The energy bill comes in the mail, it goes straight to the main office, and the store manager knows nothing about it. The owner may not even see it except in the profit-and-loss statement as a group aggregate.”

PlotWatt helps reduce energy bills by providing information that store managers can translate into waste-cutting action. They might adjust appliance start-up and shutdown schedules or air conditioner settings. They might reset timers for seasonal changes, so that outdoor lights don’t come on two hours before (or two hours after) sunset. By detecting failing and broken equipment, PlotWatt spurs restaurant personnel to take corrective action sooner.

One fast-food franchisee gained $779 in monthly savings per store; another saw monthly savings of $649 per store.

It’s not strictly about savings, though. PlotWatt also aims to improve customer satisfaction.

“When you enter that store, you expect a certain quality of product,” Arneman says. “If, say, the crew has shut down the ice cream machine early, you’re not going to get that quality. With any type of appliance, we can see and understand the behavior around it and provide feedback.”

Piecing together the data puzzle

Usage reports and suggestions for corrective action are all made possible, of course, by PlotWatt’s ability to easily gather and make good sense of data.

“We got into multi-location restaurants because of our ability to transfer what we know from one store to another,” Arneman explains. “We started with one fast-food store, then went to two, then to 10, then to 100…”

Scripting features in JMP paved the way. “When I need to deep-dive into how a store is performing,” he says, “I can take the standards I’ve used and very quickly make sense of an individual store or a group of stores by having the scripts already prepared.”

Arneman gathers a variety of data from a number of sources. Information about weather, financial, sales and more each provides a piece of the puzzle.

“The weather data combined with the energy-usage data and PlotWatt Non-Intrusive Load Monitoring algorithms tells me something about how the air conditioning is working,” Arneman says. “And the point-of-sale data joined with the energy data tells me whether sales are actually driving up energy usage or whether the two are not really related, it’s just waste.”

He does a lot of data exploration, and says he’s not always sure what he’s going to find. “In fact, I’m almost never sure what I’m going to find. JMP brings order to it all.”

JMP Graph Builder is a tool Arneman uses every day. If he is looking at a particular franchisee’s holdings, he can pull up data for performance at those stores over, say, the past three weeks and filter out the ones that are open 24 hours. Being able to do all that in one graph expedites his research.

“Having the tools to be able to quickly go through and visualize allows me to bring order to this data,” he explains. “So over time, what this has led to is a core set of views that started to make sense to me as we expanded.”

With the insight gathered, PlotWatt sends an automated daily email message to the store manager with information that can spur immediate action: “Someone left the stove on overnight.” “We’re noticing a bit more energy use on your walk-in freezer; check to see if there’s ice buildup.”

Wiser choices

Database connectivity tools in JMP allow Arneman to get to his results quickly: “The ability to open up web pages is really key, because the data does come from a lot of sources, and being able to directly access those sources saves me a step.”

He also uses JMP variability gauge charts, which let him analyze the performance of stores over extended periods of time.

Bottom line, it’s about making sense of data, communicating analysis findings, and, ultimately, making wiser choices in how to use energy.

Coming soon: wider access

Arneman believes that it will soon be easier for consumers to gain access to this type of information. Because it costs utility companies more money to provide energy at peak times, it’s sometimes in their interests to encourage consumers to reduce their energy use during those hours.

And in areas where deregulation is offering customers more choices about how they buy electricity, providers need to offer incentives. A feature that allows customers to see how their energy is being spent, itemized on their bills, would be a differentiator.

Back at PlotWatt, Arneman is developing tools to measure the effect of the messages he sends. “We can learn what types of messages get people to switch off appliances overnight, for example, or what types of messages are most effective at getting people to adjust the thermostat,” he says.

Is it better to say something positive, or something slightly negative? Should you send the same message three days in a row, or is once enough?

“We are able to mine the data we have and set up experiments to drive this behavior change in a scientific way,” he says. “JMP is my choice for doing this.”

We are able to mine the data we have and set up experiments to drive this behavior change in a scientific way.
Daniel Arneman

Director of Energy Analytics

Additional Links
Contact JMP Sales

Back to Top