Saving on Energy in the Future
Ever wanted to know just how much energy each of your appliances uses?
Henning Lange, a first-year PhD student, was confronted with this question as a master’s student living on a budget in Berlin.
Lange and his roommates had amassed an enormous energy bill, but they couldn’t determine which appliance was the culprit that caused the bill to skyrocket. To pinpoint the appliance, Lange wanted to see how much energy each of the appliances in his apartment was using individually, something that is not currently possible.
“We wanted to save energy … but we didn’t know where to start,” he said. Since Lange became interested in breaking down energy usage by appliance, he has worked at two European companies that are also trying to solve this problem. During his time at those companies, he realized it wouldn’t be easy to find an answer. To do so, he would need to perform his own research.
Lange found the perfect place to start that research within the CEE department.
He spent last summer visiting the on-campus lab of professor Mario Berges, an expert in the field of non-intrusive load monitoring (the technical term for the problem Lange is trying to solve). In his lab, Dr. Berges is researching ways to make built environments more operationally efficient and robust by using information and communication technologies.
Shortly after Lange visited the lab, he joined it and began searching for his answer.
At CMU, Lange is studying electrical currents to help him break down total energy usage by appliance. Through measurements similar to those obtained by utility meters, he is analyzing the aggregate, or total, electrical current that appliances in a home or building are drawing at a particular time. The current waveforms on their own, however, don’t tell him how much of the total power each of the devices is consuming.
With an algorithm that he created, Lange can identify characteristics of a single appliance’s power usage within the total current. With those characteristics, he can infer an individual current for the device that tells him how much power it is using on its ow
For example, if a washer, television and lamp are on at the same time, the algorithm can parse apart a current that shows how much power the devices are using combined. The algorithm can find characteristics for the washer, television and lamp in the total current that Lange can then use to determine individual currents for the devices. By doing so, he can tell how much energy each appliance is consuming.
“At the moment, you’re getting an electricity bill, but you don’t actually know where the energy went to,” Lange says. “What we are trying to do at the moment is find building blocks, or patterns, in the data that constitute current waveforms of different appliances.”
If he is successful, the benefits of Lange’s research are far-reaching for energy efficiency and the environment.
Disaggregating energy by appliance can be beneficial for savings. If people can see which of their appliances or devices in their homes are consuming the most energy, they can limit those heavy users to save both money and energy.
If, as a student back in Berlin, Lange had been able to determine which device was driving up his energy bill, he and his roommates could have more consciously used the appliance.
Lange’s research, in addition to having applications in residential buildings, can be used in commercial buildings, where identifying devices that use the most power can help businesses and other organizations rein in their energy usage.
The outcomes of the research could also help people determine at what times during the day they should use, or refrain from using, their appliances to avoid peak energy demand periods. Peak demand periods occur when people together use more energy than is usually expected. To account for increased loads, additional power stations may need to be activated during these periods. Energy usage is also less efficient, and carbon emissions increase, during these times. But if people more evenly spread out the times during which they use their plugged-in appliances, they can together one day more evenly distribute energy usage.
Currently, Lange’s model for breaking down currents depends on knowing what devices are plugged in beforehand. By the time his research is complete, he aims to make the algorithm work without that prior knowledge.
“Computationally, [energy disaggregation] is a very hard problem,” he said.
If Lange has anything to say about it, though, homeowners and businesses will one day be able to pinpoint which appliances are their biggest energy consumers – and adjust their usage accordingly.