Episode 201: Are researchers using smart grid data?
Are researchers using smart grid data? On this week’s Energy Bite, Richard Huntsinger, a recent PhD graduate of Carnegie Mellon University, has some answers.
- Grid Modernization and Smart Grid by Office of Electricity Delivery and Energy Reliability
- No time to think: How utilities are handling the deluge of grid data by UtilityDive
- Data and the Electricity Grid by More Than Smart
HOST: Are researchers using smart grid data? On this week’s Energy Bite, Richard Huntsinger, a recent PhD graduate of Carnegie Mellon University, has some answers.
HUNTSINGER: Researchers are using it a lot. At Carnegie Mellon University, for example, we developed a software system to analyze about 800,000 electricity demand forecasts made using a wide variety of forecasting processes, based on data from smart grids in Dublin, Ireland and Sydney, Australia. This is the first study to systematically look at so many forecasting processes using detailed data on real consumer electricity usage. When viewed in this broad context, the smart grid data reveals some interesting things that we otherwise wouldn’t be sure about.
HOST: What have you learned from smart grid data?
HUNTSINGER: In our study, we see that forecasters using techniques that do a good job of predicting electricity demand in one location do not necessarily do well anywhere else. However, there are some techniques that do pretty well across many locations, though they aren’t the best for any one location. We also see that you can get good forecasts with data from just a small sample of households, say 10-20% of the full population of a city – but, forecasting based on grouping households with similar electricity usage behavior surprisingly does not help much. With more research on data from more smart grids, we’ll see how well these results hold up.
HOST: Do you ever think about how your electricity usage data is used to predict electricity demand? Take our poll, see the results, and ask your energy questions at Energy Bite dot org.