In a recent article in Significance magazine, researchers addressed some of the factors affecting energy demands, with a particular focus on the role of statistics in energy forecasting.
“Energy forecasting goes beyond simply forecasting power consumption,” explained Mihai Anitescu, a senior computational mathematician in the Mathematics and Computer Science division at Argonne National Laboratory and professor in the Department of Statistics at the University of Chicago and coauthor of the article.
It includes estimates of electricity demand, prices and generation. It involves dealing with different regions with different needs, connected by diverse networks. The increasing reliance on renewables also means that controlling grid stability is more difficult. It requires considering production from consumers’ homes from solar or wind turbines, as well as energy credits that may be called in. Moreover, issues of privacy may prevent full identification of resources.
“All these factors make energy forecasting complex and challenging,” Anitescu said.
Arguably, statistics cannot predict with certainty when a power outage will occur. Statistics can, however, help energy companies better estimate the probability of outages – how often they will occur over a specific time period. Statistics can also help address the problem of shortages: it can identify optimal research storage levels to prevent power cuts without being too costly.
Anitescu and his coauthor Amanda Lenzi, a former postdoctoral appointee in Argonne’s MCS division and now at the University of Edinburgh, consider statistical analysis a powerful tool for energy forecasting. While statistics may not prevent all power shortages, they emphasize that it can certainly help energy companies be more likely to “keep the lights on.”
For more information about energy prediction and the role of statistics, see the full article by Amanda Lenzi and Mihai Anitescu, “How statistics can keep the lights on,” Significance 18(6) 28-30, December 2022, https://rss.onlinelibrary.wiley.com/doi/full/10.1111/1740-9713.01704