ORNL machine learning helps predict COVID-19 impact on fuel demand

Oak Ridge National Laboratory (ORNL) researchers have developed a machine learning model that could help predict the impact that pandemics such as COVID-19 have on fuel demand in the United States.

Called the Pandemic Oil Demand Analysis, or PODA, this model compared mobility patterns before and during the COVID-19 pandemic, analyzing historical weekly motor travel trends and projecting future usage.

“We developed this machine learning-based model by studying trip activities and corresponding fuel usage,” ORNL’s Shiqi (Shawn) Ou said. “The PODA analysis can serve as a useful tool to understand the impact of travel quarantine on fuel demand.”

In a Nature Energy study sponsored by Aramco Research Center, researchers assessing the time period from mid-May until August determined that average fuel demand is likely to remain below non-pandemic levels and is not likely to return to those levels before October 2020. Under a "pessimistic scenario," in which social distancing and other preventive policies are relaxed, the projected fuel demand in late September 2020 would be about 78% of the demand projected for a non-pandemic scenario.

The researchers also created a hypothetical scenario in which reopening is postponed by four weeks (that is, the reopening is not implemented until late May and early June 2020). The delayed reopening scenario would have a negative impact on fuel demand temporarily, but demand would likely recover to normal levels more quickly than if reopening was initiated sooner (late April and early May 2020).

These types of PODA projections could help inform economic and energy planning, both for the current pandemic and for similar situations that could arise in the future.

Read more: https://www.nature.com/articles/s41560-020-0662-1

Read the study: https://www.nature.com/articles/s41560-020-0662-1#MOESM1