The world is (once again) returning to some semblance of pre-pandemic life as the omicron variant wanes. Many are now wondering about the risk calculus for popular activities such as plane travel, which can often be a high-density environment. Researchers at the Argonne Leadership Computing Facility (ALCF) have been applying supercomputing power to examine how different boarding procedures affect the risk of Covid transmission on airplanes.

The researchers — hailing from University of West Florida, Florida State University, Embry-Riddle Aeronautical University, and Arizona State University — used a pedestrian dynamics model to simulate a series of boarding procedures. These included single-zone boarding (random order with no preference); six-zone, business-first boarding (business class boards first, then the remaining zones board back-to-front); back-to-front boarding (each row boards back-to-front); and back-to-front, business-first boarding (business class boards first, then the remaining rows — not zones — board back-to-front).

In each of these boarding scenarios, the researchers looked for the number of times two passengers came within six feet of each other. “By varying parameters such as passenger speed, line-distance threshold, intersection-speed coefficient, luggage stowing time, seat conflict team, and towards-seat-speed coefficient, we were able to generate more than 16,000 distinct scenarios,” explained Ashok Srinivasan, author of the paper and a computer scientist at the University of West Florida, in an interview with ALCF’s Nils Heinonen.

“In designing this work, we looked to molecular dynamics for inspiration,” Srinivasan said. “Our social dynamics models present pedestrians as a chemistry simulation would an atom. The momentum the pedestrian’s mass and velocity generate is directed toward the success of a given goal, the completion of which is obstructed by fixed surfaces (such as walls) and other atom-like pedestrians.”

To run these models, the researchers used Argonne’s Theta system, a 6.9-Linpack petaflops Intel-based Cray XC40 system with over 4,000 nodes. Theta ranked 70th on the most recent Top500 list.

The researchers found that limiting the number of seating zones actually helped ensure fewer clusters of passengers during boarding, while back-to-front boarding essentially doubled the rate of exposure. The models also showed a couple of other helpful policy changes that could be made.

“We concluded that prohibiting luggage from being stowed in the overhead bins can limit exposure,” Srinivasan said. “Additionally, to reduce the risk of transmission as much as possible, beyond implementing random boarding procedures, airlines should have passengers with window seats board prior to people assigned to aisle seats.”

To learn more about this research, read the reporting from ALCF’s Nils Heinonen here and read the paper, published in Royal Society Open Science, here.

This research is a follow-up to prior research conducted on the Frontera supercomputer at the Texas Advanced Computing Center (TACC). To learn more about that research, click here.