A naturalistic study into the flow of alighting and boarding passengers of commuter trains

• We study flow rates in a naturalistic environment using analysis of video material taken using a drone. • We highlight how flow rates change under different conditions. • We highlight how naturalistic flow rates differ from laboratory experiments. • Boarding flow rates are more affect by luggage and bicycles. • Alighting flow rates are more affected by passenger volumes. The alighting and boarding times of commuter trains play a major role for the length of dwell times and are directly influenced by the flow rate of both alighting and boarding passengers, measured as the number of passengers passing through a door per second. It is, therefore, important to have a good understanding of passenger flow rates to better understand and schedule dwell times. Where previous studies into passenger flow rates have mostly taken place in mock-up settings, this study focuses on the flow rate of alighting and boarding passengers in a real-world setting. Using video footage of more than one thousand alighting and boarding processes we find that the flow rate of alighting passengers increases with an increase in the volume of passengers but only up to a saturation point, whilst the boarding flow rate remains stable. The dominant flow of passengers is found to have a statistically significant effect on the flow rates of alighting and, especially, boarding passengers. These effects are found across different passenger volumes. Both alighting and boarding flow rates are affected when passengers carry luggage, whereas the presence of bicycles did not have a statistically significant effect on alighting flow rates. The findings presented in this paper suggest that different flow rates should be used when modelling dwell times depending on the number of passengers, that a saturation point for the alighting flow rates exists, and that boarding flow rates are likely more affected by other factors such as interior bottlenecks than passenger volumes.