Passenger flows and dwell times for commuter trains in Stockholm

This study investigates how the volume and flow of passengers affect dwell times, using
data from commuter trains in Stockholm. We show that dwell times often take longer than
scheduled and that most station stops involve rather limited numbers of passengers.
Improving the performance at these stops with small volumes can thus have a sizeable
impact on the overall on-time performance. We find a quadratic relationship between the
number of passengers and the likelihood of dwell times taking longer than scheduled. Dwell
times are shortest when the flow of passengers is entirely in one direction, boarding or
alighting, for both lower and higher passenger volumes. If both boarding and alighting flows
are present, it is best that they are as balanced as possible in the case of lower passenger
volumes. The influence of this proportion is less prominent when passenger volumes are
higher. Scheduled dwell times should thus be based both on the expected volume of
passengers and on the expected proportion of boarding and alighting passengers.