Including passengers’ response to crowding in the Dutch national train passenger assignment model: Fid-Bau Portal

Including passengers’ response to crowding in the Dutch national train passenger assignment model

Pel, Adam J. / Bel, Nick H. / Pieters, Marits

Highlights • We review public transit users’ perception and response to crowding. • 4 crowding models are proposed and implemented in the Dutch National Model System. • Boarding penalties capture responses to seat availability. • Time-multipliers show large responses due to anticipating crowding further along the line. • Combining boarding penalties and time-multipliers yields consistent crowding effects.

Abstract Transit passengers’ response to crowded conditions has been studied empirically, yet is limitedly included in transport models currently used in the design of policy and infrastructure investments. This has consequences for the practical applicability of these models in studies on, for instance, timetabling, train capacity management strategies, project appraisal, and passenger satisfaction. Here we propose four methods to include the effect of crowding, based on existing studies on passengers’ perception and response as well as often-used crowding indicators. These four alternative methods are implemented in the train passenger assignment procedure of the Dutch national transport model, and evaluated with respect to their impacts on the model results for the Dutch railway network. The four methods relate to four different ways in which an additive trip penalty and/or time-multiplier can be incorporated in the train utility function for different travel purposes, to capture the disutility of crowding as measured by the load factor. The analyses of the test case favor the hybrid method using both a boarding penalty (capturing seat availability upon boarding) and a time-multiplier (capturing physical comfort and safety throughout the trip). This method produces consistent results, while the additional computational effort that it imposes is acceptable. Further empirical underpinning is needed to conclusively show which of these methods best captures passengers’ response behavior quantitatively (for different travel purposes and conditions).