LAUSR

Reliability is understood in public transport as the certainty travellers have regarding the level of service they will experience when travelling. The travel time, waiting time, or the comfort level they will experience inside the vehicle are some of the most important reliability attributes. Reliability is usually neglected from travel behavioural models, since there is a lack of studies addressing its impact in travellers’ choices. This study proposes an in-depth analysis and characterization of travel time reliability in a public transport system using passive data only, which every day is getting more popular worldwide. The study is comprised of two parts, taking Transantiago (the public transport system of Santiago, Chile) as a study case, as it is a good representation of a frequency-based public transport system. In the first part a statistical and graphical analysis of travel times is conducted, focusing on characterizing headway and travel time reliability of public transport services as well as the effect of dedicated infrastructure, based on smartcard transactions and GPS information. In the second part an aggregate mode choice model based on revealed data is developed to analyse the effect of travel time reliability on travellers’ preferences. Overall, this study provides evidence of significant differences among headway variability and in-vehicle travel time dispersion different public transport modes. The standard deviation (a measure of dispersion) can be quite high for bus trips, while in the case of metro is smaller than 4 min independent of trip length. Regarding the aggregate public transport mode choice model, the coefficient of variation of headways is a key attribute to explain modal preferences. It was found that average bus users would accept traveling ~ 5 min longer to completely avoid headway irregularity. All these results were obtained by analysing passive-data from AFC and AVL systems only, which represents a novel approach for choice modelling. Transport modellers should consider the impact of service reliability to improve route choice and passenger assignment models, and better represent users’ perceptions and behaviour.