Dwell time delays for commuter trains : An analysis of the influence of passengers on dwell time delays

Sammanfattning: The thesis presented here delves into why dwell time delays for commuter trains occur, with a specific focus on the impact of boarding and alighting passengers. The overarching aim is to develop knowledge of how time delays arise to identify and describe potential ways in which dwell time delays can be reduced. In addition to this, a secondary aim is to identify how dwell times can be studied on a network-wide level. Six research papers are included in this thesis, which all contribute to the aforementioned aims. The first paper presents a literature review on the influence of passengers on dwell times. The five subsequent papers present different data analyses on the impact of passengers on dwell times and make use of several years worth of automatic passenger count data collected on board commuter trains in Stockholm and the region of Skåne in Southern Sweden. The findings from these studies indicate that although the volume of passengers is often stated as the main cause for dwell time delays, this is not necessarily the case. The results, instead, suggest that the volume of passengers acts as an accelerator for the negative impact of other aspects of the dwelling process such as the behaviour of passengers. With regards to studying dwell time delays it is important to make use of robust measures and to presentdwell time delays in terms of frequency and size rather than just an average value. In addition to this, the value of having data on a level of seconds rather than minutes is highlighted. The latter is important since a majority ofdwell time delays are smaller than one minute. Several avenues to reduce the risk of dwell time delays are proposed, based on the findings from the includedstudies. The first avenue to be explored is that of adopting a more dynamic approach to dwell time scheduling. In practice, this means that, in contrast to what is common practice in Sweden, different dwell times should be usedduring peak and off-peak hours and between different stations. In addition to this, it is important to account for the behaviour of passengers during the boarding and alighting process. This can be done by making use of platformmanagement measures. The third avenue that is identified states that dwell time scheduling should take on a more network-wide approach rather than treating stations as a single entity. This is important since there are interdependencies between stations that influence the behaviour of passengers, such as the way passengers spread out. Working on these points will help to reduce the risk of dwell time delays. Although this thesis has an emphasis on the Swedish context, given the origin of the data, the above-mentioned avenues are likely to be applicable in different geographical settings as well.