Strategic passenger-oriented timetable design : Long-term timetable designs with minimised passenger inconvenience

Sammanfattning: Timetable development and design is a complex process that is crucial for safe and efficient railway operations. The combination of steel wheels and steel rails makes it possible to create trains and to transport many vehicles, thus passengers and freight, at the same time, but it also results in longer braking distances. These braking distances often exceed sight distance, which means that sufficient distance between trains must be maintained. This requires a thorough planning of train movements in order to prevent conflicting train paths and trains stopping for red signals. This is done by creating a time schedule for different train paths along the track, the so-called timetable. The timetable forms the backbone of railway operations, because a timetable informs a passenger when a train departs and arrives. However, in order to attract passengers, the timetable should be aligned with customer demand. Unfortunately, railway operation tends to deal with great demand variations over time and within the network. In order to make clear how passenger demand is distributed, the demand is often expressed in an origin-destination matrix. Each cell of the matrix corresponds to the number of expected passengers between an origin and destination. Based upon the demand distribution, a line design is created. A line design determines the route of a train, and consists of a stopping pattern and frequency per train. Although the line plan is important for the timetabling process, an optimal line plan does not automatically result in an optimal or feasible timetable. In the past, timetable design focused on a minimisation of the total travel time in conflict-free timetables only. Nevertheless, several studies confirmed the need for periodic and symmetric timetables that come with equal levels of service throughout the day, which are easily memorisable for the passenger. These timetables must be robust, so that a high punctuality can be achieved. Additionally, an ideal timetable also takes into account factors like in-vehicle time, waiting time and number of transfers, summarised in the perceived travel time (PTT). It is, however, impossible to include all these elements in a manual timetable design. This emphasises the need for a timetabling model that combines passenger demand and line design to calculate a timetable with a minimal PTT. Several different timetable models have been developed in the past, where each model has its own area of focus. Some models focus on the optimisation of line plans, so that the line design connects most important origin-destination pairs and travel time between these pairs is minimised. However, these models do not take into account specific arrival and departure times. It might thus be that the travel time will be high for passengers that have to change trains. Other models focus on the development of conflict-free timetables, in which the infrastructure governs the timetable. Although this might result in a feasible timetable, it may not always be an optimal timetable since passenger demand is often not included. The final category of timetabling models focuses on the improvement of passenger satisfaction. These models minimise waiting time or the total journey time for instance. Nevertheless, the resistance to change trains is usually high, but often not included in the calculation. In contrary to other timetabling models, the Strategic Passenger Oriented Timetabling (SPOT) model, developed by Polinder (2020) and NS, is able to create a timetable with a minimal PTT. However, the model is currently not used within the timetable development process. Therefore, this research has investigated to what extent the SPOT model can be used in this process, and hence support and speed up the design of new timetables. The SPOT model includes the resistance to change trains in the calculation of the PTT. In the model it is assumed that each minute of in-vehicle time counts as 1 passenger-minute, each minute of waiting time corresponds to 2 passenger-minutes, and each transfer is awarded with a penalty of 20 passenger-minutes. A lower PTT is thus achieved through an optimisation of waiting times and transfer penalties. It means that the model can especially be used for determining arrival and departure times at transfer nodes. Despite the fact that the model is unable to include infrastructural limitations, the results are useful for determining which transfer possibilities are important at each node. In order to validate this hypothesis, two case studies have been performed for the transfernodes Weesp and Zwolle. These cases have been selected based upon recent problems during the development of post-COVID-19 timetable scenarios for NS. For each casestudy, several elements of the current timetable and proposed scenarios have been included in the input of the model, in order to analyse the effect on the timetable at the specific node. The output of the model, consisting of the PTT, improvement potentials for origin-destination pairs and dwell-time graphs, provided a clear overview of how each experiment scored. In the end, this study concludes that the SPOT model is especially applicable for studies in which different timetable scenarios must be compared with each other. It can help to illustrate the impact of decisions and trade-offs, so that different ideas on timetable design can be assessed before making specific, conflict-free timetables. The model can thus be used in the stage of exploratory research.