Punctuality modelling in the railway: comparative analysis of network simulation and statistical approaches on a Norwegian railway line
In this study, models have been developed to predict punctuality for railway sections. Two models have been developed and tested on the Dovre Line (Dovrebanen). Simplifications have been made, and only passenger trains run the entire section between Trondheim and Oslo. The first model that has been...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/428228 |
| Acceso en línea: | https://hdl.handle.net/2117/428228 |
| Access Level: | acceso abierto |
| Palabra clave: | Railroads Ferrocarrils Àrees temàtiques de la UPC::Enginyeria civil::Infraestructures i modelització dels transports::Infraestructures i transport ferroviari |
| Sumario: | In this study, models have been developed to predict punctuality for railway sections. Two models have been developed and tested on the Dovre Line (Dovrebanen). Simplifications have been made, and only passenger trains run the entire section between Trondheim and Oslo. The first model that has been developed is a network model where train performance for a day is simulated. The model is programmed in Python. The simulation is based on distances and speed profiles between stations where trains can cross. Rules have been added for how crossings can be changed in the event of delays. The model is calibrated so that in a normal situation, the simulation will ensure that trains follow the timetable. To analyse the consequences of errors in the infrastructure and other conditions that can affect train performance, it is possible to analyse the effects of different scenarios such as errors that stop traffic, errors that result in temporary slow running and conditions that mean that slow running has been introduced over longer periods of time. When the model is run for such a scenario, or a combination of scenarios, one can see the effect this will have on train performance and the model calculates the associated delays. Delays will often propagate throughout the day, so that it is also possible to look at consequential delays. With such a model, it is possible to assess the consequences of errors and defects in the infrastructure, which can then be used to prioritize maintenance and renewal. A statistical model has also been developed. This model is based on minute delays registered in Bane NOR’s punctuality database. In this database, a reason code is also specified behind each delay. The most common delays are linked to reason codes that represent infrastructure problems, weather conditions and cascading effects from other delayed trains. By comparing both approaches, we see that the simulation model is able to reproduce and explain the observed delays. The analysis of specific scenarios can also identify limitations in the current timetable, including tight timetables and limited possibilities to get back on track after errors have been corrected. |
|---|