Network analysis of delay propagation on Swedish railways
(2021)Department of Automatic Control
 Abstract
 Travel on railway in Sweden has increased steadily over the past three decades and there are today twice as many passengers travelling by train as there were 30 years ago. The increasing awareness of environmental issues with other methods of transportation is likely to favour railway travel, and so the number of passengers is expected to continue to rise. As the number of passengers increase, so do the requirements on keeping trains on time. Delayed trains do not only cost money for train operators, but may also affect how likely we are to choose to travel by train. Understanding where and why delay occurs as well as how this delay might spread is therefore important not only from an economic point of view but from an environmental one as... (More)
 Travel on railway in Sweden has increased steadily over the past three decades and there are today twice as many passengers travelling by train as there were 30 years ago. The increasing awareness of environmental issues with other methods of transportation is likely to favour railway travel, and so the number of passengers is expected to continue to rise. As the number of passengers increase, so do the requirements on keeping trains on time. Delayed trains do not only cost money for train operators, but may also affect how likely we are to choose to travel by train. Understanding where and why delay occurs as well as how this delay might spread is therefore important not only from an economic point of view but from an environmental one as well.
This thesis shows that behaviour of delay occurrence and propagation of delay in the Swedish railway network may be reproduced using an epidemic SusceptibleInfectedSusceptible (SIS) model with satisfying results. By optimizing the probability of a train carrying infection (delay) from an infected station to a susceptible one, the simulation can reproduce the level of delay over time as well as the geographic distribution of delay, thus capturing global as well as local delay behavior. The thesis further shows the necessity of heterogeneous delay propagation probabilities on edges in the network in order to reproduce realworld behavior.
Furthermore, the results indicate that rescaling the nodal selfinfection rate (spontaneous delay rate) improves the model and is needed when changing the number of delayed departures needed to make an infected state, i.e. marking a station as delayed. Simulations indicate that this nodal selfinfection cannot be expressed using a linear function but varies nonlinearly with the number of delayed departures. One explanation for this would be the possible dependence between selfinfection rates on trains, which could be explained by the fact that some external factors giving rise to spontaneous delay might affect entire stations rather than individual departures. However, the effect of rescaling the selfinfection rate could also be obtained by increasing the recovery rate, why further analysis is needed in order to determine which parameter should be modified to what extent.
Lastly, results also indicate that the model may be used for estimating the impact on certain delay preventive measures by manually changing parameters for chosen stations or railway lines. This is valuable as it may give stakeholders a way of prioritizing projects and resources. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/studentpapers/record/9058326
 author
 Landelius, Jacob and Wallgren, Elsa
 supervisor

 Giacomo Como ^{LU}
 Emma Tegling ^{LU}
 CarlWilliam Palmqvist ^{LU}
 Anders Rantzer ^{LU}
 organization
 year
 2021
 type
 H3  Professional qualifications (4 Years  )
 subject
 report number
 TFRT6137
 other publication id
 02805316
 language
 English
 id
 9058326
 date added to LUP
 20210715 14:49:18
 date last changed
 20210715 14:49:18
@misc{9058326, abstract = {{Travel on railway in Sweden has increased steadily over the past three decades and there are today twice as many passengers travelling by train as there were 30 years ago. The increasing awareness of environmental issues with other methods of transportation is likely to favour railway travel, and so the number of passengers is expected to continue to rise. As the number of passengers increase, so do the requirements on keeping trains on time. Delayed trains do not only cost money for train operators, but may also affect how likely we are to choose to travel by train. Understanding where and why delay occurs as well as how this delay might spread is therefore important not only from an economic point of view but from an environmental one as well. This thesis shows that behaviour of delay occurrence and propagation of delay in the Swedish railway network may be reproduced using an epidemic SusceptibleInfectedSusceptible (SIS) model with satisfying results. By optimizing the probability of a train carrying infection (delay) from an infected station to a susceptible one, the simulation can reproduce the level of delay over time as well as the geographic distribution of delay, thus capturing global as well as local delay behavior. The thesis further shows the necessity of heterogeneous delay propagation probabilities on edges in the network in order to reproduce realworld behavior. Furthermore, the results indicate that rescaling the nodal selfinfection rate (spontaneous delay rate) improves the model and is needed when changing the number of delayed departures needed to make an infected state, i.e. marking a station as delayed. Simulations indicate that this nodal selfinfection cannot be expressed using a linear function but varies nonlinearly with the number of delayed departures. One explanation for this would be the possible dependence between selfinfection rates on trains, which could be explained by the fact that some external factors giving rise to spontaneous delay might affect entire stations rather than individual departures. However, the effect of rescaling the selfinfection rate could also be obtained by increasing the recovery rate, why further analysis is needed in order to determine which parameter should be modified to what extent. Lastly, results also indicate that the model may be used for estimating the impact on certain delay preventive measures by manually changing parameters for chosen stations or railway lines. This is valuable as it may give stakeholders a way of prioritizing projects and resources.}}, author = {{Landelius, Jacob and Wallgren, Elsa}}, language = {{eng}}, note = {{Student Paper}}, title = {{Network analysis of delay propagation on Swedish railways}}, year = {{2021}}, }