| Popis: |
In this report we outline a conceptual demonstrator for advanced real time network management for freight rail traffic. The focus is on the coordination between traffic control, train drivers and yard management, three essential parts in the real time management of a rail freight network. The intention is that the demonstrator can support multiple purposes, such as education and training, demonstrating research advancements, and enabling feedback between practitioners, system developers and researchers. The proposed demonstrator has a focus on the interaction between different systems and between humans using these systems, but also on the rail freight system perspective by the inclusion of the connection between the line and the yard. We present a generic architecture and propose existing components that could be combined to such a demonstrator. Thus, even though the demonstrator may seem complex and visionary, the existence of these components makes the realization of the demonstrator realistic. The development roadmap for the demonstrator proposes both a step-wise implementation plan of the complete demonstrator, as well as several partial packages that provide useful sub-demonstrators by themselves. The appendices of the report include contributions to the continued development of two of the components that are part of the demonstrator. Firstly, in order to also better understand the type of situations that yard managers need to handle in operations and what implications these have on the traffic on the line, a Swedish case study has been conducted and the results are presented in Appendix A. More specifically, the case study analyses the factors that influence the departure time deviation for freight trains and how these can be used for predicting the actual departure time. These predictions can be used in a decision support system for yard planning at larger marshalling yards. A conclusion is that no single factor can fully explain the departure time deviation, but many different factors contribute to it, like destination, time of day, train load, number of wagons on the yard, connection time for wagons, and connection time for locomotives. Secondly, to support the traffic controllers and dispatchers with an advanced decision support tool for deviation handling, a selection of different functionalities and algorithms may be required. In Appendix B, two different approaches for disturbance management are presented. Approach 1 (ALG1) is a heuristic, parallel algorithm, while the second approach (ALG2) is an exact algorithm based on state-of-the-art commercial optimization software. In order to classify and evaluate alternative algorithms for train re-scheduling and disturbance management, an assessment framework is also proposed in Appendix B. Based on this framework, the overall strengths and shortcomings of the two mentioned train rescheduling algorithms are assessed while applied on a set of 30 simulated disturbance scenarios of various complexity. The results show that typically, ALG2 obtained good rescheduling solutions for all 30 disturbances, but compared to ALG1, ALG2 is slow in obtaining solutions.ALG1 is good at quickly finding solutions with less passenger delays while it is less effective when it is used to solve disturbances associated with an infrastructure failure. The strength of ALG2 is its ability to reschedule the traffic during infrastructure failures. A detailed presentation of the evaluation is found in Appendix B. H2020 Grant Agreement 826206Deliverable D3.2 of project Fr8Rail II |
