Journal: IVT Schriftenreihe

Abbreviation

Publisher

ETH Zurich

Journal Volumes

ISSN

Description

Filters

2010 - 20168
Reset filters

Search Results

Publications 1 - 8 of 8
  • Holistic rail network operation by integration of train automation and traffic management
    Item type: Doctoral Thesis
    Rao, Xiaolu (2015)
    IVT Schriftenreihe
    Nowadays, railways play an increasingly important role in our daily lives. Many experts and engineers are making their own efforts to optimise railways whether it involves increasing capacity, saving energy and reducing costs. However, the experts from different fields have their own intention, method and evaluation on optimisation. This thesis aims to analyse these differences systematically in order to build a more holistic optimisation method, with special focus on two specific areas: traffic management and train operation.It has been proved that traffic management and train operation work well separately, but not collaboratively. Traffic management’s strength lies in traffic conflict prevention and resolution, but it lacks sufficient on-board computing power and real-time transmitted train data, which will reduce the accuracy of traffic plan execution. On the contrary, train operation has sufficient on-board computing power and real-time train data transmission, but it has no knowledge of other trains’ status to prevent or resolve conflicts. Therefore, the optimisation strategies of traffic management and train operation are complementary.The thesis endeavours to integrate the two fields by proposing an integrated optimisation model, which highlights the importance of the bidirectional communication between traffic management and train operation. The former provides the strategy of conflict resolution for the latter, while the latter offers real-time feedback of train characteristics to the former. Moreover, the thesis introduces a decision-making structure into the integrated optimisation model by using the Analytical Hierarchy Process (AHP) method. Therefore, it can adapt to different optimisation circumstances and achieve a balance in multidimensional, at times contradicted optimisation goals. Furthermore, the thesis develops a lab-demonstrator to prove the feasibility of the integrated optimisation model.In addition, the thesis contributes to the railway capacity study. It proposes a new classification of capacity definitions, which distinguishes the dynamic capacity influenced by different optimisation methods from the theoretical capacity influenced by different infrastructure conditions. This classification can reduce the impact of confusion of capacity definitions in the discussion of capacity, especially when it involves experts from different backgrounds. Moreover, the thesis proposes a new 4-quadrant capacity model, which provides a new method to evaluate the trade-off between dynamic capacity and quality in traffic planning and operations. Furthermore, the thesis proves the feasibility of the new 4-quadrant capacity model in a practical case study.To sum up, the integrated optimisation model and the proposed 4-quadrant capacity model both indicate that the tendency of railway optimisation is not to maximise the capacity at the expense of quality, but to increase the capacity and improve the quality at the same time.
  • Algorithmic decision support for the construction of periodic railway timetables
    Item type: Doctoral Thesis
    Herrigel-Wiedersheim, Sabrina (2015)
    IVT Schriftenreihe
    This thesis addresses the problem of solving large railway timetabling problems using algorithmic methods. With the increasing demand for better and more frequent services, for higher capacity utilization and for improved reliability, railway timetabling problems become more and more complex. Algorithmic decision support is one promising way to cope with this increasing complexity, and the continuous progress of operations research methods offers a significant potential for the railway industry. The approach of this thesis concentrates on algorithms for the construction of periodic railway timetables during long- and mid-term planning. As an input, functional requirements and restrictions of the infrastructure have to be described at a model granularity suitable for the given planning stage. The algorithmic approach then creates a periodic timetable with the same model granularity and optimizes the timetable according to a given objective. Compared to a manual timetable construction, the algorithmic approach allows to compare different timetable scenarios in a shorter time. Possible modifications of the infrastructure or the service intention and their influence on the overall timetable can therefore be tested and evaluated more efficiently and also in more detail.The algorithms studied in this thesis are based on the so called Periodic Event Scheduling Problem (PESP), a mathematical model which proved to be suitable to automate the construction of periodic railway timetables already several times in research and practice. To solve the model there exist different mathematical methods. This thesis summarizes them and shows advantages of a method based on Mixed Integer Linear Programming (MILP). Compared to other solution methods it allows a direct optimization and with this several further advantages concerning a more efficient automation and a better solution quality. However there exists no practical experience of the method for larger timetabling problems and there even can be found the conjecture that the solution method is not suitable for large scale problems.The algorithms developed in this thesis are based on the so-called Periodic Event Scheduling Problem (PESP), a mathematical model which has become the standard approach for algorithmic construction of periodic railway timetables and has been used successfully already several times in research and practice. Different mathematical methods have been proposed to solve timetabling problems formulated as a PESP. This thesis summarizes the different solution approaches and shows advantages of a method based on Mixed Integer Linear Programming (MILP). Compared to other solution methods, the MILP approach allows a direct optimization, and with this several further advantages concerning a more efficient automation and a better solution quality. However, there exists no practical experience of the method for larger timetabling problems and it has been an open question so far whether and to what extent it is suitable for large-scale problems.In this thesis this gap is filled by providing the missing experience of applying the MILP solution approach for a set of increasingly large timetabling problems for parts of the Swiss railway network. To profit from the method’s advantages also for large-scale problems, an adaptation of the solution method is introduced. It allows to reduce computational complexity considerably, but still ensures good solution quality. To this end, the thesis provides three main contributions:• Implementation of the defined algorithms including an automated model construction out of real data provided by Swiss Federal Railways (SBB).• Extension of the models until a limit of computation time for the used algorithms is reached.• Development of new methods for the acceleration of the given algorithms to solve and optimize also larger models.To ensure operational feasibility for the given model granularity and a certain degree of timetable quality for our models, the resulting timetables are evaluated by the software OnTime. This way it can be ensured that the level of detail of our models is rich enough to represent realistic timetable scenarios. However, the models in this thesis are not constructed to study and evaluate a concrete timetable scenario but rather to represent different model sizes to study the computational performance of the proposed algorithms. After an evaluation of different parameters for the algorithm and the used MILP solver, the best parameter setting is used to determine a computational limit of the given solution strategy for larger model sizes. Although using strong computer servers and state-of-the-art commercial MILP solvers, this limit can be reached for our largest models.To accelerate the algorithms, two decomposition methods are proposed and studied. The first one is motivated from optimization theory. The PESP-graph corresponding to a PESP model is split into different subgraphs and used to define independent MILPs. Two iterative methods are introduced to coordinate the subproblems in order to find feasible global solutions of sufficient quality. The decomposition approach is successfully applied to a smaller test model splitting it into two subproblems. However, the generalization of the method for a larger number of subproblems or larger model sizes leads to several difficulties.
  • Methodik zur Effizienzbeurteilung der Kapazitätsnutzung und -entwicklung von Bahnnetzen
    Item type: Doctoral Thesis
    Frank, Patrick (2013)
    IVT Schriftenreihe
    Wachsende Zugzahlen und vielfältige Anforderungen der Netznutzer führen auf historisch gewachsenen Bahnnetzen zwangsläufig zu Nutzungskonkurrenz und damit zu Engpässen. Neben dem reinen Mengenwachstum und den sich wandelnden Verkehrsströmen, erfordern gesteigerte Flexibilitäts- und Pünktlichkeitsanforderungen Anpassungen in den Bereichen Betrieb, Produktion und Infrastruktur. Für eine langfristig orientierte, pro-aktive Netzentwicklungsstrategie ist das frühzeitige und zuverlässige Erkennen von Kapazitätsengpässen und Stabilitätsrisiken somit von entscheidender Bedeutung, um einen optimalen Mitteleinsatz zu erreichen. Die übergeordnete Forschungsfrage der vorliegenden Arbeit lautet daher: Wie können Kapazitätsengpässe in einem Eisenbahnnetz für mittel- bis langfrsitige Planungshorizonte beurteilt und effizient beseitigt werden?Zur Beantwortung dieser Frage werden in der vorliegenden Arbeit Methoden zur Beurteilung von Kapazitätsnutzung und –entwicklung in einem dem langfristigen Horizont gerechten Massstab entwickelt und kombiniert, insbesondere bezogen auf die Granularität und Modellierung der Infrastruktur- und Betriebs-Daten. Sie baut auf einer eingehenden Analyse der wichtigsten Einflussfaktoren von Kapazität und Stabilität auf und typisiert Netznutzungsformen anhand von sieben charakteristischen Netznutzern und deren Zusammenwirken anhand von spezifischen Nutzungsmustern.Anhand einer Analyse existierender Methoden zur Kapazitätsermittlung und Definition der Anforderungen im langfristigen Planungshorizont wird eine vereinfachte Kapazitätsermittlung für Strecken und Knoten entwickelt. Deren Vorteil liegt darin, dass trotz einer geringen Anzahl an benötigten Inputparametern eine gegenüber reinen Schätzverfahren belastbarere, quantifizierte Berechnung möglich ist, die Aussagen über zulässige Zugzahlen auf einem Streckenabschnitt oder innerhalb eines Knotenbereichs ermöglicht. Sie liefert damit einen Beitrag zur Schliessung der momentan noch immer bestehenden Lücke zwischen Expertenwissen und IT-basierten Methoden.Die zur Effizienzbeurteilung notwendige quantitative Nutzenbewertung von Kapazitätsnutzungen baut auf einem mehrstufigen Akteursmodell der folgenden vier wesentlichen Sichtweisen auf: Die Sichtweise der Eisenbahninfrastrukturunternehmung, der Eisenbahnverkehrsunternehmung im Personenverkehr, im Güterverkehr und schliesslich die Sichtweise des integrierten Konzerns. Dabei wird bewusst auf eine Monetarisierung der eisenbahnbetrieblichen Nutzenkomponenten verzichtet, um Nutzenwirkungen möglichst in ihrer originären Dimension beschreiben und vergleichen zu können. Hierzu dient ein System von drei Nutzenkomponenten: Der Kapazitätssteigerung, der Flexibilitätssteigerung und der Fahrzeitverkürzung. Diese lassen sich durch direkte und indirekte Nutzenindikatoren quantifizieren, welche in einer Kosten-Nutzwertanalyse beispielsweise den Kosten durch einen Kapazitätsausbau gegenübergestellt werden können. Zur parallelen Kostenbeurteilung einer Massnahme wird ein neu entwickeltes Modell zur Kostenermittlung von Bahninfrastrukturelementen herangezogen, das auf einem mesoskopischen Massstab die rasche Ermittlung zu erwartender Investitionskosten und Annuitäten ermöglicht.Es wird gezeigt, wie die anhand der neu entwickelten Methoden Erkenntnisse für weitere Aspekte im Bereich der Kapazitätsnutzung und –entwicklung gewonnen werden können. Dazu zählen ein Vorschlag zur Kostenumlegung des Kapazitätsverzehrs als Bestandteil eines anreizorientierten Trassenpreissystems, die systematisierte Massnahmenentwicklung im Zuge von Engpassbehebungen und ein Ansatz zur Einteilung von Netzeffekten infrastruktureller Massnahmen innerhalb integral vertakteter Systeme.Schliesslich werden die Methoden in einem generischen Planungsprozess zusammengefasst und anhand einer Case Study, dem nördlichen Teil des schweizerischen Nord-Süd-Korridors Basel – Erstfeld als Teil der Achse Rotterdam - Genua, erprobt. Für die anhand der vereinfachten Kapazitätsermittlung prognostizierten Engpässe im Jahr 2030 werden Lösungsvorschläge sowohl infrastruktureller, als auch betrieblicher Art entwickelt und bewertet. Die Funktionalität der (Gesamt-)Methodik kann grundsätzlich nachgewiesen werden und liefert damit einen Beitrag zur verbesserten Berücksichtigung eisenbahnbetrieblicher Zusammenhänge in der strategischen Infrastrukturplanung.
  • Impacts of non-Aeronautical Development at Airports on Public Transport
    Item type: Doctoral Thesis
    Orth, Hermann (2016)
    IVT Schriftenreihe
  • Failure and degradation prediction by artificial neural networks
    Item type: Doctoral Thesis
    Fink, Olga (2014)
    IVT Schriftenreihe
  • Investigation of space-time structures in public transport networks and their optimization
    Item type: Doctoral Thesis
    Alt, Bernhard (2010)
    IVT Schriftenreihe
  • Planning, operation, and the influence of national culture on public transport service reliability
    Item type: Doctoral Thesis
    Carrasco, Nelson (2014)
    IVT Schriftenreihe
  • How international borders affect local public transport
    Item type: Doctoral Thesis
    Barth, Emanuel (2014)
    IVT Schriftenreihe
Publications 1 - 8 of 8