Projects per year
This thesis is composed of three papers each dealing with different aspects in optimization of intermodal transportation and a summary introducing the perceived issues within intermodal transportation and placing the three papers into context. The summary starts by introducing the congestion and environmental problems seen in transportation in Europe and why the European Union sees the reestablishment of the rail sector in an intermodal setting as the solution to the problems. The summary continues by illustrating some of the measures and initiatives that are taken to improve intermodal transportation. The summary presents the concepts behind developing a freight route planner similar to route planners seen in public transit and discusses how that could be beneficial to the transportation sector as a whole while presenting some of the barriers that may be expected in case of implementation. The summary continues into discussing more advanced methods for planning intermodal transportation with scheduling of transportation services as the main focus and gives pointers to the three papers. The first paper present a mathematical programming model to determine optimal timetables for public transit systems with respect to passenger transfer waiting time. By adopting a value of time cost the model opts to minimize the total sum of the transfer waiting time cost for transferring passengers between public transit routes. The model is solved using a Tabu search heuristic on a large scale network instance taken from the public transit system of the greater Copenhagen area. The results show that there is a potential benefit for passengers in applying models for transfer optimization. The savings in transfer waiting time for passengers may account for as much as 4 million € a year expressed in value of time cost. The second paper presents a mathematical programming model to determine intermodal freight train schedules in a European setting. By the latter is meant that trains are assumed to run on an infrastructure divided into train paths as is becoming common practice in European railways. Furthermore the model includes terminal operations on an aggregated level to capture the transfer costs at terminals. Finally, the model introduces a value of time cost for freight. The level of the value of time determines the trade-off between operational cost of trains and the total transit time, i.e. a low value of time cost means low operational cost and high transit times, while a high value of time means higher operational costs and lower transit times. The model shows that it can be used as a decision support system for intermodal train carriers to determine their trade-off between customer service (in form of transit time) and operational cost. The model is solved using XpressMP’s mixed-integer programming solver which not surprisingly due to the complexity of the model did not prove to be an efficient solution method. The third paper presents a Tabu-search based algorithmic framework to solve a modified version of the fixed-charged capacitated multi-commodity network design model (CMND). The modification is derived from the model presented in the second paper where vehicle balance constraints are added in nodes. The constraints add a restriction on the design arcs requiring that the number of open arcs entering a node must be equal to the number of arcs leaving a node. These constraints are dubbed design balance constraints and are added as a new set of constraints to the CMND model resulting in a model denoted the design balanced capacitated multi-commodity network design model (DBCMND). The new set of design balance constraints prevents the use of existing solution methods developed for the CMND model and thus requires a new algorithmic framework. The Tabu search framework presented in the paper offers a solution method to solve the DBCMND model. The algorithm is tested on previously used network design instances in the literature and the computational results are compared to results achieved using Xpress-MP’s mixed integer programming solver. The results show that the algorithm produces good solutions to the DBCMND model and that it is applicable to large-scale instances. The algorithmic framework thus creates a starting point to create solution methods network design models with design balance constraints that can be applied to scheduling problems in transportation.
|Place of Publication||Kgs. Lyngby|
|Publisher||Technical University of Denmark|
|Number of pages||158|
|Publication status||Published - Sep 2005|