Modellering av tågnät på Arriva DK

Sammanfattning: Purpose: The purpose of this thesis is to develop two simulation models of Arrivas train network in Denmark to facilitate the process of timetabling. One of the models assumes deterministic times in the other model these times are stochastic. The purpose of the deterministic model is to evaluate if timetables are free from conflicts, a process that currently involves a lot of manual labor. The main goal of the stochastic model is to be able to compare different timetables, using a more realistic model of the train network that compensates for the inherent uncertainties in travel and stop times. The goal is not to develop a finished product for immediate practical implementation. The aim has been to develop a platform and process for modelling and serve as first version for further improvement. Delimitations: The models focus on normal operations. Therefore unexpected events like signal errors, accidents etc. are disregarded. Another delimitation is that the availability of trains and staff, and how it affects the timetable, are disregarded. Additional delimitations are that the model only accounts for Arriva trains and not for their competitors’ trains, and that the decision process for prioritizing trains when conflicts occur is simplified in the simulation models. v Method: The method used in this thesis is a modified version of the 8-step process for large simulation projects in Hillier and Lieberman (2010, pp. 959-963). The process has been implemented in two stages, one for each model. Conclusions: Verification and validation show that the deterministic model is consistent with reality. The model is therefore considered ready to be used as an aid in the process of making new timetables. The stochastic model has been more difficult to validate. This is because the material plan and service plan is not yet included in the simulation model, which makes it hard to compare the simulation results to the observed data for the real system. However, validity has been achieved for the logic of the system, and for the distributions of transport times and number of passengers. We can also conclude that the distributions for the time spent on stations needs to be further investigated and validated. There is also an uncertainty about the rules of prioritization of trains in case there are delays. During the validation of the model, the dataset has shown lack of quality in large parts of the total set. Therefore we recommend Arriva to investigate their data collection. After the quality of the data is improved, we recommend Arriva to implement the service plan and material plan into the simulation model to be able make a fair comparison between observed data and simulation results.