https://www.mdu.se/

Electric traction is the most efficient traction system in the railway transportation. However, due to the expensive infrastructure and high power demand from the grid, the share of electric trains in the railway transportation is still lower than other trains. Two of the possible solutions to increase the share of electric trains are: optimal train operation to minimize energy consumption, the use of batteries as the energy source for driving electric trains on non-electrified lines. This thesis aims to extend the knowledge in the field of energy optimal operation of electric trains and battery-driven electric trains.

Energy optimal operation of electric trains is supervised using a driver advisory system (DAS), which instructs the driver to operate the train in an energy-efficient manner. This thesis contributes to DAS technology under two topics: the increase of energy efficiency and the design of DAS.

This thesis presents a complete procedure of designing a DAS from the mathematical formulation to application on the train. The designed DAS is in the form of an Android application and is based on a dynamic programming approach. The computational performance of the approach is enhanced using heuristic state reducing rules based on the physical constraints of the system. The application of the DAS shows a potential reduction of 28% in energy consumption.

This thesis considers the detailed energy losses in the whole propulsion system using a regression model that is generated from validated physical models. The application of the regression model instead of a previous constant efficiency factor model results in 2.3% reduction in energy consumption of the optimum speed profiles.

Based on the solution for the normal electric trains, a solution is also offered for the optimal operation of battery-driven electric trains, in which the characteristics of the battery as one of the main components are considered using an electrical model. The solution presented in this thesis, is to combine the popular single mass point train model with an electrical circuit battery model.

Furthermore, this thesis evaluates the performance of the optimization approaches and validates the models against the measurements from actual drives of a real-life battery train. The results show a potential of around 30% reduction in the charge consumption of the battery.

The results of this thesis (algorithms and the Android application) are provided as open source for further research in the field of energy efficient train control.