Genetic Algorithm Optimization of Charging Infrastructure Locations for the Operation of Battery-Electric Trains in German Regional Passenger Rail.

Kurzfassung

Responding to global climate change and political aims at the European level, the German government aims to achieve carbon neutrality by 2045. Embedded in this aim lies a necessity for converting diesel multiple units to electric multiple units, especially on regional rail passenger routes where diesel multiple units are often used. Making great use of existing electrification and longer layover times, battery electric multiple units are particularly well suited for these routes. However, identifying the best the spatial allocation of charging infrastructure to allow for their operation constitutes a complex non-linear optimization problem. This thesis develops a standardized methodological concept based in a genetic algorithm to optimize this spatial distribution with the least infrastructure cost using vehicle circulation plans, simulated vehicle power at wheel, OpenStreetMap data, a digital terrain model, a timetable dataset, and route geometries. The approach is developed and tested on a circulation plan for the German subnetwork of Pfalznetz. The methodology successfully identifies optimal spatial distributions of charging infrastructure but suffers from shortcomings relating to the quality and availability of input data, the computational efficiency of the algorithm and the approaching of local minima. Specific minor alterations to the code may significantly improve the quality of the results. The flexibility and broad applicability of the method primes it for expansion accounting for a wider range of aspects relating to charging infrastructure placement.