Cost-Efficient Robust Network Design for BEBs: Tackling Energy Uncertainty with Limited Data

Abstract

This research focuses on electrifying an existing bus network under energy consumption uncertainty between stops, with limited data. The study evaluates the optimal locations and types of charging stations, as well as battery sizes, to minimize electrification costs. Three optimization models are explored: nominal (deterministic model considering expected values for energy consumption), robust optimization with uncertainty budget (BoU), and distributionally robust chance constraint (DRCC), which utilizes observed energy consumption data. Regarding the optimal design, the BoU model opts for more flash-feeding stations to handle greater uncertainties, while DRCC tends to minimize the number of charging stations overall. The performance of the models are compared based on their electrification costs as well as conceived battery longevity in terms of charge-discharge cycle, finding that larger battery capacities in robust models (BoU and DRCC) extend battery life compared to the nominal model. Compared to BoU model, the DRCC achieves comparable improvements in battery life at a lower cost (for similar battery capacity). For the observed energy consumption in this study, nearly 40 data points are found to be sufficient for robust network design which is feasible for 90% of observed energy consumption data. In conclusion, the DRCC model is particularly efficient in designing robust and less conservative network design.