An activity-based latent class modelling approach to assess the impact of hybrid working on travel demand in the Netherlands after COVID-19

Abstract

Zero emission policies in urban centers are promoting the conversion of transit agencies fleets to battery electric buses (BEBs). This transition raises questions about battery management and more specifically about the best way to mathematically model this resource in order to respect energy feasibility constraints while being as little conservative as possible. In an attempt to partially answer these questions, this work presents a two-stage stochastic model with recourse for the multiple depot electric vehicle scheduling problem with stochastic travel time and energy consumption (S-MDEVSP). Vehicles are allowed to be partially recharged and a non-linear charging function is considered. Our model takes advantage of the full information on the current state of charge that is available in operation by allowing planned charge time to be extended when energy consumption deviations are observed. We propose a column-generation-based heuristic featuring stochastic pricing problems to solve a real-life instance from the city of Montréal, Canada. An analysis of the relevance of our approach for different commercially available BEBs is also provided.