Integrated and agent-based charging demand prediction considering cost aware and adaptive charging behavior

With the projected growth of electric vehicles to meet net-zero emission targets, the accurate prediction of future charging demand is essential for optimal infrastructure planning. This study delivers an integrated and scalable agent-based modeling framework for future spatiotemporal estimation, which simultaneously captures heterogeneous cost aware charging behaviors, daily activity patterns, and route and mode choices. Meanwhile, the framework employs a stochastic, adaptive smart charging module that incorporates diverse charger types and dynamic ToU electric tariffs, enabling users to probabilistically shift charging decisions to minimize costs and mitigate range anxiety. The framework was applied in a case study of Gothenburg, Sweden, under near-future scenarios with 50 % EV penetration. Results indicate that introducing charger-type prices with residential ToU tariffs shifts charging toward home, and probabilistic ToU-aware deferral reduces the residential peak by upto 20 % relative to the cost aware but immediate charging scenario.