CargoNetSim: A hybrid agent-based and system dynamics framework for cost–energy–emissions optimization in multimodal freight transport

Freight transportation contributes approximately 40% of transport-sector CO2 emissions globally, yet existing simulation tools fail to capture how operational dynamics in multimodal networks affect energy consumption and emissions. This paper introduces CargoNetSim, an open-source framework integrating agent-based modeling (ABM) with system dynamics (SD) to jointly optimize and evaluate energy consumption, carbon emissions, and costs across rail, maritime, road, and terminal systems. Validated, modular sub-simulators are coupled with a distributed SD layer that models congestion-driven throughput degradation, delay propagation, and their cascading energy and emissions effects across network nodes. A generalized cost engine incorporating energy, emissions, delay, and monetary components pre-screens feasible routes before running a detailed simulation. Applied to transcontinental container transport from Madrid, Spain, to multiple U.S. destinations, results show that static models underestimate total costs by up to 3×, driven by terminal dwell times, customs delays, and modal constraints that compound energy consumption through congested nodes. The SD congestion feedback endogenously amplifies energy penalties downstream. Sensitivity analysis reveals volume-dependent modal energy efficiency: rail becomes cost- and energy-competitive with trucking only beyond consolidation thresholds (8–14 containers depending on time valuation). These findings demonstrate that static models are inadequate for energy-aware freight planning and that dynamic hybrid simulation is essential for evaluating decarbonization pathways, modal shift strategies, and carbon pricing policies under realistic operational constraints.