Developing a Large-Scale Multi-Modal Modeling and Optimization Framework for Freight Transport Network Analysis

Abstract

Global freight transportation is confronted with unprecedented challenges from heightened logistical complexity, operational uncertainty, and the pressing necessity to minimize environmental footprints. Current simulation models tend not to represent very well the dynamic interaction between micro-level agent behavior and macro-level system dynamics of multi-modal freight networks and therefore fall short to holistically optimize energy consumption, emissions, costs, and delays. This dissertation presents CargoNetSim, an open-source simulation software that combines agent-based modeling (ABM) and system dynamics (SD) to simulate the movement of containers through multi-modal freight transport networks. CargoNetSim is composed of independent modules—NeTrainSim for rail, ShipNetSim for shipping, INTEGRATION for trucking, and TerminalSim for terminal operations—all orchestrated by a central integration hub to simulate multi-modal transport dynamics. A cost optimization module is used to identify efficient routes by calculating energy consumption, emissions, and operational costs, thereby improving computational efficiency. The framework is validated and its capabilities are illustrated through extensive validation and field applications. For rail transportation, the model captures freight train dynamics for six powertrain technologies—diesel, biodiesel, their hybrid versions, electric, and hydrogen fuel cell—predicting energy consumption to within 4.5% of empirical data and CO$_2$ emissions. Comparative analysis reveals that electric trains conserve energy by 56% over diesel trains and reach zero emissions based on renewable fuels. Biodiesel-hybrid and diesel-hybrid engines cut emissions by 21% and 16%, respectively. Utilizing a multi-objective optimization method and the A* algorithm, one realizes the potential for 47% savings in hybrid rail powertrains and 26% using hydrogen and electricity-based solutions and only 7% travel time increases but fulfilling energy efficiency in addition to speed demands. For maritime transport, the framework comprises hydrodynamic simulation with high fidelity and cybersecurity threat modeling, such as GPS spoofing, to analyze operational resilience. A test case of an S175 container vessel from Savannah, USA, to Algeciras, Spain, accurately predicted fuel consumption within 13.1% of operational records, with consistent performance (1.5% variability) under multiple environmental scenarios. The terminal and trucking components capture the road transport dynamics and intermodal transfer inefficiencies to represent real-world congestion and customs delays. A comprehensive case study from Madrid, Spain, to Kansas City, Missouri, USA, validated the integrated framework, finding optimal routes and revealing that simulated costs were up to 1.98 times higher than initial estimates, largely due to terminal delays and rail disruptions. Sensitivity analyses in additional U.S. cities—Chicago, Dallas, and Los Angeles—confirmed the strength of the model emphasizing the value of dynamic simulation in representing the intricacy of operations. By bridging the divide between single agent and system-scale dynamics, CargoNetSim offers a powerful decision-support tool for policymakers and logistics planners. Its ability to quantify trade-offs between economic efficiency and environmental sustainability addresses international decarbonization ambitions, such as those regulated by the International Maritime Organization. CargoNetSim is open-source to guarantee availability and flexibility to foster the collaborative development of sustainable freight transport practices and facilitate the evolution towards cleaner and more efficient global logistics and commerce.