Search

An urban logistics system is composed of multiple agents, e.g., shippers, carriers, and distribution centers, etc., and multi-modal networks. The structure of Physical Internet (PI) transportation network is different from current logistics practices, and simulation can effectively model a series of PI-approach scenarios. In addition to the baseline model, three more scenarios are enacted based on different characteristics: shared trucks, shared hubs, and shared flows with other less-than-truckload shipments passing through the urban area. Five performance measures, i.e., truck distance per container, mean truck time per container, lead time, CO2 emissions, and transport mean fill rate, are included in the proposed procedures using real data in an urban logistics case. The results show that PI enables a significant improvement of urban transportation efficiency and sustainability. Specifically, truck time per container reduces 26 percent from that of the Private Direct scenario. A 42 percent reduction of CO2 emissions is made from the current logistics practice. The fill rate of truckload is increased by almost 33 percent, whereas the relevant longer distance per container and the lead time has been increased by an acceptable range. Next, the dissertation applies an auction mechanism in the PI network. Within the auction-based transportation planning approach, a model is developed to match the requests and the transport services in transport marketplaces and maximize the carriers’ revenue. In such transportation planning under the protocol of PI, it is a critical system design problem for decision makers to understand how various parameters through interactions affect this multi-agent system. This study provides a comprehensive three-layer structure model, i.e. agent-based simulation, auction mechanism, and optimization via simulation. In term of simulation, a multi-agent model simulates a complex PI transportation network in the context of sharing economy. Then, an auction mechanism structure is developed to demonstrate a transport selection scheme. With regard of an optimization via simulation approach and sensitivity analysis, it has been provided with insights on effects of combination of decision variables (i.e. truck number and truck capacity) and parameters settings, where results can be drawn by using a case study in an urban freight transportation network. In the end, conclusions and discussions of the studies have been summarized. Additionally, some relevant areas are required for further elaborate research, e.g., operational research on airport gate assignment problems and the simulation modelling of air cargo transportation networks. Due to the complexity of integration with models, I relegate those for future independent research.