Why System Requirements Must Influence the Selection of Causal Models to Address Congestion in Freight Logistics

Abstract

There is urgent need to identify goods which may be diverted to an alternative mode of freight movement due to the disruptions to priorities and decisions of logistics entities by Covid-19, container shortages, labor shortages, etc. Here diversion refers to the shift of goods in the logistics network that are typically moved via one freight mode - such as rail, highway, or air - to another mode. There has been significant effort to model this behavior for purposes of infrastructure planning by both national and regional transportation agencies. In the United States this has culminated in a gold standard behavioral/agent-based model supported by the Federal Highway Administration (FHWA) and implemented by state transportation agencies. This model is well established and demonstrated to successfully predict the diversion of goods from one mode to another in response to changes in the freight network. However, this model is limited to consideration of the requirements state or national transportation agencies value. As a result, the model requires large teams, years of work, and millions of dollars in funding. This scale is not acceptable for modeling of transient events such as container shortages which would dissipate before model completion. This work advances a novel approach to model this behavior that is sensitive to the requirements of a diverse set of stakeholders and conditions.