Realistic Flow Analysis Using a Simple Network Model

Simple topological data structures consisting of links and nodes have often been used for modeling flows through a network because of their processing efficiency and ease of implementation. These representations of networks are typically assumed to be flat and metrical; links have the same cost of travel in either direction and no costs are associated with traversing nodes. This paper describes three additions to the common model that retain its advantages while offering a capability for much more realistic analysis of network flows. The first enhancement of the simple network model is the assignment of separate costs for traversing a link in either direction. This simple modification allows consideration of physical constraints such as slope and temporal contraints such as traffic. Second, costs are associated with each link-to-link turn. This allows consideration of the delays experienced at network nodes from congestion and controls, and permits impossible turns (e.g. from an overpass) to be removed from paths of flow. Finally, it is shown through examples that integration of an enhanced network model with a relational database management system simplifies the assignment of appropriate travel costs enough that a single network can support realistic analyses that consider the unique characteristics of various types of links and nodes and flows across them in various congestion and control scenarios