This paper follows on the work of Daganzo [1,2], who used a cell transmission model for traffic to create evacuation plans. In this model, a cell corresponds to a segment of a street and the street cells are connected into a network. An extended version of the model accommodates varying cell sizes and multiple-lane streets, and also accounts for some areas carrying higher risks.

The drawback of Daganzo’s work is that it is computationally intensive. This paper introduces a heuristic procedure that allows the model to be used for larger-scale evacuation scenarios. The effectiveness of this heuristic is shown in a number of real-world scenarios using the city of Duisburg as an example.

The extended cell transition evacuation model, which incorporates multiple cell sizes, road lanes, and traffic flow limits, is described in detail (27 sets of equations are required to do this). The paper then describes two heuristics used to obtain computationally tractable solutions, one based on shortest path and an iterative one that minimizes risk to the evacuees. The heuristics are used to solve a static version of the extended model for which the flow connections are fixed. These solutions are then fed into the original extended model.

The nine scenarios in the paper range in size from network lengths of 23 to 83 kilometers, from 8,750 to 25,856 vehicles, and from 107 to 339 cells. The shortest path heuristic is disappointing in all but one case, while the iterative heuristic produces good results in all cases except one where the memory requirements proved excessive. The paper contains detailed tables describing the results.