Base for calculating the dynamic assignment

The principle of dynamic assignment is illustrated in a flow chart (Flow diagram dynamic assignment).

The Quick Start-contains the most important steps and settings required for dynamic assignment, after you have modeled the Vissimnetwork (Quick start guide Dynamic assignment).

The dynamic assignment is calculated in Vissim based on the iterated simulation. Thereby the modeled road network is simulated not only once but repetitively. The drivers choose thereby their paths through the network based on their experiences from the preceding simulations. The modeling of this "learning process" consists of the following subtasks:

• Vissim takes into consideration in searching for the paths from sources to destinations, that not every driver chooses the best path. Some drivers will use less attractive paths. That means that not only the best paths must be known for each origin-destination relation but also a set of paths. Ideally, the number of the k best paths would be used. Since there is no efficient method for direct calculation of this quantity, which would be useful for traffic assignment, the following approach is used in Vissim:

In each iteration of the simulation, the best paths are calculated respectively. More than one optimal path would be found, because traffic conditions change from iteration to iteration. In the course of the iterations the number of different paths and the archive of known paths, from which the drivers can choose, increases (Path search and path selection).

• For the paths found an evaluation must be calculated, on which the drivers base their choice. In Vissim, the so-called generalized costs are computed for the paths. These comprise travel time, travel distance and other costs (for example, tolls) are calculated. Travel distance and costs are defined directly in the network model whereas the travel time can be determined only with the help of the simulation. Therefore travel times will be measured in individual parts of the network during a simulation run in Vissim, and can be considered for the path selection in subsequent simulations.
• The selection of a path from a set of possible paths is a special case of the discrete decision problem (discrete choice problem). From the set of paths and their generalized costs, the percentage of drivers that choose the path is calculated. The mathematical function which is by far the most commonly used to represent this type of selection is the Logit function. Also Vissim uses a variant of the Logit model for the path selection (Method of path selection with or without path search).

The road network is modeled in Vissim in great detail to provide a reproduction of the traffic flow as precisely as possible, in high temporal and spatial resolution. However, this detailed modeling is not necessary for any of the three subtasks listed above. For example, the decision which path through a city is chosen does not depend on which lanes the vehicles travel, or how the intersections on the path look like exactly. For the assignment it is enough to reference an abstract description of the road network, where the intersections are nodes and the links between the intersections are edges of an abstract network graph. On this abstract network graph, the assignment procedures can work much more efficiently. The abstract network graph also correlates with the human understanding: For example, to describe a path to someone, it is sufficient to mention the sequence of intersections and to add if he must turn there; a detailed description is not necessary.

In Vissim an abstract network is built for dynamic assignment. To do so, in the detailed model, highlight the parts you want to use as abstract nodes by drawing in network objects of the type node. For dynamic assignment with microsimulation, these will normally be network sections that correspond to real intersections (Building an Abstract Network Graph).

The simulation is iterated until there are no more significant changes from one iteration to the next in the congestions and travel times in the network. This situation is called in Vissim as convergence. You can set the criteria for convergence (Attributes for achieving convergence).