Modeling traffic demand with origin-destination matrices

An OD matrix defines the travel demand based on the number of trips between the zones (Defining an origin-destination matrix). The zones are the starting points and end points of the trips. The number of trips applies to each pair of districts for a given time interval. You define the time interval in the matrix attributes (Matrix attributes). You edit the number of trips of the OD matrix in the matrix editor (Editing OD matrices for vehicular traffic in the Matrix editor). Origin-destination matrices are also called OD matrices, demand matrices or trip matrices.

You can specify multiple origin-destination matrices for a simulation with dynamic assignment. Each origin-destination matrix can contain a different vehicle composition or apply to a different time interval. The time intervals can overlap arbitrarily because the traffic generated at any time is always the result of the total traffic from all matrices that include this point in their validity interval.

Demand values with decimal places

The demand values in origin-destination matrices can contain decimal places:

• When matrices are imported from Visum with the ANM import

• When values are entered in the matrix editor

• When the function Scale total volume to is selected in the parameters of the Dynamic assignment

At the start of the simulation run, Vissim scales the demand for each origin-destination relation and rounds it to an integer value. Only then does Vissim calculate the trips. The rounding depends on the random seed specified in the simulation parameters. Example: The demand of a origin-destination relation is 100.27 veh/h. Vissim generates 100 trips/h in 27% of the simulation runs and 101 trips/h in 73% of the simulation runs.

Demand values with decimal places do not affect the achievement of convergence in most cases, or only very slightly. The effects also depend on the Vissim network: If paths have a very low volume or if a single additional vehicle considerably affects travel times on many paths, this may require more simulation runs until convergence is reached. In such cases, you can reduce the Required share of converged paths/edges in the Dynamic assignment parameters on the Convergence tab.