Transportation planning considers questions of the long-term impact of infrastructure or policy changes - such as an additional subway line or an increase of gasoline costs. Typical questions are: Where does traffic/congestion shift to? Will trips be dropped or added? Might people change where they live or work as a longterm response to changes in the transportation system?
Traditional models for transportation planning typically put most of their emphasis on car traffic. Other modes are often considered via modal split calculations only - that is, once transportation demand has been calculated, some of the demand may decide to take other modes instead of the car, usually via a discrete choice model based on travel times and different values of time on the different modes.
The reason for this is that static assignment models do not have a realistic representation of the dynamics. For example, time-dependent effects of car traffic, such as queue build-up and spill-back during the rush-hour, cannot be represented. Similarly, more complicated effects in other modes, such as congestion of a pedestrian facility, or bus-car interaction, cannot be modeled.
The approach is nevertheless justified as long as the other modes quantitatively do not play a large role, and it may be justified under certain circumstances if car mode is the only congested mode. In many metropolitan regions, this is not true: other modes besides cars are contributing more than 25% of all trips, and these other modes are heavily congested.
As a result of these and other thoughts, there is currently a push towards microscopic (or agent-based) transportation planning models. In these models, all entities including the travelers themselves are individually resolved. This makes is in principle straightforward to include arbitrary elements of reality. The downside is that it may be hard to implement, it may use considerable computer resources, and data for calibration of these effects may be hard to obtain. Nevertheless, microscopic transportation simulation models are feasible, and several groups are working on them.
This paper will first give an introduction into how such microscopic transportation simulation packages are designed. This is followed by a section about a possible implementation of multiple modes into such a package. The paper is concluded by a discussion and a summary.