Overview of the model structure
The ABM is a 4-step model that is disaggregated in three ways: spatially, temporally, and in terms of the trip chain (see Snethlage 2023b).
The spatial resolution is not defined by zones but by so-called locations. Locations represent individual buildings, building blocks, settlements, or even city districts. The resolution depends on the requirements and the data situation. Not all data need to be known at the finest resolution. Thus, different resolutions can be selected for different activities. It is also often possible to disaggregate structural data at rougher resolutions to finer resolutions using building floor plans.
The model is based on a synthetic population, i.e. persons with individual characteristics and a "schedule skeleton". Such a skeleton is a detailed daily schedule that lacks the data to be computed by the model: Destinations and modes as well as start times.
The interlinked destination, mode, and start time choice is based on the schedule skeleton. The sequence of the three choice levels can be freely selected. A typical sequence is time – destination – mode, which is useful for many situations. However, there are also activities, such as shopping, where a different sequence may be useful, e.g. mode - time - destination.
If required, the calculation of a choice level can be fully or partially deactivated, similar to how the start time choice is taken as given in traditional aggregated demand models. If the Use existing choices option is selected, only the choices that are still open are simulated.
Note: If the Use existing choices option is selected, any inconsistencies are not checked or corrected. If, for example, the demand segment of the trip is "car" but that of an existing trip is "public transport", this inconsistent state is not adjusted. In this case, the demand segment of the tour would be decisive for the mode choice of further trips. |
The utilities are based, among other things, on time-differentiated shortest path searches between the locations. The temporal resolution usually consists of 5-8 time slices.