Transportation Modeling and Planning Software

Modeling Approaches

The general structure associated with transportation modeling was developed in the 1960s and has remained mainly unchanged despite major technological advances in the last decades. The approach concerns a sequence of four sub-models: trip generation, trip distribution, mode choice, and traffic assignment over a network.

Because travel decisions are not actually taken in this kind of sequence, several contemporary approaches on transportation modeling consider the simultaneous processes of trip generation, trip distribution, and mode choice. Other recent approaches focus on the activities of travelers and the choices they entail, aiming at a clearer understanding of travel behavior.

Model Taxonomy

Transportation models are classified in various categories in order to determine their appropriate use in the planning process and to identify their properties and differences. Their classification can thus be based on various dimensions.

A deterministic model does not include any randomness. Thus, an event can either occur or not occur. Repeated applications of a deterministic model will produce the same results. However, if certain attributes or inputs of the model are subject to any randomness (or not known with certainty), the model is defined as stochastic. In such models, different random number sequences will produce different results, indicating a probability of certain outcomes.

A model is defined as time-based when time advances from one point in time to the next based on constant steps. A model is event based when time progresses based on the occurring events, thus skipping certain points in time.

Off-line models are mainly used for evaluating what-if scenarios or testing various alternatives, while online models open up the possibility of continuous data input. If these data are real-time data, then real-time models can be developed that progress according to the rate at which time actually advances.

Real-time models are mainly used for managing and controlling traffic at real-time level (for example, traffic signal controllers) and are often attached to a physical traffic control device. Because computational requirements are considerably lower than those of real-time models, off-line models can simulate the behavior of a system at a rate often faster than real time.

Analytical models deal with the prediction of the performance of a system based on the validation, calibration, and statistical analysis of field data. If the modeled system has a logical structure and is not governed by discontinuities, an analytical solution can be obtained.

When analytical models are not applicable, simulation models are used. Such models use mathematical and numerical techniques for creating an inventory of possible events and assessing their impact for one or more variables of the model. However, they require a rather significant amount of input data, which in turn need to be validated and calibrated before used.

...