Traffic Simulation

Transportation planning and management is a very complex field that needs to consider a large number of interacting phenomena concurrently: land use and transportation infrastructure, and transportation demand and supply are only two such examples. The timescale of transportation interventions can range from the short term (for example, minutes or hours in the case of real-time incident response) to long term (for example, years or decades in the case of forecasting the traffic of a planned freeway or rail connection). Considering the size of most transportation problems, which can range from a single intersection or corridor to a city, a metropolitan area, or a region, spatial interactions also need to be considered. It becomes apparent that these interactions cannot be dealt with easily without the help of specialized tools. Furthermore, objective and factual evidence needs to be provided in order to accompany and support related investment and policy decisions. In light of the high costs of developing transportation infrastructure, accurate forecasts of their expected impacts can play a significant role in properly allocating resources.

The field of traffic simulation has emerged over the past few decades as a response to these needs. A large number of traffic simulation models and tools have been developed during this time, ranging from simplified analytical representations of individual phenomena to integrated simulation-based models that aim to capture all aspects of traffic and traveler behavior. In order to capture these various interactions and satisfy these numerous requirements, traffic simulation tools comprise a large number of models, each of which typically has several parameters that govern its behavior.

While traffic simulation has been increasingly used in all aspects of transportation planning and operations, traffic simulation models have also received criticism. George Box famously said (decades ago), “Essentially, all models are wrong, but some are useful.” While this famous aphorism might seem like a negative statement, in essence it provides valuable insight into the benefits and limitations of traffic simulation.

Traffic simulation is a valuable tool that allows transportation modelers to quantify the impact of different strategies and policies in an unprecedented level of detail. Of course, there are significant caveats in the development of traffic simulation models. If the basic requirements (that is, reasonable data, reliable calibration, and validation, etc.) are not performed, there are [Page 1404]considerable dangers of creating unrealistic models. Current traffic simulation models offer very powerful graphical user interfaces (often offering three-dimensional photorealistic rendering) that make challenging these results very hard.

A successful simulation project comprises three main components: (1) a traffic simulation model, (2) adequate suitable data, and (3) a proper calibration and validation of the model, prior to its application, to ensure that its output matches the real world.

In order to meet the broad needs for traffic simulation, a wide range of traffic simulation models with different properties and characteristics have been developed over the last several decades. Traffic simulation models can be classified according to multiple dimensions, such as level of granularity or fidelity with which they attempt to model traffic. Choosing the appropriate model is a fundamental decision for the success of any simulation project. Often, modelers tend to know a tool well and use it for all tasks, irrespective of its suitability, but, ideally, one should aim to use the right (level of) tool for each task.

...