We propose and analyze a generic multi-class kinematic wave traffic flow model: Fastlane. The model takes into account heterogeneity among driver-vehicle units with respect to speed and space occupancy: long vehicles with large headways (e.g. trucks) take more space than short vehicles with short headways (e.g. passenger cars). Moreover, and this is what makes the model unique, this effect is larger when the traffic volume is higher. This state dependent space occupancy is reflected in dynamic passenger car equivalent values. The resulting model is shown to satisfy important requirements such as providing a unique solution and being anisotropic. Simulations are applied to compare Fastlane to other multiclass models. Furthermore, we show that the characteristic velocity depends on the truck share, which is one of the main consequences of our modeling approach.
FLM van Wageningen-Kessels et al 3
INTRODUCTIONTraffic flow theory aims to describe human driving behaviour on road networks, including consistent explanations of observed phenomena such as stop and go traffic, capacity drop and traffic hysteresis. Traffic flow models are traditionally classified into microscopic models, which describe the behavior of individual vehicles, mesoscopic models, which describe traffic on the basis of probability distributions of 'packets' of vehicles, and macroscopic models, which describe traffic as a continuum flow (1, 2, 3). Our focus is on macroscopic traffic flow models, which are widely used to describe and predict traffic flows in larger networks, both in the context of traffic and transportation planning, as well as in management of traffic operations.Macroscopic models describe aggregate driving behavior and typically include an average (equilibrium) relation between traffic density ρ (number of vehicles per unit length) and flow q (number of vehicles per unit time). In kinematic wave (KW) models, traffic is assumed to always be in a state described by this fundamental relation. However, observed density-flow plots usually show wide scatter. One reason for this scatter is that not all the data represent steady-state conditions. Higher-order models explain and reproduce (at least partly) this scatter by assuming accordingly that the traffic state tends towards the fundamental relation but is usually not on it, due to for example anticipation and relaxation effects.A more complete explanation for the scatter in density-flow plots is that it is also related to heterogeneity among drivers and vehicles. Ossen and Hoogendoorn (4) discuss heterogeneity in relation to microscopic models and distinguish between intra-and interdriver heterogeneity. The first relates to changes in behavior of a single driver over time. Additionally, inter-driver heterogeneity relates to structural differences in behavior and/or capabilities between vehicles and drivers. For example, trucks are usually longer and slower than cars, and have different drive characteristics (e.g. maximum acceleration and deceleration capabilities). As a result...