Free flow speed and drivers behavior in Italian arterial roads

Free flow speed is a very used transportation parameter but there are many ways to define it. In this paper "space mean speed" is used to characterize free flow conditions. The study overall objectives are, first, to determinate the time intervals that define isolated vehicle and then to develop models to estimate free-flow speed. The methodology is based on geometric features and space mean speed data collected in urban roads tangents located in Cagliari, Italy. The data were analyzed with statistical tools. It was used a multiple linear regression that allowed to obtain different models. On these models, a diagnostic analysis and residual tests was performed to assess adequacy of models (model verification phase). At last, a model validation phase was developed. The first study result shows that a vehicle is isolated when the vehicle ahead is at a time distance greater than or equal to six seconds and the following vehicle is at a time distance greater than or equal to four seconds. Based on this assumption, isolated vehicles have been identified through a survey. Three statistical models are constructed, and all these have passed the diagnostic analysis showing to be adequate to interpret the phenomenon. Nevertheless, only two are validated to predict free flow speed. Variables influencing free-flow speed are lanes width, type of median, tangent length, type of left-lateral obstacle, lateral clearance on the right side, access points and presence of crosswalk. The paper shows that even if all statistical verifications are good, it is important to validate models with a sample not used in the model construction. In fact, validation phase can ensure model replicability in different contexts.