Validation Procedure for Predictive Functions of Driver Behaviour on Two-Lane Rural Roads

Abstract

The study presented here aims to validate some operating speed prediction models calibrated on two-lane rural roads by using speed data collected in Northern and Southern Italy. Operating speed is defined as the speed at which drivers of passenger cars travel on a dry road in free flow conditions during daylight hours and it is calculated using a specific percentile of speed distribution, typically the 85th. Speed measurements were carried out by using laser detectors in connection with previous environmental and traffic conditions. The study is addressed to emphasize the reliability and easy application of one predictive speed model working both on tangent segments and on circular curves. The calibration phase involved roads in the Northern Italy, while the validation phase involved roads in the Southern Italy. Three models were validated applying them on eight two-lane rural roads falling within the road network of the Province of Salerno with features that reflect those adopted in the calibration phase; the selected models to be validated present the simplest analytical structure for type and number of explanatory variables and for the performance diagram shape of the operating speed values. The validation procedure was to estimate some synthetic statistical parameters as mean absolute deviation, mean squared error and coefficient of variation. The results allow in a simple way to trace continuous operating speed profiles on two-lane rural roads and to carry out safety analyses on the horizontal alignment.