Use of Toll Transaction Data for Travel Time Prediction on National Highways Under Mixed Traffic Conditions

Knowledge of accurate travel time is an important aspect for both users and traffic engineers/planners. Hence, the travel time models should predict the travel time more accurately by considering different traffic characteristics. Limited studies are carried out for travel time distribution prediction for different vehicle classes plying on National Highway (NHs), under mixed traffic conditions. Hence, an attempt is madeh in the present study (a) to evaluate the travel time and (b) to develop the vehicle classwise travel time prediction model. For this, a study stretch of 38 km of National Highway (NH)-48 between Karjan Toll Plaza and Narmada Toll Plaza located in the western part of India is considered. Travel time of different vehicle classes obtained from toll transaction data for seven days continuous count has been used for analysis. It is observed that there is a wide variation in the travel time of different vehicle classes and within the same vehicle class. Further, distribution analysis was carried out for the best-fitted distribution, and it was found that the Generalized Extreme Value (GEV) distribution explains the travel time variation well. Using the base of GEV distribution, the GEV parameters, i.e., location, scale, and shape parameter equations are developed based on traffic volume and traffic composition. The results showed that the developed equations predict well for travel time.