Grey model for asphalt pavement performance prediction

Abstract

Most existing rutting models are either based on an empirical or mechanistic-empirical modeling approach. The mechanistic-empirical model can be used easily and effectively on the relationship between rutting and the pavement response such as vertical strain, asphalt mix properties, and ambient condition. However, the cause of rutting is sophisticated and is the function of molding variables, which results in the poor performance of many of the existing models. In addition, some data or properties for pavement evaluation relating to the environment and structure may not be very clear, and so the pavement system is grey in its nature. Due to the advent of laboratory facilities and pavement management systems, a large number of pavement databases have become available. Thus, in this paper, the prediction-modeling approach is to simplify with two components and to emphasize the phenomenon of cause and effect by the data of rut depth versus traffic loadings. The model based on grey system theory is, then, developed to predict the rut depth of the asphalt pavement performance. The algorithm of model represented GM (1, 2) is presented and the applicability of the model is the data of rut depth obtained from the wheel-tracking device test. Field test data of rutting performance is collected from FHWA research data to evaluate the accuracy and effectiveness of the new prediction model and to confirm the model's validity. A statistical linear regression analysis was conducted for testing the accuracy of prediction. The comparison between the measured and predicted rut depth is within the tolerance level of /spl plusmn/2.5 mm. It is to believe that GM (1, 2) is useful for making a prediction of rut depths.