Development of pavement deterioration models by combining experimental and field data.

摘要

Pavement deterioration models are an important input to pavement management systems and pavement design procedures. In pavement management systems, they are used to determine the cost-efficient maintenance schedules and activities. In pavement design procedures, they assist to achieve the most economical pavement structure design to satisfy given requirements.; Rutting refers to permanent depressions along the wheel paths and is considered one of the most important pavement distresses. This distress causes ponding of water, which is a safety concern, and increases road roughness, which is related to user comfort. In the past, mechanistic-empirical models and empirical models developed for rutting only achieved limited success. In this dissertation, a model was developed to predict rutting using joint estimation techniques with experimental and field data.; In this dissertation, the experimental data came from the AASHO road test while the field data was obtained from the MnRoad test site. Experimental data and field data have their own strengths and weaknesses. By combining both data sets, joint estimation provides the following benefits: higher efficiency, identification of the effects of attributes unobserved in the field data, and correction of the prediction biases in the experimental data. This dissertation is one of the very first efforts to systematically achieve bias correction in pavement deterioration models by using rigorous statistical tools instead of shift factors that are conventionally used.; The unobserved heterogeneity among different pavement sections in the data is accounted for by using random-effects approaches in the estimation process. In addition, by predicting rut depth increments for each time periods, the models are particularly useful for pavement management systems.