Development of rutting progression models by combining data from multiple sources

Pavement deterioration models are an important element in the design and management of pavement systems. They predict the development of distresses in the pavement structure as a function of pavement characteristics, loading, and environmental variables.; Rutting, a specific type of pavement distress, manifests itself through longitudinal depressions along the wheel paths of asphalt concrete pavements. Historically, this distress has been a primary criterion of structural performance and represents a serious safety issue for road users. Empirical models developed to date have a limited range of applicability and in most situations cannot predict rutting accurately. In this dissertation, structured econometric techniques are used to estimate new empirical pavement rutting models. The model specifications are non-linear in the variables and the parameters and are estimated using data from the AASHO and WesTrack Road Tests. Fixed-effects and random-effects approaches are used to account for unobserved heterogeneity of the different pavement sections.; Two different sources of rutting are modeled: rutting in the underlying layers of the pavement structure (i.e., base, subbase and subgrade) and in the asphalt concrete layer. Different parameters for load equivalency and material hardening are estimated for these two sources of rutting. The models predict rut depths by adding the predicted increments of rut depth for each period, which is particularly advantageous in a pavement management context, and they account for relevant environmental effects in both test locations. The data sources have complementary characteristics that are exploited using joint estimation. Joint estimation from both data sources allows the incorporation of the effects of layer thicknesses, mix characteristics, loading and the effects of thawing and high temperatures in a single rutting model.; The models presented in this dissertation can be used in pavement design for the determination of an economically optimum thickness of the asphalt concrete layer for rutting performance. In a pavement management system, they can be used for the determination of strategies for the original construction and rehabilitation for optimal rutting performance. In addition, the models have important implications for asphalt concrete mix design and the allocation of cost responsibilities to different vehicle classes.