Pore Structure Models to Predict Hydraulic Conductivity of Recycled Asphalt Pavements

Recycled asphalt pavement (RAP) is increasingly being used due to the economic savings and environmental benefits. RAP contains removed or reprocessed asphalt binder and aggregates instead of virgin material. In order to minimize the damage due to water migration in RAP, pavement design requires accurate quantification of its hydraulic properties. This paper first presents theoretical formulations that enable finding the most accurate and applicable models for RAP. Hydraulic conductivity of RAP is predicted using models containing asphalt content and associated parameters. Results of several widely used empirical and semiempirical models were also compared to identify the one that is most reliable for RAP. A systematic evaluation is presented based on root-mean-square error (RMSE) analyses of predicted and experimentally measured hydraulic conductivity values. The influence of bitumen content and grain-size distribution factors on hydraulic conductivity of RAP were evaluated by the addition of new variable combinations to account for these effects on the best widely used model. Logarithmic linear regression was used to develop new relationships between hydraulic conductivity and different variable combinations to determine the influential variables. It was determined that factors related to grain-size distribution play a prominent role in hydraulic conductivity of RAP. It was observed that the RMSE values of models that contain asphalt content were comparable to those of the Kozeny-Carman model, but lower than those of the Hazen, Slichter, Terzaghi, Beyer, Kruger and Fair-Hatch models. The latter do not contain asphalt content, showing that models that contain it gave better predictions.