Development of a Mechanistic Prediction Model and Test Protocol for the Permanent Deformation of Asphalt Concrete

Rutting (permanent deformation) is one of the major distresses in asphalt pavements. Rutting is caused by permanent deformation and flow around the wheel path that occurs at the surface of the asphalt concrete pavement. Rutting can cause serviceability issues such as hydroplaning and water spray.;A simple but accurate permanent deformation model is needed that can simulate various loading conditions, such as load time, deviatoric stress, and temperature. Two mixtures are used for model development in this study. The FHWA ALF (Federal Highway Administration Accelerated Loading Facility) Control mix is a coarse dense-graded mixture with PG 70-22 binder and nominal maximum size of aggregate (NMSA) of 12.5 mm. The NY9.5B mix is a dense-graded mixture with polymer-modified PG 64-22 binder with a NMSA of 9.5 mm. Triaxial repeated load permanent deformation (TRLPD) test results provide a constant slope in a log N-log &egr;vp plot; accordingly, all the strain curves translate to construct one representative curve, which is called the permanent strain mastercurve. This phenomenon is supported by two superposition principles, i.e., time-temperature superposition (t-TS) and time-stress superposition (t-SS). The logarithmic horizontal distance between the mastercurve and the strain curve determines the reduced load time shift factor or the deviatoric stress shift factor. The proposed shift model is composed of one mastercurve and two shift functions.;A composite loading test is introduced to reduce the number of required tests. The state variable, i.e., permanent strain, serves to link one loading block of the composite loading test to an entire TRLPD test; thus, one composite loading test can substitute for several TRLPD tests. The composite loading test is employed within the overall test protocol, which is composed of triaxial stress sweep (TSS) tests at three different temperatures (TH, T I, TL, reflecting high, intermediate, and low temperatures) and a reference test (i.e., a TRLPD test). The proposed test protocol method can be completed within 8.8 hours, so two or three days are needed to complete them, depending on the number of replicates. The proposed model and test protocol are verified using composite loading tests and random loading tests.;The shift model is implemented in the layered viscoelastic continuum damage (LVECD) program to predict the rut depth of asphalt pavement. The rut depth measurements taken at the FHWA ALF test sections and the National Center for Asphalt Technology (NCAT) test track are evaluated using the model. The model can successfully evaluate rut depth within a few millimeters, which proves the capability of the model implemented in the LVECD program. The model slightly over-predicts the rut depths for the NCAT test track. Wandering and aging associated with the NCAT track make it relatively resistant to rutting; thus, the measurements are lower than the predicted ones. Predictions made from the University of California, Davis (UC-Davis) full-scale tests (which use a PG64-28PM mix) show good agreement with the measurements, even though shear flow-related deformation is observed.;These simulation results support the hypothesis that TSS tests with confinement can represent behavior in the field. In this regard, excessive shear flow may be the reason for the under-prediction of the FHWA ALF mixtures. For better predictions, a correction factor (i.e., a transfer function) is suggested, which seems to be quantified via the ratio of shear stress to shear resistance. After applying an individual transfer function, the permanent deformation model in the LVECD evaluates the growth of the rut depth. Therefore, even though the shift model is a uniaxial model, the model can predict the rut depth of asphalt concrete by employing the transfer function.