Fatigue-Transfer Functions: How Do They Compare?

Abstract

The primary objective of this work was to investigate which of several fatigue-transfer functions predict the failure of mixtures and expected fatigue life most accurately. A second objective was to study existing fatigue models and how they relate to mixture volumetrics. This work was done by employing layered elastic analysis to obtain stresses and strains in the pavement under monthly temperature variations. The calculated tensile strains were then used in the fatigue-transfer functions to predict the fatigue life of the pavement. The mix and distress data were obtained from the WesTrack project. In general, all fatigue-model predictions were poor to fair; even the locally calibrated models (WesTrack) yielded at best 57% accuracy in predicting failing or surviving mixtures. The Strategic Highway Research Program (SHRP) model predicted failures with 72% accuracy; the Asphalt Institute model, 39%; Medani and Molenaar's model, 31%; and the Shell models, 28% accuracy. The SHRP and Asphalt Institute models used suggested field shift factors (FSFs). The fatigue models studied were originally developed for predicting crack initiation in the laboratory. In using these models to predict fatigue life in actual pavements, crack propagation is usually addressed through an empirically determined FSF. An attempt was made to calibrate and recalibrate the fatigue-transfer models using the WesTrack data. This analysis indicates that there is a trade-off in predicting failing or surviving mixtures correctly and the accuracy of the predicting fatigue life compared with the applied traffic loading. Because the nonfailed mixtures were trafficked with only 5 million traffic repetitions, the error associated with the fatigue-life predictions was unknown. Therefore, the overall improvement of the model accuracy through this calibration process cannot be assessed. For several reasons, this study recommends using the Asphalt Institute fatigue-transfer equation in layered elastic analysis to evaluate the fatigue-cracking potential of asphalt mixtures.