Fatigue is one of the major distresses occurring in asphalt concrete pavement. Repeated traffic loading causes escalated structural damage which results in the formation of cracks. There is a strain level below which the Hot Mix Asphalt goes under fatigue failure, and it is called endurance limit. In this study, a predictive model is developed to predict endurance limit strain values by using artificial neural network. Uniaxial tension-compression fatigue test results conducted under NCHRP Project 9–44 A were utilized in the model development process. An equation is also extracted from the model along which gives the exact values as the model. The coefficient of determination (R2) for the ANN predicted strain value and laboratory-measured strain value is 0.96. The model performed better than the previous in predicting fatigue endurance limit strain value. A separate Monte Carlo model is developed to highlight the impact of the variability of the individual parameters on the tensile strain predictions. The Monte Carlo Analysis based on 1,000 simulations revealed that predictive tensile strain models can lead to underestimation of output as they do not account for the variability of input parameters.
Acharjee et al. (Thu,) studied this question.