Backcalculation of Thickness and Dielectric Properties of Asphalt Concrete Using Numerical Simulation of Ground Penetrating Radar

Abstract

Ground penetrating radar (GPR) is an essential tool for the non-destructive evaluation of pavement conditions. The accuracy of GPR analysis depends on the correct prediction of the layer dielectric property. In this study, a finite-difference time domain simulation-based two-step backcalculation algorithm was developed to determine the layer thickness and dielectric constant from air-coupled GPR data. The two-step backcalculation algorithm was developed by combining a surrogate-based global optimizer and artificial neural network for efficiency and accuracy. The GPR backcalculated results were compared with the field conditions of three pavement sections. The results show that the backcalculation algorithm could predict the layer thicknesses with almost 95% to 99.7% accuracy, whereas the prediction accuracy of the dielectric constants was between 90% and 98.5%. Furthermore, it was observed that the variations of dielectric constants in each lift of the asphalt concrete layer could be used to predict the presence of probable distress in the field.