An investigation into the railway ballast grading using GPR and image analysis

This study reports on an investigation into the grain size distribution of the railway ballast using ground-penetrating radar (GPR) and image analysis. The proposed approach relies on the hypothesis that the dimension (grading) of the ballast aggregates can influence the back-reflected spectrum received by the use of GPR. This assumption was confirmed by the finite difference time-domain (FDTD) simulations of the GPR signal, which were run by using the numerical simulator package gprMax 2D. A regression model was developed which related the "equivalent" diameter of the ballast aggregates and the frequency of the peak within the received spectrum. The model was validated in the laboratory environment by means of a 155 cm x 155 cm x 50 cm methacrylate tank, filled up with railway ballast. An air-coupled GPR system equipped with a 2000 MHz central frequency antenna was used for testing purposes. A total of three spatial distributions of the ballast aggregates within the tank were investigated, by emptying out and filling up thrice the tank with the same material. The geometric information on the ballast grading obtained from the simulation-based regression model was compared to the actual grading curve of the ballast. To this effect, an algorithm based on the automatic image analysis was developed. The comparison showed that the modelled aggregate diameter corresponded to the 70 % of the grading of the material sieved out in the laboratory. This contribution paves the way to new methodologies for the non-destructive assessment and the monitoring of segregation phenomena within the railway ballast layers in railway track-beds.