ABSTRACT
Ground-Penetrating Radar (GPR) investigations of railway track bedsare becoming more important nowadays in civil engineering. The manufacturing of representative full-scale scenarios in the laboratory environment for the creation of databases can be a critical issue. It is difficult to reproduce and monitor the effect of differing physical and performance parameters in the ballast layer as well as to evaluate the combination of these factors in more complex scenarios. In addition, reproducing full-scale tests of railway ballast implies to handle huge amounts of aggregates. To this effect, the use of the Finite-Difference Time-Domain (FDTD) simulation of the ground-penetrating radar signal can represent a powerful tool for creating, extending or validating databases difficult to build up and to monitor at the real scale of investigation. Nevertheless, a realistic three-dimensional simulation of a railway structure requires huge computational efforts. This work focuses on performing simulation of the ground-penetrating radar signal within a railway trackbed by using a two-dimensional cross-section model of the ballast layer, generated by a Random Sequential Adsorption (RSA) paradigm. Attention was paid on the geometric reconstruction of the ballast system as well as on the content of voids between the aggregate particles, which complied with the real-world conditions of compaction for this material. The resulting synthetic GPR signal was subsequently compared with the real signal collected within a realistic trackbed scenario of ballast aggregates recreated in the laboratory environment.
