PSI - Issue 84

Ilaria Catapano et al. / Procedia Structural Integrity 84 (2026) 119–126

124

Fig. 3. ERT results referred to a pile foundation (CS2): measurement configuration - left; resistivity map for scenario (a) - central, resistivity map for scenario (b) - right.

3.3. Forward modeling of GPR data and Microwave tomographic imaging Full 3D GPR data were simulated considering a simplified scenario made of a single pile foundation having a square cross-section with a side of 1 m. The pile is made of concrete (with a relative dielectric permittivity of 6, typical of unsaturated concrete) and is embedded in a soil whose relative permittivity is equal to 9 (unsaturated and slightly moist sedimentary soils). Concrete and soil have an electrical conductivity of 1 mS/m (i.e., 1000 Ω∙m ). GPR data were simulated by considering surface (S) and borehole (BH) measurement configurations. Both configurations operate in reflection multi-monostatic mode (i.e., one transmitting and one receiving antenna, separated by a negligible spatial offset relative to the probing wavelength, are jointly moved in M evenly spaced measurement points). The primary source is an x -oriented Hertzian dipole working at the central frequency of 300 MHz, moved along a straight line with a spatial offset of 0.08 m. For each source position, the backscattered signal is measured by assuming a 0.04 m spatial offset between the transmitting and receiving points. Fig. 4 and Fig. 5 show sketches of the two reference scenarios. For the S measurement configuration, data were gathered along the y -direction at M = 45 measurement points at a distance of 1 m from the pile along the x -direction. For the BH measurement configuration, data were gathered along the z -direction at M = 38 measurement points at x m = 2.0 m and y m = 0.5 m. GPR data have been simulated for several extents of the pile into the soil. Fig. 4 and Fig. 5 also show the raw radargram referred to the S and BH measurement configuration when the pile depth is 2 m (center panels) and 3 m (right panels). These figures confirm that the pile depth can be easily estimated from the raw radargram in the case of BH data. Indeed, although the radargram is characterized by several signal contributions, a hyperbola branch having an apex at the end of the pile appears as its unique fingerprint. Conversely, in the S configuration (Fig. 4), although the raw radargrams corresponding to different pile depths look different, mainly after about 40 ns, it is hard to distinguish the signal contribution characteristic of the end of the pile. To improve the GPR imaging capabilities in this latter case, a well-assessed microwave tomographic approach facing the imaging as a linear inverse scattering problem under the Born Approximation (Catapano et al., 2019) has been applied.