Ground penetrating radar bridge deck investigations using computational modeling

Infrastructure in the United States is failing. According to a 2005 study by the American Society of Civil Engineers, over a quarter of the bridges are structurally deficient or functionally obsolete. Condition assessment without the assistance of subsurface sensing techniques leads to poor detection and quantification of damage because much of the damage and precursors to damage is hidden beneath the surface. Ground Penetrating Radar (GPR) a popular choice for bridge deck assessment, depends on a subjective process, which is the trained eye of a technician. The ability to simulate a GPR investigation provides insight into the response from bridge deck elements, as well as the interaction among the elements and changes due to the presence of an anomaly and supports defect detection.;A subsurface modeling tool is developed with physical modeling components available for general applications but extended to meet specific requirements for geometric modeling of civil infrastructure. The simulation component implements the 2-dimensional Finite Difference Time Domain (FDTD) method for electromagnetic modeling. Comparisons between 2D and 3D simulations show that, for bridge deck analysis, 2D modeling is adequate for condition assessment.;A model-based assessment augments the conventional approach to analysis by using iterative computational models to reconstruct the bridge deck in a healthy condition. To identify areas of suspect condition, the response from the computed healthy deck can be compared to the response collected in the field.;The effect of the presence of rebars on the scattering from an anomaly can be significant, and is not easily removed from GPR data. In the computational model, the strong scattering rebars are replaced with an excitation source that results in wave propagation equivalent to the scattering from the rebar. This technique makes the GPR bridge deck problem better suited to the traditional inversion algorithms that are often complicated by strong scatterers.;Through experimentation, the GPR antenna can be characterized to determine a virtual sensor for the 2D FDTD model. The resulting sensor allows for a significantly smaller geometry, which saves time and computational resources while reducing differences in propagation associated with using a 2-dimensional instead of 3-dimensional model.