Reconstruction algorithms to improve nondestructive evaluation of reinforced concrete

Abstract

Two reconstruction algorithms have been developed to improve nondestructive evaluation of reinforced concrete structures. Both algorithms implement computationally demanding strategies to estimate the properties of potentially complex systems. These algorithms are: (1) a two-dimensional magnetic algorithm to identify the locations and sizes of reinforcing bars in concrete, and (2) an algorithm to estimate the material properties and geometry of a concrete member by comparing the experimental and simulated dynamic responses to a surface impact.The magnetic algorithm uses magnetostatic relationships to relate measured distortions in the magnetic field outside a concrete member to the distribution of steel within the member. An optimal solution for the locations and sizes of the reinforcing bars is determined by minimizing the difference between measured and computed distortions in the magnetic field. In simulated problems, where the concrete was assigned the magnetic properties of air, the algorithm accurately identified the location and size of a single bar, even in the presence of induced random errors. The algorithm was also able to identify two bars simultaneously, even when they were closely spaced. When realistic magnetic properties were used for modeling the concrete, it was found that the concrete had a significant effect on the measured magnetic distortions. A convenient method was found to compensate for the magnetic distortion caused by the concrete.The dynamic response algorithm consists of the following steps: (1) recording an impact and the dynamic response of a concrete member at multiple locations, (2) comparing these responses with those obtained by simulating the experiment with a finite-difference model and (3) varying the properties of the model until the difference between the measured and simulated responses is minimized. Applying the algorithm to experimental measurements of dynamic response demonstrated the algorithm's ability to identify the thickness and Young's modulus of a slab. When applied to simulated responses of a slab containing a partial depth, open, surface crack, the algorithm also demonstrated the potential to estimate crack depth and location.