GPR-based Detection of Voids and Evaluation of Grouting Under Semi-rigid Basement

The void underneath semi-rigid base is a common defect in roads. There are some difficulties in the detection and repair for this kind of hidden damage, as well as in the evaluation of the effects of grouting treatment. For the detection and maintenance of roads, it is essential to study the detection and judging for voids underneath base and the evaluation of the spread of grout. Through theoretical analysis, numerical simulation and analysis of real data, this research generated the characteristics of under-base voids of different types and dimensions on GPR images, proposed the detecting and dimension-measuring methods for under-base voids, and studied the process and effects of data analysis techniques. (1) The characteristics of under-base voids of different types (air-filled, water-filled or grout-treated) and dimensions (height and horizontal dimensions), on A-scan and B-scan GPR image respectively, were analyzed theoretically. The gprMax software which is based on the FDTD method was employed to simulate the transmission of GPR wave within the road structure, which certified the conclusion of theoretical analysis of the image characteristics of voids. In addition, the influence of antenna frequency on the detection for voids are also analyzed.(2) Approaches for detecting voids and for estimating its height were studied, focusing on voids with a height ranging from 0.01m to 0.3m. The Least Squares Method of System Identification and the Tikhonov Regularized Deconvolution were both successfully applied to the detection and dimension estimation of air-filled voids, and their application conditions were discussed. As for water-filled and grout-treated voids, the reflection-amplitude-based dielectric constant method was used for void detection.(3) The approach for estimating the horizontal dimension of voids was studied, focusing on voids with a length ranging from 0.04m to 0.52m. According to the simulating results of air-filled voids, the estimation index was selected, and the linear calculation formula for length of voids was generated by regression analysis. (4) The data processing process was discussed. Also, the effects of different data processing techniques were studied in terms of noise filtering and attenuation compensation, and their influence on the image characteristics was also discussed.