Detection of internal defects of concrete for noncontact acoustic inspection method using healthy part extraction

Sugimoto, Kazuko; Toin University of Yokohama; Japan

Sugimoto, K.; Toin University of Yokohama; Japan
Sugimoto, T.; Toin University of Yokohama; Japan

ID: ECNDT-0560-2018
Download: PDF
Session: Civil Infrastructure - AE
Room: J1
Date: 2018-06-12
Time: 13:30 - 13:50

In recent years, degradation of concrete structures built in the period of high economic growth has become a social problem in Japan. We have studied a noncontact acoustic inspection method to detect internal defects of surface layer (~ 10 cm) of concrete such as bridge and tunnel from long distance (5-33 m away). The surface layer of concrete is acoustically vibrated with strong aerial sound waves, and the vibration velocity is measured at a two-dimensional lattice point with a laser Doppler vibrometer. If there are internal defects such as peeling and cracks in the concrete surface layer, flexural vibration, which is the same physical phenomenon as the hammering test, occurs. Obtained data is signal processed by time-frequency gate, noise is reduced, and a change in vibration state due to flexural vibration is obtained. We introduced two acoustic feature quantities (vibration energy ratio and spectral entropy) and proposed a defect detection algorithm. Then internal defects of concrete were visualized by vibrational energy ratio. Generally, it can be said that a healthy part of concrete is acoustically homogeneous and isotropic. From measured results, it has been found that the distributions of two acoustic feature quantities follow a normal distribution, against a healthy part of concrete with a smooth surface. By drawing a scatter diagram using two acoustic feature quantities in case of a simple shaped specimen such as cavity defect, a healthy part of concrete and a defective part can be separated clearly. However, in case of actual concrete structure, measured points, which are difficult to distinguish whether it is healthy or defective, exist. Therefore, we propose a method to statistically extract and evaluate a healthy part of concrete. By statistically evaluating a healthy part of concrete, it is possible to detect concrete internal defects even for actual concrete structures.