Reinforcing steel bars (Rebar) in the concrete need to be inspected to prevent the structure from collapsing. Magnetic non-destructive testing methods can inspect the rebar without the influence of concrete. However, since the inspection is performed at a high lift-off on concrete, a strong applied magnetic field is required in order to measure a sufficient magnetic signal generated from the rebar. In addition, if a single sensor is used to measure the rebar, this requires the scanning of extensive area along the invisible rebar covered with the concrete. To overcome these problems, we have developed a method to detect and image the defect signal using unsaturated AC magnetic flux leakage (USAC-MFL) testing combining a magnetic sensor array probe. The USAC-MFL is a method that uses an alternating magnetic field in the initial magnetization characteristics of steel, which does not require a strong applied magnetic field compared to the conventional magnetic flux leakage method. By measuring the magnetic flux from the rebar in a wide area using the magnetic sensor array probe, the defect signal was able to be detected and imaged. The defect signal was able to be confirmed up to the lift-off of 100 mm even when the defect was half length of rebar or located on the back side.

Fatigue cracks occurred in the welds of steel structure are easily propagated. Therefore, it is important to detect the fatigue cracks at an early stage. However, most of the steel structures are protected by surface coating, which requires removing and restoring the coating for the inspection. This makes the inspection time-consuming. Magnetic methods such as eddy current testing (ECT) and magnetic flux leakage (MFL) can be used without removal of surface coating. However, these magnetic methods are difficult to apply to the shape of the welds and affected by nonuniform magnetic permeability. In this study, we developed a sensor probe adapted to the shape of the welded joint using the unsaturated ac MFL (USAC-MFL), which is less affected by the nonuniform magnetic permeability. To evaluate the performance of the developed sensor probe, the welded joint samples with the surface coating and shallow cracks of which depth ranges from 1 to 5 mm were measured. The measurement results showed a strong correlation between the intensity of the MFL and the crack depth. The detection performance was evaluated using the confusion matrix and the receiver-operating characteristic (ROC) curve. The area under curves (AUCs) of the ROC curves were around 0.9. In order to maximize the true positive rate (TPR) and minimize the false negative rate, the threshold of the intensity to identify a crack was obtained using the ROC curve. As the results of evaluating the detection performance using the obtained threshold, the four metrics (accuracy, precision, recall, and specification) were all above 0.8.

Conventional eddy current testing (ECT) using a pickup coil probe is widely employed for the detection of structural cracks. However, the inspection of conventional ECT for steel structures is difficult because of the magnetic noise caused by the nonuniform permeability of steel. To combat this challenge, we have developed a small magnetic sensor probe with a dual-channel tunneling magnetoresistance sensor that is capable of reducing magnetic noise. Applying this probe to a complicated component of steel structures—such as the welds joining a U-shaped rib and deck plate together—requires the reduction of signal fluctuation caused by the distance (liftoff) variations between the sensor probe and the subject. In this study, the fundamental crack signal and the liftoff signal were investigated with the dual-channel sensor. The results showed that the liftoff signals could be reduced and differentiated from the crack signals by the differential parameters of the dual-channel sensor. In addition, we proposed an extraction technique for the crack signal using the Lissajous curve of the differential parameters. The extraction technique could be applied to the inspections not only for flat plates but also for welded angles to detect cracks without the influence of the liftoff signal.

Infrastructure facilities that were built approximately half a century ago have rapidly aged. Steel sheet piles, the inspection object in this study, are severely corroded, resulting in cave-in damages at wharfs. To solve such a problem, non-destructive inspection techniques are required. We previously demonstrated plate thickness measurement using extremely low-frequency eddy current testing. However, when the steel sheet piles are located in water, shellfish adhere to their surface, causing a lift-off of several tens of millimeters. Therefore, this large lift-off hinders the thickness measurement owing to fluctuations of magnetic signals. In this study, sensor probes with different coil diameters were prototyped and the optimum size for measuring steel sheet piles at high lift-off was investigated. Using the probes, the magnetic field was applied with a lift-off range from 0 to 80 mm, and the intensity and phase of the detected magnetic field were analyzed. Subsequently, by increasing the probe diameter, a good sensitivity was obtained for the thickness estimation with a lift-off of up to 60 mm. Moreover, these probes were used to measure the thickness of actual steel sheet piles, and measurements were successfully obtained at a high lift-off.