| Abstract [eng] |
This work aims to evaluate the reliability of different ultrasonic features with model-assisted probability of detection (MAPOD) curves for adhesive bonding quality assessment. Although adhesive bonding has advantages such as homogenous load distribution, high strength to weight ratio, and ability to join dissimilar materials and complex structures; applications are limited in industries like aerospace due to the lack of reliability in non-destructive evaluation of bonding quality. Hence, we propose a cost-effective reliability assessment methodology with feature-based ultrasonic non-destructive testing technique. <br><br>In order to estimate the detectability of debonding defects and weak bonds, numerical models for aluminum-epoxy-aluminum single-lap joints were built with three different bonding quality: debonding at the interface, weak bond due to contamination, weak bond due to faulty curing. Scanning acoustic microscopy set-up with 50 MHz focused transducer is selected for ultrasonic testing. Semi-analytical finite element method with CIVA is used to model high-frequency ultrasonic wave propagation. Numerical models have been validated with experiments. Five different ultrasonic features are considered to be interest: (1) peak-to-peak amplitude, (2) time delay, (3) frequency (Amax) domain absolute maximum amplitude, (4) frequency shift of Amax, (5) attenuation in adhesive layer. According calculated MAPOD curves, it is shown that different ultrasonic features dominate the detection of debonding defects and weak bonds. Also, frequency based ultrasonic features can improve the weak bond detection reliability. |
