Development of a piecewise nonlinear approximating transformation method for ultrasonic assessment of axial holes in large-scale structures

Abstract

Holes and cracks can form at the bottom or inside of a structure due to the influence of external forces. They can also appear during processes such as exploration drilling and pipeline installation, and fatigue cracks or local separation may occur due to natural phenomena. Defects like hidden holes and cracks are often detected using ultrasonic waves. Detecting such defects typically requires expensive equipment or multiple transducers, but in this study, we develop a novel detection system for hidden holes, utilizing a single ultrasonic transducer and applying a piecewise nonlinear approximating transformation (PNAT) method combined with the modal assurance criterion (MAC). The present study aims to detect the hidden holes in relatively large-scale specimens (300 mm x 100 mm x 100 mm) utilizing information from small-scale specimens (300 mm x 50 mm x 50 mm) along with some assumptions. To implement the proposed method, reflected longitudinal waves are analyzed and employed. A signal processing method is developed and applied to mitigate the influences of the DC components and noises in the measured ultrasonic signals. The filtered signals are then approximately classified through the PNAT process. Finally, various shapes and temperature conditions of hidden holes are evaluated using the MAC method. To validate the proposed approach, several experimental examples considering various diameters, angles and temperatures are conducted. As a result of the validation, the large-scale specimen cases including diameter cases of 13 mm, 7 mm and 4 mm, angle cases of 140 degrees, 120 degrees and 100 degrees, and temperature cases of -30 degrees C, 30 degrees C and 100 degrees C could be classified and detected within relative errors of 5.6 %, 4.5 % and 6.3 %, respectively, compared with the reference signals of lab-scale specimens.