This paper presents a method for imaging defects within 3D curved metallic components using laser-induced phased array (LIPA). Beyond extending LIPA to curved samples, a strategy for accelerating data acquisition is introduced by selecting the most favourable generation-detection locations for a given inspection volume, while reducing false positives by mapping geometry-induced artifacts and regions of low-sensitivity (‘dead-zones’). To move beyond planar or near-planar geometries, a system using two robot arms is introduced to position laser generation and detection points on the sample surface in near-arbitrary fashion, limited only by access. Analytical predictions of detection sensitivity are combined with finite element modelling to map sensitivity throughout the component volume. This mapping guides the robots in selecting optimal pairs to maximize signal-to-noise within the inspection zone. The approach is experimentally validated on cylindrical and semi-cylindrical samples with side-drilled defects. Using only optimal data pairs, 75%–85% of total measurements is eliminated, achieving four- to six-fold speed-ups. The effect of surface waviness is also modelled, predicting enhanced detection by partially removing ‘dead-zones’, at the cost of requiring surface measurement. These combined advances mark a significant step toward in the non-destructive evaluation of complex real-world components. • Laser-Induced Phased Arrays can detect flaws on arbitrary non-flat surfaces. • An adaptive method is presented to scan strategically areas of interest efficiently and with high SNR by choosing the most sensitive generation-detection locations for the region of interest. • Analytical and finite element acoustic modelling are used to map regions of low sensitivity (‘dead zones’) and artifacts. • The method detects flaws in the region of interest 75%–85% faster choosing the most optimal generation-detection locations.
Fuentes-Domínguez et al. (Sun,) studied this question.