[4C4] Component shape optimisation for improved laser-induced phased array non-destructive ultrasonic testing
A Keenan, G Davis, P Lukac, D Pieris, J Singh, K Tant and T Stratoudaki
University of Strathclyde, UK
Component design is a primary aspect of many industries, including construction, automobile and aerospace. Recent developments in manufacturing methods have made it possible to construct intricate parts that offer major challenges to standard non-destructive testing methods. As a result, it is often challenging to reliably inspect parts with complex profiles. For NDT methods based on ultrasound, the ability to effectively image a defect is dictated by the sensitivity and coverage of propagating ultrasonic waves. Most parts have a region of low sensitivity to detection by ultrasonic methods. Regions with low sensitivity can be considered as blind spots; thus, a defect in this area may go undetected. Typically, low-sensitivity areas include corners and zones far from ultrasonic generation and detection sites. Hence, it is vital that a design-for-testing approach be included during the design of components.
This study optimises an arbitrary aluminium component cross-section using a genetic algorithm (GA) for enhanced ultrasonic imaging. In this approach, the sensitivity is maximised for compliance with ultrasonic testing employing laser-induced phased arrays (LIPAs). An arbitrary component cross-section is initially considered and its ultrasonic sensitivity image is plotted based on laser ultrasonic directivity and sensitivity patterns to identify regions of minimum sensitivity. The shape is optimised using a GA as follows. The part boundary is parameterised and parameters are varied within a user-defined envelope size to find an optimum solution. During optimisation, a cross-sectional area constraint is applied so that any area change is within ±10% tolerance of the initial shape cross-sectional area. In this study, shape optimisation was applied for different envelope sizes ranging from 4 to 12 mm. The GA used for each envelope size resulted in a cumulative increased minimum sensitivity for each case. For each iteration of the GA, the optimisation algorithm demonstrated that a region of low sensitivity was transformed into a region of higher sensitivity. For an envelope width of 12 mm, an increase in sensitivity of 7.7 dB was observed. Sensitivity images will be presented according to the original shape and corresponding optimised shape, along with their appropriate sensitivity values. Test samples with a single side-drilled hole (SDH) defect were fabricated to prove experimental verification of the optimisation process; these samples were assessed using LIPAs where the data acquisition method was full matrix capture on the top three planar surfaces. Acquired data were then post-processed using the total focusing method and images were created. The images were analysed and an increase of 5.9 dB defect signal amplitude was observed between the original and optimised design.
This study optimises an arbitrary aluminium component cross-section using a genetic algorithm (GA) for enhanced ultrasonic imaging. In this approach, the sensitivity is maximised for compliance with ultrasonic testing employing laser-induced phased arrays (LIPAs). An arbitrary component cross-section is initially considered and its ultrasonic sensitivity image is plotted based on laser ultrasonic directivity and sensitivity patterns to identify regions of minimum sensitivity. The shape is optimised using a GA as follows. The part boundary is parameterised and parameters are varied within a user-defined envelope size to find an optimum solution. During optimisation, a cross-sectional area constraint is applied so that any area change is within ±10% tolerance of the initial shape cross-sectional area. In this study, shape optimisation was applied for different envelope sizes ranging from 4 to 12 mm. The GA used for each envelope size resulted in a cumulative increased minimum sensitivity for each case. For each iteration of the GA, the optimisation algorithm demonstrated that a region of low sensitivity was transformed into a region of higher sensitivity. For an envelope width of 12 mm, an increase in sensitivity of 7.7 dB was observed. Sensitivity images will be presented according to the original shape and corresponding optimised shape, along with their appropriate sensitivity values. Test samples with a single side-drilled hole (SDH) defect were fabricated to prove experimental verification of the optimisation process; these samples were assessed using LIPAs where the data acquisition method was full matrix capture on the top three planar surfaces. Acquired data were then post-processed using the total focusing method and images were created. The images were analysed and an increase of 5.9 dB defect signal amplitude was observed between the original and optimised design.
