Error characterization in CFRP shell components using derivative-based optimization processes

Ultrasonic approaches are often used for the non-destructive testing of parts and components. The standard test methods usually only allow statements about the reflectivity or sound attenuation of a defect, but not about its exact size and shape. In order to overcome this limitation, image-based reconstruction methods are used for 3D components. However, they cannot be transferred easily to modern plate or shell-shaped components made of fiber-reinforced plastics with anisotropic acoustic properties. The use of inverse methods for error reconstruction has so far failed due to the lack of available efficient methods for a forward simulation of the ultrasound propagation. In the planned project, novel semi-analytical approaches for the efficient simulation of ultrasound propagation will be used together with
calculus-based optimization strategies. As a result, an inverse method is available such that error geometries, error sizes and error locations can be reconstructed from the ultrasonic signals in shell-shaped, acoustically anisotropic components. In addition to algorithmic differentiation (AD) as basis for the solution of the inverse problem, the Scaled Boundary Finite Element Method (SBFEM) should be used to simulate sound propagation.