Defect characterisation in FRP shell components using derivative-based optimisation methods

部件的无损检测在许多情况下使用超声波技术进行。标准化的检查技术通常允许对缺陷的反射率或声音衰减得出结论，而不是提供有关其大小和形状的信息。为了克服这一限制，成像技术被应用于大体积部件的缺陷重建。然而，这些技术不一定适用于由纤维增强塑料（FRP）制成的具有各向异性声学特性的现代板状或壳状部件。由于缺乏有效的技术来模拟超声在这些结构中的传播（正演模型），迄今为止应用逆技术进行缺陷重建失败。造成这种情况的原因是波长与部件尺寸的不利比例，这使得成熟的数值技术达到了极限，因为这些数值技术使整个检测体积离散（例如FEM）。提出的研究项目应用新颖的半解析方法来有效地模拟导波超声传播。分析优化策略将补充这些。该方法将产生一种逆向技术，该技术允许利用壳形各向异性部件中获得的超声数据重建缺陷的几何形状、尺寸和位置。为了实现这一目标，计划开展一个合作项目，将各种方法相互联系起来。求解逆问题的算法微分法（AD）和模拟波传播的尺度边界有限元法（SBFEM）将得到应用。利用不同的离散化方案，SBFEM可以非常有效地模拟换能器产生的超声波在无缺陷区域的传播，以及波场与缺陷的相互作用。为了使技术能够解决问题，必须对其进行相应的扩展和修改（例如，通过考虑各向异性和阻尼以及优化网格）。用于计算优化过程的衍生信息的AD程序也必须扩展并适应模拟代码。这两种技术将分别适应，然后结合起来。实验将支持这项调查。在项目的第一阶段，将进行各向同性板的开发和验证。在第二阶段-将在第一阶段成功验证后申请-该项目将逐步扩大，以处理具有可变截面的各向异性壳形结构。敏感性研究将评估反向方法的可靠性。主要目标是开发一种可靠的成像技术，用于检查FRP结构的缺陷重建，但此外，该项目的科学权重也与所关注的数值模拟和优化技术的进一步发展有关。

Non-destructive testing of components is in many cases performed using ultrasonic techniques. Standardised inspection techniques generally allow drawing conclusions on the reflectivity or sound attenuation of a defect, rather than providing information about its size and shape. To overcome this restriction imaging techniques are applied to reconstruct defects in voluminous components. These techniques, however, can not necessarily be applied to modern plate-like or shell-shaped components made of fibre-reinforced plastics (FRP) which exhibit anisotropic acoustic properties. The application of inverse techniques for defect reconstruction so far fails due to the lack of efficient techniques to simulate the propagation of ultrasound (forward models) in these structures. The reason for this is the unfavourable ratio of wavelength to the components’ dimensions which brings well-established numerical techniques to their limits since these discretise the whole inspection volume (e.g. FEM). The proposed research project applies novel semi-analytical approaches for efficient simulation of guided wave ultrasound propagation. Analytical optimisation strategies will complement these. The approach will result in an inverse technique which allows for the reconstruction of defect geometries, sizes, and positions using ultrasonic data acquired in shell-shaped anisotropic components. To achieve this goal a cooperative project is planned to interconnect various approaches. Algorithmic Differentiation (AD) to solve the inverse problem will be applied as well as the Scaled Boundary Finite Element Method (SBFEM) for simulating the wave propagation. Using different discretisation scenarios SBFEM provides benefits for the extremely efficient simulation of transducer-generated ultrasonic wave propagation in defect-free areas, as well as wave field interaction with defects. To empower the technique for the addressed issue, it has to be expanded and modified accordingly (e.g. by considering anisotropy and damping as well as optimising the meshing). The AD procedures for calculating derivative information for the optimisation process also have to be expanded and adapted to the simulation code. Both techniques will be adapted separately and then be combined. Experiments will support the investigation. In the first stage of the project, the development and the validation will be performed for isotropic plates. During a second stage - to be applied for after successful validation of the first phase - the project gradually will be expanded to cope with anisotropic shell-shaped structures with variable cross-sections. Sensitivity studies will evaluate the reliability of the inverse approach. The primary goal is the development of a reliable imaging technique for defect reconstruction for the inspection of FRP structures, but moreover, the scientific weight of the project also links to the further development of the numerical simulation and optimisation techniques under concern.