Analytische Chemie
Filtern
Dokumenttyp
- Beitrag zu einem Tagungsband (1)
- Vortrag (1)
Sprache
- Englisch (2)
Referierte Publikation
- nein (2)
Schlagworte
- Algorithmic Differentiation (2) (entfernen)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (1)
In recent years carbon über polymers have become a popular light-weight substitute for high-weight materials such as steel. One advantage of carbon fiber polymers is the high strength-to-weight ratio, thus some popular application areas are weight sensitive such as aeronautics or automobiles. As these application areas are especially sensitive to material failure it is of significant interest to characterize material defects which may arise. In this talk we will propose a method to characterize material defects in carbon fiber reinforced polymers using gradient-based optimization methods. The procedure is based on the solution of an inverse problem where simulation data and experimental data is fitted. Here,
gradients of the simulation will be supplied by an Algorithmic Differentiation (AD) tool which greatly enhances the quality of the solution. Numerical examples will be provided.
In non-destructive testing and structural health monitoring with ultrasonic waves, the quantification of damage in components is one of the main tasks. In many shell-like structures, such as plates, pipes, or laminate components, ultrasonic waves propagate as guided waves. Although guided waves enable the testing of large areas, their multimodal and dispersive properties make it challenging to analyze signals. So, there is a need for more advanced algorithms to handle these properties, especially when reconstructing damage position and geometry.
The reconstruction can be formulated as an inverse problem where the measured signals are fitted with a simulative forward model. Due to the small wavelength of ultrasonic waves, classic forward models based on, e.g., the Finite Element Method are computationally intensive. In contrast, the authors use the semi-analytical Scaled Boundary Finite Element Method (SBFEM) to reduce the computational effort. The SBFEM approximates arbitrary long, undamaged parts of the structure with only a few degrees of freedom.
This contribution summarizes a general inverse procedure based on algorithmic differentiation in combination with the SBFEM. Results are presented for damaged 2D cross-sectional models of waveguides. These results include an analysis of the robustness of the proposed algorithms against noise.