In this thesis, a quaternion-based configuration design optimization framework is developed using a gradient-based optimization method with continuum-based configuration design sensitivity for geometrically exact beam and shell structures. In order to...
In this thesis, a quaternion-based configuration design optimization framework is developed using a gradient-based optimization method with continuum-based configuration design sensitivity for geometrically exact beam and shell structures. In order to accurately represent finite deformation, a singularity-free quaternion is adopted for the parameterization of cross-sectional rotations in shear deformable beam theory. This parameterization preserves the objectivity and path-independence of the theory, which are violated in the conventional rotation vector-based formulations. The variational equation and its corresponding linearization are explicitly derived with the quaternion framework. The unit norm of the quaternion is maintained through an optimal parameterization and an exponential map, which implicitly ensures rotational continuity in built-up structures. The developed formulation is discretized using isogeometric analysis (IGA), providing high-order continuity and geometric exactness. Configuration design sensitivity equations are derived using both the direct differentiation method (DDM) and the adjoint variable method (AVM) based on the material derivative concept. For a given design velocity field, orientation changes are automatically determined by employing the parallel transport frame.
The geometrically exact shell theory uses vector equations rather than curvilinear component equations, which simplifies the governing equations by avoiding the explicit appearance of the Christoffel symbols. The inextensible director vector and its corresponding orthogonal frame are parameterized using the unit quaternion, and the proposed optimal parameterization enables a drilling-free description using only two parameters.
Numerical examples verify that the proposed parameterization preserves the objectivity and path-independence of the formulation, while conventional parameterization based on an incremental rotation vector fails to maintain these properties. Design sensitivity analyses further demonstrate that the violations of objectivity or path-independence lead to inaccurate sensitivities that cannot be corrected by mesh refinement.
Finally, the developed DSA framework is applied to the gradient-based design optimization of metamaterials. The proposed method is used to maximize the twisting response of a translation-twist coupled metamaterial under prescribed axial strain and to design multi-bandgap structures by exploiting Bragg scattering and local resonance mechanisms.