Development of Methods and Tools for the Design of Programmable Materials

Within the thesis at hand, materials are interpreted as arrays of unit cells, similar to mechanical metamaterials. Additionally, each cell is equipped with design parameters, allowing it to adjust its effective behavior. Due to their internal design, these materials have the capability to fulfill a desired function. The simulation and the design of those materials cannot be handled with computational methods for common structures.
To accomplish this task, a data driven simulation and optimization scheme is presented. The aim is, to find a distribution of design parameters, leading to a desired material behavior under given boundary conditions. Therefore, each cell is considered as homogeneous material, with its effective behavior being represented by a data driven surrogate model. With this, it is possible to consider almost arbitrary nonlinear structures with elastic but also viscoelastic behavior.
The usage of the data driven surrogate model speeds up the simulation by a factor of magnitude 1,000 and more. Thus, it is possible to consider optimization problems with more than 105 cells and 2x105 design parameters.