Robust Structural and Process Optimization in the Context of Additive Manufacturing

Refined Cantilever

In the first funding period of the CRC 814, novel optimization methods were developed and analyzed, that make it possible to solve material and topology optimization problems with locally varying material properties. Based on this mathematical foundation and different material models, special goals in the field of additive manufacturing can be considered through suitable nonlinear material parametrizations. These include the optimization of the build direction of components in the build chamber or the optimal control of graded material stiffness. Extending these models, in the second funding period the following objectives are pursued:


- Creation of a Robust Model for Material and Topology Optimization
While the focus in the first funding phase was clearly on modeling the design freedom available in additive manufacturing, experiments have made it clear that at the same time desired material properties and structures cannot be reproduced accurately. Therefore, from the viewpoint of mathematical modeling, a model is required which is able to separate the resulting uncertainties of the freedom in design. Hence, robust design models and associated solution methods need to be developed.


- Optimization of Multi-Material Components
The material optimization models and algorithms will be extended for the simultaneous optimization with several different materials. The resulting problem is here: which material should be used where and how should the material properties be graded in the resulting phases?


- Integration of the Process Optimization
Instead of only using simple constitutive formulas for the material parametrizations as in the first funding period, the material optimization models will now be more interlinked with the process simulation models and their results. For this purpose, on the basis of the results of various projects (A6, B4, C3, C5), effective models for process-structure-property relations will be created. Based on these models, in future it should be possible to obtain optimized scanning strategies for specific components directly in the optimization process.

 

Professor Stingl
Daniel Huebner


Prof. Dr. Michael Stingl
Professur für Mathematische Optimierung
Friedrich-Alexander-Universität Erlangen-Nürnberg
Cauerstraße 11
91058 Erlangen
michael.stingl@fau.de

Daniel Hübner, M.Sc.
Professur für Mathematische Optimierung
Friedrich-Alexander-Universität Erlangen-Nürnberg
Cauerstraße 11
91058 Erlangen
daniel.huebner@fau.de