Highly efficient numerical model for forward and inverse problems in the physics-based simulation of metal additive manufacturing processes on part scale

• Applicants: Dr.-Ing. Christoph Meier; Professor Dr.-Ing. Wolfgang A. Wall
• Subject Area: Mechanics
• Term: since 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 437616465

Additive Manufacturing (AM) aims at the production of high-performance functional parts (end products) with mechanical properties comparable to processes such as casting, milling or forging. As compared to these classical processes, however, AM offers highest production flexibility and almost unlimited freedom of design, which enables the generation of highly complex geometries or substructures (e.g. lattice-based, honeycomb-like, bionic designs) that cannot be obtained by conventional manufacturing processes. Among the manifold of existing additive manufacturing processes, powder-bed fusion additive manufacturing (PBFAM) of metals has attracted significant scientific and industrial attention because it offers near-net-shape production of near-limitless geometries, and eventual potential for pointwise control of microstructure and mechanical properties. However, these processes are complex and governed by a variety of physical mechanisms. A sub-optimal choice of process parameters might lead to diminished material properties or even to failure of the part during the manufacturing process. The complexity of these processes prohibits an identification of optimal process parameters in trial-and-error manner, resulting in the inevitable need for predictive process simulation.The core objective of this project is the development of a physics-based model embedded in a highly efficient numerical formulation allowing unprecedented predictive simulations of PBFAM processes on part scale while still preserving essential thermo-mechanical effects and process characteristics on the length scale of the powder layer thickness. To achieve this goal, cutting-edge formulations for non-conforming and adaptive multi-resolution finite element discretization schemes as well as novel techniques for combined projection-based and physics-inspired model order reduction are proposed, which optimally exploit the characteristics of PBFAM processes. While the resulting forward model enables the prediction of residual stresses, dimensional warping and material behavior on part level, modern schemes for inverse analysis are incorporated to also address inverse thermo-mechanical challenges. Inverse problems such as the optimal compensation of dimensional warping by a properly adapted initial geometry are of highest practical relevance and can be solved very efficiently by combining these inverse analysis schemes with the proposed model order reduction techniques. The accuracy, efficiency and predictive ability of the overall modeling approach will be assessed on the basis of high resolution numerical and experimental reference data generated for selected benchmark tests.

• DFG Programme: Research Grants
• International Connection: USA
• Cooperation Partner: Professor A. John Hart, Ph.D.