Periodic Reporting for period 2 - MA.D.AM (Modelling Assisted Solid State Materials Development and Additive Manufacturing)
Período documentado: 2022-12-01 hasta 2024-05-31
The MA.D.AM project aims to establish novel scientific knowledge for the fabrication of customized aluminum wires and their application for AM via a fully solid-state based processing route, allowing the production of AM parts with customized mechanical properties, providing at the same time sustainable solution for lightweight design. For this purpose, innovative solid-state materials development and AM processes are utilized to obtain alloys beyond known borders. The solid-state Friction Extrusion process allows for generating phases via thermo-mechanical processing, i.e. severe plastic deformation and elevated temperature, leading to so far unexplored microstructural states, enabling the production of high-performance wire material with tailored properties. To avoid microstructural deterioration and to preserve or even improve the properties of the designed rods, Solid State Layer Deposition, i.e. Friction Surfacing is employed as AM process. To exploit the full capabilities of these processes, the underlying physical relationships along the complete manufacturing chain will be explored via physical models to assist in the experimental alloy and process development. Concepts of machine learning will be employed to establish a digital twin of the process chain.
Therefore, the overarching objective of this project is to establish this experimental reality paired with numerical approaches, leading to a digital twin of the full process chain to ensure its translation to different alloys and AM strategies and to obtain a so far unavailable decryption of the composition-process-(micro)structure-property relationships.
In terms of solid-state AM, the friction surfacing process was studied, gaining an in-depth understanding for single and multi-layer structures. Next to analyses of the temperature evolution and process efficiency, the effect of pre- and post-processing techniques in terms of consolidating volumetric defects was successfully investigated, allowing the fabrication of large-scale structures. Based on in-depth microstructure analyses of the deposited material, a periodic variation in the average grain size along the build direction could be proven. This is attributed to the complex material flow, resulting in varying strain and temperature conditions along the height of each individual layer during deposition, representing a characteristic feature of the process independent of the specific alloy.
In terms of modelling, a primary thermodynamic assessment of the unary systems of the considered aluminium alloys was performed via the CALPHAD method. A phase-field formalism for describing precipitation and the grain size evolution has been implemented. In terms of process simulation, a thermo-mechanical process simulation of friction extrusion has been validated against experimental data. A heat transfer model for friction surfacing has been developed, which has been employed to build a data-enhanced hybrid surrogate model, based on limited available experimental data. Shapley Additive Explanation values were used to evaluate the built machine learning model and most particular findings were confirmed by existing knowledge in literature, thereby allowing for an extension of this knowledge.
Until now, two fundamentally different extrusion types were identified during friction extrusion for the first time, one leading to a fully recrystallized wire material. This is related to established sticking friction condition at the interface between the extrusion die and charge material. Under these friction conditions, the material flow follows a layer-by-layer pattern, for which a novel material flow model has been proposed. It has been shown for the first time that the deposited aluminium alloys via friction surfacing show a characteristic periodic pattern in terms of grain size distribution across the building direction, which is related to the thermo-mechanical loading condition during the deposition process, independent of the specific alloy system. The process-characteristic phenomenon observed is expected to be in direct relation with the material flow during FS layer deposition, which is aspect of ongoing research. Furthermore, fundamental knowledge and understanding have been gained with perspective to the mechanical properties of multi-layer friction surfacing structures, especially with regard to the role of layer interfaces. For the first time, a data-enhanced hybrid surrogate model for the friction surfacing process has been established, requiring much less data compared to a fully experimentally based machine learning model.