Phase Field/DG-FEM Modeling of Metal Additive Manufacturing
Phase Field/DG-FEM Modeling of Metal Additive Manufacturing
Disciplines
Computer Sciences (30%); Mechanical Engineering (50%); Physics, Astronomy (20%)
Keywords
-
Phase Field Method,
Discontinuous Galerkin Finite Element Method,
Melt Pool Dynamics,
Multi-Phase Flow,
Powder Bed Fusion Additive Manufacturing
Additive Fertigung ist ein neuartiges, vielversprechendes Produktionsverfahren von 3D-Bauteilen. Ausgehend von einem digitalen 3D-Modell werden hochkomplexe, leichtgewichtige Geometrien schichtweise hergestellt. Für Metallbauteile hat sich dabei insbesondere das selektive Laserschmelzen (SLM) als attraktiv herausgestellt. Innerhalb einer Schutzgasatmosphäre wird ein dünnes Metallpulverbett auf einer Grundplatte aufgetragen. Darin scannen gebündelte Laserstrahlen anschließend die Konturen der finalen Geometrie, wobei die Pulverpartikel durch den Wärmeeintrag geschmolzen werden und ein Schmelzbad entsteht. Nach dem Scan einer Pulverlage erstarrt das noch flüssige Metall und fusioniert mit den darunterliegenden Schichten. Die Grundplatte wird um die Schichtdicke gesenkt und der Prozess wird wiederholt, bis das Bauteil fertig ist. Da das SLM-Verfahren noch recht jung ist, sind die zugrundeliegenden physikalischen Phänomene noch nicht vollständig untersucht. Somit ist der Zusammenhang zwischen den Prozessparametern und der Bauteilqualität nur unzureichend verstanden. Gegenüber herkömmlichen Technologien ermöglicht die zyklische thermische Prozessgeschichte zwar eine Verbesserung der Materialeigenschaften. Gleichzeitig birgt diese erhöhte Komplexität, wie unvollständig geschmolzene Pulverpartikel, durch Verdampfung entstandene Poren oder verzogene Geometrien, welche zu unzureichenden mechanischen Eigenschaften des Bauteils führen. Die Ermittlung optimaler Prozesseinstellungen ist daher für die Fertigung strukturell einwandfreier Bauteile unerlässlich. Experimentell ist diese sehr aufwendig, da für jedes Bauteil individuell die Prozessparameter durch Versuch und Irrtum neu bestimmt werden müssen. Daher besteht ein enormer Bedarf an präzisen computergestützten Berechnungsmodellen, die eine digitale Prognose der finalen Bauteilqualität anhand der Prozessparameter ermöglichen. Ziel des Projektes ist es, ein computergestütztes Berechnungsmodell zu entwickeln, das die genaue Beschreibung der physikalischen Prozesse im Schmelzbadbereich, d.h. Bildung des Schmelzbades, Wechselwirkung der flüssigen und dampfförmigen Metallphase sowie des Schutzgases, welche maßgeblich die finale Bauteilqualität beeinflussen, ermöglicht. Die zugrundeliegende hohe Dynamik der Prozesse (Zeitskala Mikrosekunde) erfordert robuste und effiziente, neuartige Berechnungsverfahren, um praktisch relevante Längen- und Zeitskalen abzubilden. Die extremen Unterschiede in den Materialeigenschaften an den Phasengrenzen (Dichteverhältnis Metallschmelze zu -dampf ~10000) werden mit der Phasenfeldmethode erfasst. Kombiniert mit der DG-FEM-Methode, ein neuartiges, hochleistungsfähiges numerisches Verfahren, soll somit ein realitätsnahes Berechnungsmodell erstellt werden. Dieses soll ein Schlüsselmodell für das mittelfristige Ziel einer ganzheitlichen, simulationsbasierten Prognose des SLM-Prozesses bilden, um die gesamte Prozesskette von der Pulverbeschickung bis zu den finalen Bauteileigenschaften digital abzubilden.
Additive manufacturing is one of the most promising technologies for producing 3D parts. Starting from a digital 3D model, highly complex, lightweight geometries are built layer by layer. For metal components, laser-based powder bed fusion (LPBF) has proven particularly attractive. In LPBF, a focused laser beam, melts pre-defined contours in thin layers of spread metal powder on a build platform. The heat-induced melting of powder particles creates a melt pool near the laser. After scanning one layer of powder, the molten metal solidifies and fuses with the layers below, gradually forming the geometry of the final part. Subsequently, the build platform is lowered, and the process is repeated until the part is completed. LPBF is a relatively new technology, and the physical phenomena driving the process are not yet fully understood. This lack of knowledge, especially regarding the interplay between process parameters and final part quality, limits the full exploitation of PBF-LB/M. Compared to conventional manufacturing technologies, the cyclic thermal history can lead to improved material properties. However, it also introduces increased complexities, such as evaporation-induced pores and powder redistribution, incompletely melted powder particles or warped geometries, which can lead to insufficient mechanical properties of the final part. The determination of optimal process parameters is therefore indispensable to produce high-quality parts meeting structural requirements. However, this is an extremely cumbersome task when approached by trial-and-error via experiments. Hence, accurate computational models that can digitally predict part quality based on process parameters are needed. In this project, a cutting-edge computational model has been developed to resolve detailed physics of melt pool dynamics in LPBF and to predict defect formation mechanisms, for example, gas inclusions or evaporation-induced gas/vapor flows and thereby induced powder redistribution dynamics. To achieve this, the model resolves the solid, the liquid and vapor/gas phases, as well as liquid-vapor phase transitions, with high spatial resolution. Given the highly dynamic nature of melt pool dynamics, robust and efficient computational tools are crucial for capturing practically relevant length and time scales. We utilize the (dis)continuous Galerkin finite element method together with a diffuse interface tracking scheme based on phase-field approaches to model extreme differences in material properties at phase boundaries. This approach inherently captures complex topological changes of the melt pool, such as keyholes, pores, spattering, etc. To increase computational efficiency, we utilize adaptive meshing schemes and highly efficient matrix-free solvers, in line with current trends in exascale finite-element algorithms. As a central result, new mathematical formulations have been proposed to accurately resolve the strong coupling between molten metal and vapor flow dynamics. The developed computational melt pool model will be a key sub-model for establishing a holistic simulation-based high-fidelity characterization of LPBF.
- Gerhard Leichtfried, Universität Innsbruck , national collaboration partner
Research Output
- 1 Citations
- 15 Publications
- 1 Datasets & models
-
2024
Title Improved accuracy of continuum surface flux models for metal additive manufacturing melt pool simulations. DOI 10.1186/s40323-024-00270-6 Type Journal Article Author Much N Journal Advanced modeling and simulation in engineering sciences Pages 16 -
2024
Title Improved accuracy of continuum surface flux models for metal additive manufacturing melt pool simulations DOI 10.48550/arxiv.2401.12114 Type Other Author Much N Link Publication -
2024
Title A consistent diffuse-interface model for two-phase flow problems with rapid evaporation DOI 10.48550/arxiv.2401.07805 Type Other Author Munch P Link Publication -
2024
Title A consistent diffuse-interface model for two-phase flow problems with rapid evaporation Type Journal Article Author Munch P. Journal Advanced Modeling and Simulation in Engineering Sciences (accepted for publication) Link Publication -
2024
Title Quantitive benchmark for laser powder bed fusion melt pool scale models Type Conference Proceeding Abstract Author Alphonius A. Conference 16th World Congress on Computational Mechanics (WCCM)}, Vancouver, July 2024 -
2024
Title A consistent diffuse-interface finite element approach to rapid melt-vapor dynamics in metal additive manufacturing Type Journal Article Author Much N. Journal in preparation for submission to Journal of Computational Physics Link Publication -
2024
Title The deal.II Library, Version 9.6 Type Journal Article Author Africa P. C. Journal Journal of Numerical Mathematics (accepted for publication) Link Publication -
2025
Title A consistent diffuse-interface finite element approach to rapid melt-vapor dynamics with application to metal additive manufacturing Type Journal Article Author Much N. Journal Computer Methods in Applied Mechanics and Engineering (accepted for publication) Link Publication -
2023
Title Physics-Based Modeling and Simulation of Metal Additive Manufacturing Across Scales Type Conference Proceeding Abstract Author Much N. Conference 2nd German-Italian Workshop on Computational Mechanics (GACM-GIMC), Aachen, Germany -
2023
Title Metal additive manufacturing melt pool modeling based on a novel FEM-based multi-physics model including melting and evaporation Type Conference Proceeding Abstract Author Much N. Conference Sim-AM 2023, Munich, Germany, July 26 - 28, 2023 -
2023
Title A novel finite element formulation for thermal multi-phase flow including melting and evaporation with application to metal additive manufacturing melt pool modeling Type Conference Proceeding Abstract Author Much N. Conference GAMM: 93rd Annual Meeting, Dresden, Germany, May 30 - June 2, 2023 -
2023
Title Mathematically consistent representation of interface fluxes for finite-element-based melt pool modeling in metal additive manufacturing Type Conference Proceeding Abstract Author Much N. Conference Sim-AM 2023, 26.07.-28.07.2023, Garching b. München -
2023
Title The deal.II tutorial step-87: evaluation of finite element solutions at arbitrary points within a distributed mesh with application to two-phase flow Type Other Author Munch P. Link Publication -
2022
Title Efficient Simulation of Complex Capillary Effects in Advanced Manufacturing Processes using the Finite Volume Method DOI 10.1109/iceccme55909.2022.9988504 Type Conference Proceeding Abstract Author Zimbrod P Pages 1-6 Link Publication -
2022
Title A novel finite element formulation for thermal multi-phase flow including melting and evaporation: application to additive manufacturing melt pool modeling Type Conference Proceeding Abstract Author Much N. Conference 9th GACM Colloquium on Computational Mechanics, 21.09.-23.09.2022, Essen, Germany