Metal-based additive manufacturing (AM) is an emerging technology that combines precision control of materials deposition with high-resolution local melting for the direct fabrication of functional components. The layer-by-layer approach inherits high degrees of design freedom and enables the production of complex geometries with potentially location-specific materials properties.
Controlling the microstructure and eliminating defects during printing is critical to achieving the desired properties. Numerical modeling of the fabrication process and solidification microstructure offers an effective method to explore the effect of alloy composition and process parameters on microstructure formation, e.g., columnar-to-equiaxed transition, and understand and avoid the formation of solidification defects, e.g., cracks and pores. At the Metal Additive Manufacturing Lab, a suite of computational tools has been developed for process and materials development. This toolset includes a macroscale process model for melt pool dynamics, residual stress and distortion, mesoscale solidification models for grain growth and a microscale solidification model for nucleation and dendritic growth.
The modeling capability will assist in interpreting experimental observations and minimizing the number of experimental tests. Moreover, it compensates for experimental limitations and uncovers new insights into the dynamic solidification process in melt-based AM. The modeling will answer fundamental solidification questions: whether and how This work is a critical part of the Integrated Computational Materials Engineering (ICME) approach. It offers a path to design and optimize a solidification-based manufacturing process from alloy compositions to the design of microstructures.
Project Goals
- A set of validated physics-based models to predict the complex phenomena during AM processes, including melt pool dynamics, heat transfer, solidification, microstructure and defects.
- High throughput methodology based on multiscale models and machine learning algorithms for process optimization and materials development.
- Real-time process interpretation and control enabled by multi-fidelity data from simulations and experiments.
Sponsors
Department of Energy, Oak Ridge National Laboratory, Savannah River National Laboratory