Mesh Generation from Energy Dispersive Spectroscopy Data of Multiphase Microstructures

Author(s):

Efraín Hernández-Rivera, S. Gary Hirsch, and Brady Butler

Abstract:

Achieving performance of materials when subjected to extreme environments necessitates advanced materials or processing techniques. Bonding of dissimilar materials via brazing is a promising route, but results in complex microstructures with multiple intermetallic phases that are challenging to model. This work presents a methodology for generating digital twins from energy dispersive spectroscopy (EDS) data, enabling accurate numerical simulations of material behavior within the FEniCS framework. The process involves phase identification and clustering from EDS elemental maps, followed by a novel microstructure cleansing approach using a Potts-like Monte Carlo model to smooth interfaces and remove spurious features. Finally, the cleansed microstructure is converted into a suitable computational mesh. This workflow provides a cost-effective and efficient means of predicting material response, supporting the development of advanced materials for demanding applications.