Alexander Mavromaras Presents Innovations in Corrosion-Resistant Alloy Engineering at 19th Middle East Corrosion Conference

Advancing Corrosion-Resistant Alloys Through Multiscale Modeling and Machine Learning

At the 19th Middle East Corrosion Conference in Dhahran, Alexander Mavromaras presented how our understanding of corrosion can be improved through multiscale modeling, combining advanced electronic structure computations and atomistic modeling with continuum phase-field simulations, AI, and machine learning to achieve quantitative predictions and design corrosion-resistant materials.

Corrosion is fundamentally a multiscale problem: atomic interactions drive chemical reactions that ultimately shape microstructural changes and long-term degradation. Alexander highlighted how integrating atomistic simulations, physics-based descriptors, and machine learning enables faster and more accurate predictions of alloy performance. Machine learning models trained on high-quality electronic-structure and molecular-dynamics data can efficiently screen alloy compositions, identify key corrosion-resistant features, and reduce reliance on costly physical testing.

A central component of his talk was the use of phase-field modeling to capture microstructure evolution—grain growth, precipitate formation, and oxidation pathways—that ultimately govern corrosion behavior. Coupled with atomistic insights, phase-field simulations provide a continuous bridge from quantum-scale mechanisms to mesoscale degradation patterns.

These capabilities are brought together within the MedeA environment, which integrates electronic-structure methods, molecular dynamics, Monte Carlo sampling, finite-element and phase-field modules, and automated workflows. This multiscale framework allows researchers to predict corrosion-related phenomena with fine spatial and temporal resolution and to design alloys engineered for specific service conditions.