From Microstructure to Megastructures in Welding: Machine Learning-Enabled

Discover how cutting-edge modelling and machine learning are transforming our understanding of welding across materials and scales.

This lecture explores multi-scale weld modelling — grounded in continuum mechanics and linked to crystal-plasticity and phase-field approaches — and shows how rigorous R6-standard validation ensures real-world reliability. Drawing on projects such as NNUMAN, NeT, and SINDRI, examples include ferritic low-alloy steels (SA508), austenitic stainless steels, nickel, titanium, and tungsten alloys. Learn how machine learning can validate heat source models, link microstructural and structural simulations, and dramatically accelerate computation. The role of data uncertainty will also be discussed, highlighting how it shapes the journey from modelling to performance across processes (electron beam, laser, multi-pass arc), materials, and industrial applications.