|About this Abstract
||2016 TMS Annual Meeting & Exhibition
||Computational Materials Engineering for Nuclear Reactor Applications
||First Principles Neural Networks and Diffusion in Nuclear Structural Materials
||Par Olsson, Luca Messina, Christophe Domain, Nicolas Castin, Giulio Imbalzano
|On-Site Speaker (Planned)
Here we will showcase and discuss recent progress in modeling diffusion driven processes in nuclear structural materials. We will show the first application of a newly developed first principles based neural network for kinetic Monte Carlo studies of solute precipitation and radiation enhanced diffusion in ferritic alloys. Using high throughput density functional theory calculations instead of conventional semi-empirical interatomic potentials for the migration barrier predictions ensures increased fidelity. Results are compared to earlier simulations and to experiments and the consequences will be discussed. Moreover, we will discuss a revival of classical experimental techniques for analyzing radiation induced defects from the point of view of experiment-oriented modeling. We will show how to calculate the residual resistivity of defects and solutes in metallic matrices from first principles and how these calculations can improve the analysis of resistivity recovery experiments.
||Planned: A print-only volume