Abstract Details
Abstracts
Author: Rinkle Juneja
Requested Type: Consider for Invited
Submitted: 2024-04-09 08:08:39
Co-authors: M.Alnaggar, C.M.Parish, S.Reeve, P.Seleson, J.Wienmeister, Y.L.Pape, J.W.Coenen, M.Diez, M.Richou, M.Lemetais, A.Durif, M.Wirtz
Contact Info:
Oak Ridge National Laboratory
1 Bethel Valley Rd
Oak Ridge, Tennessee 37830
United States of Ame
Abstract Text:
The deployment of sustainable fusion pilot plants demand addressing the challenges associated with the lifetime of fusion wall materials under extreme reactor conditions. Exposure to extreme heat fluxes result in thermal-fatigue induced cracks in the plasma-facing materials (PFMs). These cracks initiate from the microscopic scale and propagate to the macroscopic scale. Since the microstructure of a material largely determines its risk for cracking, in this talk, I will introduce a combined machine learning and modeling framework to predict cracking thresholds of PFMs under reactor-relevant heat fluxes, starting from the microscopic scale. A comprehensive database for 2D tungsten microstructures corresponding to different fabrication conditions is curated using machine learning (ML) generative adversarial networks. The training dataset includes experimental scanning electron microscopy (SEM) images for tungsten obtained from the electron-beam facility JUDITH and WEST tokamak. The robustness of the ML-generated data is ensured by minimizing the loss functions and by statistical validations against experimental data. From the resulting 2D microstructure database, 3D microstructures are generated using DREAM-3D software. The 3D microstructure database is used to inform the mesoscale modeling framework, namely, Multi-Physics Lattice Discrete Particle Model. The computed mesoscale homogenized constitutive relations are incorporated into the peridynamics-based fracture mechanics code CabanaPD to simulate macroscale fracture and component failure. The developed multi-scale modeling framework will enable understanding of important plasma-material interactions that determine the lifespan of fusion wall materials, thereby supporting the component design developments for fusion pilot plants.
*Research sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the U. S. Department of Energy.
Comments:
I would like to request scheduling of my talk either May 6 (Monday) or May 7 (Tuesday) due to my travel to another conference.