Plasma Surface Interactions: Predicting the Performance of PFC Surfaces

PSI2
Plasma Surface Interactions: Predicting the Performance of PFC Surfaces

Summary

The PSI2 project is developing and integrating computational tools for simulating plasma-surface interactions in future magnetic fusion devices.

The PSI2 project is developing and integrating high-performance simulation tools capable of predicting plasma facing component (PFC) operating lifetime and the impact of the evolving surface morphology and composition of tungsten-based PFCs on plasma contamination, including the dynamic recycling of fuel species and tritium retention, in future magnetic fusion devices. This work will enable discovery of phenomena controlling critical PFC performance issues, and quantitatively predict their impact on both steady-state and transient plasma conditions. The outcome of this project will be a suite of coupled plasma and materials modeling tools, and a leadership-class PFC simulator to predict PFC evolution and feedback to the boundary plasma. Success in the proposed research tasks will enable the prediction of both plasma fueling and the sources of impurity contamination that impact core plasma performance, and will lay the foundation for understanding, designing, and developing the materials required to meet the performance objectives of future fusion reactors.

The partnership between FASTMath and PSI2 involves work on multiple applied math fronts, with a particular focus on uncertainty quantification (UQ) and unstructured meshing technologies. We are working on the estimation of uncertainties in models, their parameters, and their predictions in the broad context of the PSI2 project. UQ has already been the subject of a SciDAC partnership activity under SciDAC-3, focusing on the PSI2 code Xolotl. This work continues in the present context and is extended to include other codes within PSI2. In particular, the current project involves applications with different and more complex materials systems and involves UQ thinking across a hierarchy of models over a wide range of scales. The UQ work will rely on the utilization of state-of-the-art FASTMath UQ libraries, including both DAKOTA and UQTk, along with reliance on other available code resources as necessary. Building on the PUMI and MeshAdapt tools, FASTMath will provide unstructured mesh technologies, including particle tracking with interactions with curved boundaries, for use in the fully 3D impurity transport, GITR, and sheath modeling, hPIC, codes.

A Weibull PDF fit to a multi-model data of helium implantation on a tungsten surface. The fit is obtained with Bayesian inference of an embedded model error representation, leading to full uncertainty decomposition including both data noise and model error.

 

Team Members

Barry Smith
Habib Najm
Khachik Sargsyan
Mark Shephard
Onkar Sahni