Center for Integrated Simulations of Fusion Relevant RF Actuators
This project will, for the first time, make it possible to explore the self-consistent interaction of RF power with the short mean free path scrape-off layer, including the effects of plasma sheaths, ponderomotive forces near an antenna, and turbulence and transport.
The capability to reliably heat and drive current using radiofrequency (RF) power is essential for steady state operation of future magnetic fusion reactor options. Reliability is presently limited by our understanding of how applied RF power interacts with the scrape-off-layer (SOL) plasma and surrounding material surfaces. The leadership-class computing systems offer the computational resources needed to execute to predictive simulations of the controlling processes. This understanding will allow the design of efficient and robust RF actuators. For the first time, we will make it possible to explore the self-consistent interaction of RF power with the short mean free path SOL, including the effects of plasma sheaths, ponderomotive forces near an antenna, and turbulence and transport. The new simulation capability will make it possible to answer critical questions relating to how RF power modifies properties of the SOL, and how, in turn, the SOL affects the propagation and absorption of RF waves. The high geometric fidelity required to describe the magnetic field in the 3D solid geometries of the RF launching structures and the surrounding vacuum vessel amplify for the high fidelity simulation tools to be developed.
FASTMath team members will develop a high order finite element analysis tool for solving the RF equations. The high order discretization methods in the Modular Finite Element Method (MFEM) library are ideally suited for use in the development of a scalable RF field simulation tool for general domains including fully detailed antenna, wall and SOL geometries; nonlinear sheath boundary conditions; etc. We will also provide unstructured mesh tools to generate and adapt the meshes over the complex geometric models involved. The ability of the Parallel Unstructured Mesh Infrastructure (PUMI) to interact directly with CAD geometry in the support of distributed curved high-order meshes will provide the meshes to be analyzed in MFEM. The MeshAdapt tools will be further extended to accept initial lower-order meshes, curve them to ensure sufficient geometric approximation (at least to 6th order), and to adapt the high-order curved meshes as needed to control the mesh discretization errors.