Speaker
Description
The limited understanding of fast-ion distribution functions represents a
significant barrier to optimizing fusion performance. More than just increased diagnostic capabilities, any prior information, in addition to the measurement data, can significantly improve reconstructions of the distribution function. Here we present a method of using ASCOT simulations to help reconstruct the fast-ion distribution function in the 4D phase-space. The role of ASCOT simulations is to encode the correlations between phase-space elements caused by the Coulomb collisions in a suitable form. In this work, this encoding was achieved by solving the reconstruction problem on a basis of slowing down distributions calculated with ASCOT. Due to the sparse nature of the 4D reconstruction, all available prior information needs to be included. For ASCOT simulations, this means that the simulations should be run based on the present fast-ion sources and sinks in the plasma. In this work, we used a source distribution from neutral beam injector (NBI) divided into different phase-space volumes to localise the correlations in the phase space. This localisation implies that regularising the basis vector coefficients regularises the localised correlations.
Here we present reconstructions using ASCOT simulations as bases created with two different localisation schemes on JET: based on the distance along the NBI beamline and based on ρ (normalized poloidal magnetic flux). These results indicate that the encoding of collisional physics using orbit following codes (like ASCOT) into fast-ion distribution reconstructions enhances the diagnostic capabilities essential for future
fusion experiments.
| Presentation type | Poster |
|---|
