Description
This work presents the KINetic Deterministic NEutral Solver (KINDNES) and its first benchmark with EIRENE (Reiter et al., 2005). Neutral particle dynamics are crucial for fluid and gyrokinetic edge plasma simulations in magnetically confined devices. While most existing neutral models fall into either deterministic fluid approximations or Monte Carlo kinetic models, KINDNES offers the advantages of a kinetic model without the drawback of the statistical error inherent to Monte Carlo codes. KINDNES, originally developed as the neutral model for GBS (Giacomin et al., 2022)(Coroado & Ricci, 2022) has been made available as a stand-alone tool. The model implemented consists in a discretization of the Boltzmann equation for each neutral species, integrated along the characteristics. In this way, the problem is reduced to the inversion of a linear system, whose solution has been recently optimized in terms of memory and computational time through the use of Hierarchical Matrices (Guido et al., 2025). This first benchmark uses a deuterium plasma background based on the TCV X21 experimental database simulated with SOLPS-ITER. From this, the particle densities, fluxes, energy densities of atomic and molecular deuterium and source terms due to the neutrals, are calculated using the two codes and are compared. This analysis is repeated for different scenarios, including baffled and unbaffled TCV geometries, in attached and detached divertor conditions. In the preliminary analyses conducted, molecular deuterium is injected into the domain from a source localized on the boundary, and the equilibrium particle and energy densities calculated with both models are in agreement both in terms of profiles and order of magnitude. The hypotheses on particle reflection at the wall boundary are found to have a significant impact on the overall simulation results. A first comparison of the performance of the two codes is presented with a scan over grid fineness and the number of particles used for the EIRENE simulation.