Validation of SOLPS-ITER and EDGE2D-EIRENE simulations for H, D, and T JET ITER-like wall low-confinement mode plasmas

Abstract

Both experiments and simulations with SOLPS-ITER and EDGE2D-EIRENE show that the onset of detachment for the low-field side (LFS) divertor – defined here as the line-averaged upstream density (${〈n_e〉}_{\mathrm{edge}}$) at which the plasma flux to the LFS target ($I_{\mathrm{LFS-plate}}$) starts to decrease with increasing ${〈n_e〉}_{\mathrm{edge}}$ – is independent of the isotope mass. However, there are three major simulation-experiment discrepancies: (i) the absolute values of $I_{\mathrm{LFS-plate}}$ and the electron density ($n_e$) in the LFS divertor at the onset of detachment are significantly lower in simulations, i.e., approximately a factor of 2 for $I_{\mathrm{LFS-plate}}$ and a factor of 3-4 for $n_e$; (ii) the degree of detachment – defined here as the difference between $I_{\mathrm{LFS-plate}}$ at the onset of detachment and at an ${〈n_e〉}_{\mathrm{edge}}$ value close to the density limit – is smaller in simulations compared to experiments; and (iii) the experimentally observed larger degree of detachment for D and T plasmas compared to H plasmas cannot be clearly distinguished from the simulation results. There are strong indications that discrepancy (i) is to a large extent caused by neglecting Lyman-opacity effects in our simulations. The simulations predict a similar net volumetric recombination source for all isotopes due to the fact that molecule-activated recombination (MAR) compensates for the reduced electron–ion recombination (EIR) for H, whereas MAR is negligible for D and T. This similar net volumetric recombination source for all isotopes leads to an isotope-independent degree of detachment in simulations. An analysis of the Balmer-$\alpha$ and Balmer-$\gamma$ emission confirms the underestimate of MAR in simulations (especially for D and T) for the JET metallic wall, which was previously observed for devices with a carbon wall. The underestimate of MAR is an important cause for discrepancy (ii) and the fact that there is a stronger underestimate of MAR for D and T than for H explains discrepancy (iii). Extending the plasma grid to the vessel wall increases $I_{\mathrm{LFS-plate}}$ and $n_e$ at the onset of detachment by 25%, and the EIR source increases by 80% in detached conditions. Hence, while the extended grid results are closer to the experimental observations, the previously described qualitative discrepancies still persist.