Speaker
Description
Experiments and modeling in DIII-D highlight the key parameters determining the dynamic change in divertor conditions with pellet fueling. In future reactors, pellet injection will be necessary for effective core fueling[Kukushkin2003NF] since a high-performance scenario typically necessitates a scrape-off layer (SOL) that is opaque to neutrals. Due to increasing alpha heating with density, the detachment level in reactors will depend on the density evolution cycle imposed by pellet ablations[Wiesen2017NF]. Integrated modeling in ITER finds that the associated increase in power crossing the separatrix ($P_{SOL}$) via convection may result in divertor reattachment with each pellet.
To explore this phenonmenon, experiments have been performed in DIII-D injecting up to 4.5$\times10^{21}~$atoms/s of D pellets from the high-field side, in a range of plasma scenarios ($B_t=+/-2.2~$T, $|I_p|=0.6~-~2.0~$MA, $P_{inj}=3.0~-~9.0$~MW, $n_G=0.24~-~0.82$). These include type-I ELMy and ELM-free (via negative triangularity, RMP suppression and QH-mode) scenarios. In the majority of cases, pellet injection (with density rise up to $\psi_N\sim0.6$) results in transient detachment of the divertor if starting attached, and a transient deepening of the detachment level if starting detached, lasting $\sim25~$ms. This is consistent with results in EAST attached plasmas with shallow core fueling ($\psi_N>0.94$)[Deng2017PST]. Well-diagnosed ELMy plasmas demonstrate transient detachment with target temperature $T_{e,t}<2~$eV and up to 35% reduction in outer target power loads ($P_{targ}$). Meanwhile, ELM-free scenarios reveal that the transient detachment is preceded by a transient increase in $P_{targ}$ lasting $\sim1~$ms. However, in an ELM-free scenario with relatively steep edge density gradients, no detachment enhancement is observed with pellet injection and $P_{targ}$ increases over $\sim25~$ms.
Time-dependent SOLPS-ITER simulations have been performed to study these dynamics, including time-dependent EIRENE. The pellet is emulated with a transient plasma source inside the confined region, and the radial source profile can be scanned. Experimentally observed divertor dynamics are qualitatively reproduced in simulation, with divertor conditions impacted over two distinct timescales. On shorter timescales ($\tau_1$), transient increases in $T_{e,t}$ and $P_{targ}$ are observed, with the magnitude determined by the pellet-induced $P_{SOL}$ increase. On longer timescales ($\tau_2$), transient reductions in $T_{e,t}$ and $P_{targ}$ are observed, with the magnitude determined by the increase in particle flux crossing the separatrix. The simulated timescales are governed by the pellet deposition location and the energy and particle confinement times, spanning $\tau_1\gtrsim2~$ms and $\tau_2\gtrsim10~$ms in a H-mode plasma, increasing with pellet penetration depth. To ensure continuous core-edge integration in future devices, the pellet deposition location may need to be optimized to the plasma scenario.