Description
Disruptions in ITER present the risk of generating large Runaway Electron (RE) beams that may cause untolerable damage when impacting plasma facing components. Assuming short Thermal Quenches (TQs) lasting around 1 ms, the hot tail mechanism, driven by the rapid temperature drop and the resulting non-Maxwellian electron population, may produce dangerous amounts of RE seeds [1]. This motivates modelling with 3D non linear MHD codes to accurately account for electron losses due to magnetic field stochasticity and other 3D aspects of the dynamics. The related physics is highly difficult to capture and remains virtually unexplored.
We leverage the JOREK 3D non-linear MHD code to post-process MHD simulations with test electrons. A dedicated computational framework has been developed to simulate a hot test electron population evolving in the pre-calculated MHD fields and undergoing collisions with the bulk electrons and ions. It has been verified in a reduced phase space against DREAM, and preliminary tested with 3D fields [2].
To approach experimental validation, the DIII-D discharge 178682, identified as producing a hot-tail RE seed [3], is modelled. A 3D MHD simulation including pellet injection is first performed, capturing phenomena typically observed during disruptions, such as stochasticity, double tearing modes, helical cooling, and 𝐸⃗ × 𝐵⃗ mixing. By using markers, and considering or neglecting the 3D components of the fields, the simulations demonstrate the critical role of stochastic losses in reducing the RE seed by orders of magnitude.
Finally, the new framework is employed to perform the first full-geometry estimates of mitigated ITER 15 MA H-mode scenarios, using MHD simulations presented in [4]. We observe that Shattered Pellet Injection (SPI) enables, for a degraded H-mode case, a safe disruption with a seed current below 10e−10 A. However, the baseline H-mode case produces a potentially dangerous seed. Unlike the DIII-D case, including the 3D effects increases the generated seed from 10 mA to 670 A. For the first time, these simulations reveal a generation boost caused by the formation of acceleration cells due to the (1,1) helical cooling, as well as the transport of hot electrons from the core into these regions which overcome stochastic losses.