Resolidification-controlled melt dynamics under fast transient tokamak plasma loads

Abstract

Studies of macroscopic melt motion induced by fast transient power loads and the ensuing damage to plasma-facing components are critical for ITER. Based on state-of-the-art experiments, heat transfer is argued to be strongly entangled with fluid dynamics. The physics model of the MEMOS-U code is introduced. Simulations are reported of recent tokamak experiments concerning deliberate transient melting of beryllium main chamber tiles (JET) and tungsten divertor components (ASDEX Upgrade, JET). MEMOS-U is demonstrated to capture the main physics responsible for melt dynamics and to reproduce the observed surface deformation. The intricate role of resolidification is elucidated.

Numerical benchmark of transient pressure-driven metallic melt flows

“…This benchmark case is to serve as a basis for future upgrades of the underlying physical model, aimed at simulations of fusion-relevant scenarios, such as unipolar arcing on tungsten components and transient beryllium melting during disruptions. Said upgrades concern in particular the inclusion of surface cooling due to vaporization, electron emission and thermal radiation [7,28,29,52], as well as the Lorentz force due to thermionic or disruption currents [6,16,17,27,53,54]. The latter may also require coupling with electromagnetic field equations, typically in the magnetostatic limit [28,53].…”

“…Said upgrades concern in particular the inclusion of surface cooling due to vaporization, electron emission and thermal radiation [7,28,29,52], as well as the Lorentz force due to thermionic or disruption currents [6,16,17,27,53,54]. The latter may also require coupling with electromagnetic field equations, typically in the magnetostatic limit [28,53].…”

“…This is due to the multi-scale nature of such flows, for which an illustrative example is provided by the observations of molten beryllium on upper dump plates in JET [3]. In this case, the melt pool is shown to have a typical depth of the order of 100 µm, while its extent ranges between 1 cm and 10 cm [28] and some small-scale features of the flow require a resolution of 10 µm or below. These scale separation arguments are also expected to hold for ITER, where millimeter-deep pools covering several tens of centimeters are predicted [55].…”

“…These scale separation arguments are also expected to hold for ITER, where millimeter-deep pools covering several tens of centimeters are predicted [55]. Therefore, a combined modelling approach is necessary, wherein the macroscopic characteristics of the flow calculated by costefficient codes [28] are fed into more detailed simulations of restricted spatial domains.…”

“…As a result, compromises must be made depending on which results are sought. For example, if the focus is set on the macroscopic flow features and large-scale melt displacements, simplifications such as the shallow water approximation [24] have been shown to reproduce the experimental data accurately and at reasonable computational costs [25][26][27][28][29]. However, the shallow water equations are not suited to study melt ejection, flow instabilities or strong deformations of the free surface.…”