Burn control mechanisms in tokamak fusion reactors

Abstract

The D-T fusion cross section has a significantly positive temperature dependence in the range of temperatures that ITER is expected to operate in. As a result, ITER must have active and preferably also passive control mechanisms that will limit inadvertent plasma power excursions that could trigger runaway fusion heating. Existing predictions of thermal stability are based on models that fail to consider many important physics mechanisms, and the impending operation of ITER provides a strong incentive to revisit this issue. We have identified and investigated the potential of several "burn control" mechanisms including electron cyclotron radiation (ECR), ion-orbit loss (IOL), impurity seeding, and Multi-faceted Asymmetric Radiation From the Edge (MARFE) instabilities that could limit sudden increases in fusion power in the inner core. ECR is the most significant passive burn control mechanism identified, and it becomes particularly important at higher temperatures (30+ keV). EC losses are a function of wall reflectivity and the amount of EC radiation that is generated in the inner core and absorbed elsewhere in the plasma. Because ECR generated in the core by hot electrons can be absorbed in other plasma regions, it can function as an instantaneous transfer of power from fusion alpha-particles in the central core to other regions of the plasma. This would have the effect of instantaneously cooling the center of the plasma and heating the outer core and edge regions, in contrast with the way energy transport is typically modeled. Several active burn control mechanisms are also investigated including adjusting fuel pellet composition, controlling impurity concentrations using on-axis EC current drive, and deliberate MARFE-initiated H-L transitions. Finally, it is concluded that we have likely exhausted the utility of simple 0-D treatments of the plasma for modeling burn control scenarios, and that these should be replaced with multi-nodal dynamics models that treat various important plasma regions in a more granular way. The equations for such a model are developed using two nodes for the confined plasma, one node for the scrape-off layer, and one node for the divertor regions.