In first approximation, charged particles spiral along magnetic field-lines. To avoid particle losses at their ends, field-lines have to be closed. To contrast the tendency of the plasma to expand like a gas of finite pressure, field-lines have to be helically twisted. Closed, helical field-lines are the basic principle by which we confine plasmas.
In a widespread declination of such principle, called tokamak, an electrical current is needed in the plasma, which has the shape of a torus. Plasmas, however, are unstable to excessive currents and pressure gradients. As a result, and, in an effort to reach a lower energy state, these plasmas tend to reorganize themselves in different topologies, often featuring "magnetic islands". These are caused by magnetic reconnection, an ubiquitous phenomenon, visible for example in solar flares.
Islands are undesired because they “short circuit” heat and particles. This degrades confinement, plasma density and temperature. Furthermore, islands interact with the wall surrounding the plasma and with imperfections or “errors” in the confining field. As a result of such interaction, islands can “lock” in a specific toroidal position, grow rapidly, and cause a rapid loss of confinement known as disruption. Disruptions are widely considered one of the main concerns for the success of magnetic confinement fusion.
At the DIII-D National Fusion Facility, we demonstrated the first technique for the suppression of locked islands and avoidance of the associated disruptions: 1) we detect imminent locking by means of magnetic sensors, 2) apply magnetic perturbations to control the toroidal location where the island is going to lock and, finally, 3) we suppress it by means of resonant wave-particle interactions. As the resonance is localized both in real and velocity space, wave absorption results in a localized current. This current generates a field that cancels the magnetic island field.
We also explained why, when magnetic islands rotate (whether viscously dragged by the plasma, or dragged by applied rotating magnetic fields, or by imparted torques) they rotate non-uniformly. The explanation is that the field errors, mentioned above, act like “bumps” along the island trajectory. The reverse is also true: from the location and "height" of the bumps -identified by rotation- we can characterize and quantify error fields, and decide how to correct them.
Non-Fourier-based methods to extract “modes” from magnetic sensor array data and to accurately analyze the topology of islands and other modes of magneto-hydrodynamic activity when this is not well-described by simple mode numbers (in other words, when mode numbers are not “good quantum numbers”). Such methods helped characterize rotating islands prior to their locking.
At present we are compiling and analyzing a database of locked modes and disruptions at DIII-D. The objective is to better understand both phenomena and how they correlate, and to inform algorithms for their prevention and control, also under development in the group.