MARS-F modeling of post-disruption runaway beam loss by magnetohydrodynamic instabilities in DIII-D

Abstract

A drift orbit model for relativistic test electrons has been incorporated into the MARS-F code (Liu et al 2000 Phys. Plasmas 7 3681), in order to study the runaway electron (RE) behavior in the presence of magneto-hydrodynamic perturbations computed by MARS-F. By implementing the model directly into the MARS-F curve-linear magnetic coordinates, maximal accuracy in representing the full field perturbation is preserved. The updated code is utilized to study the high current RE beam loss in a post-disruption DIII-D plasma, revealing that a fast growing, n = 1 (n is the toroidal mode number) resistive kink instability, at ∼100 Gauss level, can induce significant fraction of RE loss, largely by perturbing drift orbits of REs. A ∼1000 Gauss perturbation fully terminates the RE beam, as found in both experiment and modeling. The 3D field induced loss increases with the perturbation amplitude but decreases with the particle energy. The loss fraction is generally not sensitive to the initial particle pitch angle. The particle velocity change, due to electric field acceleration/deceleration, small pitch angle scattering, synchrotron radiation and Bremsstrahlung, further perturbs the RE trajectory but plays a minor role in prompt RE loss within microseconds time scale. Therefore, the dominant dependencies are simply the RE energy and instability strength. For comparison, a resonant magnetic perturbation field, generated by 4 kAt n = 3 even parity I-coil currents in DIII-D and with the plasma response field included, is found to induce almost no loss for the same RE beam.

Self-consistent simulation of resistive kink instabilities with runaway electrons

“…8 shows the radial structure of plasma displacement on the direction normal to flux surfaces (ξ n ) decomposed into different poloidal harmonics. The displacement is also localized near q = 2, which is in agreement with the linear results found with MARS-F [16].…”

“…The simulation is based on a DIII-D shot 177040, in which a large RE current is generated from the avalanche after the initial disruption, and finally leads to a "second disruption" when the edge safety factor (q a ) approaches 2, and causes the sudden loss of all RE [8]. The linear simulation shows the dominance of the (2,1) kink mode near the edge, in agreement with previous work using MARS-F [16]. In the nonlinear simulation, it is found that as the (2,1) resistive kink mode grows exponentially, more than 95% of the REs can get lost to the wall due to the breaking of flux surfaces and formation of stochastic field lines.…”

Runaway electron beam dynamics at low plasma density in DIII-D: energy distribution, current profile, and internal instability

“…A prompt (on a µs time scale) RE loss domain is computed assuming n = 1 internal kink mode eigenfunction using a drift orbit model for relativistic test electrons recently incorporated into the MARS-F code [84]. Almost no RE loss is found when perturbations of the poloidal magnetic field with a magnitude δB p = 0.1 G are applied at the HFS inside the DIII-D vessel as shown in Figure 10(d).…”

Runaway electron deconfinement in SPARC and DIII-D by a passive 3D coil

“…Experiments aimed at mitigating tokamak disruptions with massive material injection (gas or pellets) have long sought unsuccessfully to exceed the so-called Rosenbluth density to maintain E < E crit even at postthermal quench temperatures [3][4][5]. Injection of either high-Z (Ne, Ar, Kr, Xe) [6,7] or low-Z (He, D 2 ) [7,8] material into an existing runaway electron beam has also been pursued, with D 2 injection showing strong promise experimentally for benign termination of a runaway electron beam by a kink instability, supported by extended-MHD modelling [9][10][11], which has been extended to ITER scenarios [12]. Strategies to prevent the formation of a mature RE beam involve enhancing transport of the seed REs in physical or momentum space, e.g., by stochastic magnetic fields [13][14][15][16][17][18] or wave-particle interactions [19,20], to produce a loss rate that exceeds the avalanche growth rate.…”