Beta limit studies and the effect of error fields at low collisionality on COMPASS-D

Recent experiments to determine the operational limits on β at low collisionality on COMPASS-D (R0 = 0.557 m, a = 0.17 m, Bφ(max) = 2.1 T, Ip(max) = 320 kA, k = 1.0-1.7) in ITER-like single null divertor (SND) plasmas have shown deviations from standard ideal MHD β-limits. This is an important issue for ITER since ITER will operate at lower collisionality than most high β tokamak experiments to data and has a design point with a normalized β (βN ≡ aBφB/Ip) of ∼ 2.5. The source of this deviation from the ideal β-limit is neoclassical bootstrap driven islands. Data are presented which demonstrate that the observed dynamics of the instability are well modelled by neoclassical theory. An important issue is the mechanism that determines the onset of the instability since the purely neoclassical theory predicts instability at all values of βp. Data from a density scan are compared to two theories which predict a critical island width for instability. A theory based on the effects of ion polarization currents compares well with the available measurements. The results of a q-scan show that the parametric dependence of the operational β limit is not linear in I/aB and appears to have a peak value when Ip ∼ 150 kA (which corresponds to a q95 of ∼ 4.3). A model based on neoclassical tearing modes is presented that reproduces the general trend of the data as a function of q95. Experiments intended to demonstrate the sensitivity of neoclassical islands to applied error fields show that these modes can be triggered for values of β and ne that are stable in the absence of such magnetic perturbations. Data are presented showing that the scaling of the error field amplitude required to trigger an island is consistent with the predictions of the ion polarization current threshold model. (author). 10 refs, 6 figs