Table of contents

In the rotamak concept, a rotating magnetic field is used to drive toroidal plasma current in a compact torus device in a non-inductive manner. The latest results from a 0.5 m diameter rotamak apparatus are presented. These show that, for a given filling pressure of hydrogen, it is possible to drive more current, whilst simultaneously preserving the compact torus configuration, by increasing the amount of RF power transferred to the plasma. Attention is drawn to the fact that a fair evaluation of the rotamak concept requires experimentation at higher RF power levels than are presently available

A finite difference scheme that allows significant reductions in computational cost with no practical loss of accuracy is presented to solve, in non-uniform grids, two dimensional (2-D) Fokker-Planck equations in velocity space describing RF heating and current drive (H&CD). Appropriate weighting coefficients are introduced to discretize 2-D Fokker-Planck equations on non-uniform grids and, for a standard lower hybrid RF CD calculation, gains in central processor unit (CPU) time of more than 50% are achieved, with an increase in error of less than 4% for the RF driven current and absorbed power densities and an increase of less than 1% for the CD efficiency

The magnetic field of a low aspect ratio tokamak (LART) deviates rather significantly from the approximately 1/R dependence of conventional tokamaks. The parameterization of current drive efficiency on the basis of conventional large aspect ratio tokamaks becomes inaccurate. The physics of high harmonic ion cyclotron fast wave current drive (FWCD) and low frequency FWCD in LARTs is examined using exact evaluations of the current drive efficiencies from the adjoint technique. The results indicate that the high harmonic ion cyclotron current drive cannot penetrate to the plasma core at the high beta values typical of LART reactors, and hence this scheme may only be applicable for off-axis steady state current drive, while the low frequency current drive penetrates well into the plasma core and is appropriate for on-axis current profile control. The implications for advanced scenarios of LART are discussed

The temporal evolutions of the operating point during the ignition access and ignited operation phases are analysed on the basis of zero dimensional (O-D) equations in order to clarify the requirements for safe control of ignited operation and for the development of diagnostic systems in ITER. A stable and safe method of reaching the ignited operating point is identified as the `higher temperature access' method, being compatible with the H mode power threshold constraints. It is found that the ignition boundary can be experimentally determined by a `thermonuclear oscillation' of the operating point without knowing the power balance equation. On the other hand, the ignition boundary determined by the power balance equation has a larger error bar depending on the accuracy of the diagnostic system. The plasma waveform response to sudden changes in the various plasma parameters during ignited operation is also calculated, and fusion power regulation is demonstrated by feedback control of the fuelling and auxiliary heating power

At the GOL-3 facility, experiments on the interaction of powerful hot electron streams with various materials have been performed. For energy densities of the hot electron stream above 10 MJ/m² an explosive-like erosion was observed, which at energy densities of 30 MJ/m² reaches 500 mu m for fine grain graphite and 200 mu m for tungsten. Under these conditions, the corona of the carbon vapour cloud has temperatures below 1.2 eV and densities up to 10¹⁷ cm^-3. It propagates along the magnetic field lines with maximum velocities of 2.1*10⁶ cm/s. The longitudinal and transverse (along and across magnetic field lines) vapour velocities of the colder bulk plasma are about 10⁶ cm/s. A model for explosive-like erosion was developed and tested against the GOL-3 results. For graphite the destruction threshold is 10 kJ/g. This value is considerably lower than the vaporization enthalpy of 20.5 kJ/g for three atomic vaporization. The validated model was applied to a numerical analysis of the occurrence of explosive-like erosion for ITER disruptions and runaway electrons. If the energy density of the runaways remains below 30 MJ/m², explosive-like erosion of graphite occurs for electron energies below 20 MeV. For the energetic tail of Maxwellian plasma electrons with temperatures up to 20 keV and power densities of 10 MW/cm² without any angular spread, explosive-like erasion becomes comparable to erosion by vaporization

It is found that low-n toroidicity induced Alfven eigenmodes (TAEs) excited by ion cyclotron resonance heating can be stabilized by counter-tangential neutral beam injection in JT-60U. The deformation of the radial structure of the TAE mode due to the shear in toroidal rotation is a plausible stabilizing mechanism. This mechanism is also expected to be effective in suppressing high-n TAE modes. Real time control of TAE mode amplitudes using toroidal rotation control is possible for burn control in fusion reactors

A one dimensional (1-D) model is presented for the investigation of the time evolution of enhanced reverse shear (ERS) confinement modes observed in recent tokamak experiments. The cases of frozen and evolving magnetic shear are studied and effects due to pressure gradient driven bootstrap currents are incorporated. The formation and evolution of transition fronts as a result of the bifurcation in the core solution has been investigated in detail

Operation of the tokamak ISTTOK in a multicycle alternating square wave plasma current regime is reported. Discharges with seven half-cycles without dwell times, over a total time span of about five times the maximum duration of a single DC discharge, were obtained by feeding the primary of the transformer with an electrolytic capacitor bank switched in polarity by a fast insulated gate bipolar transistor (IGBT) H bridge

In reversed magnetic sheaf-plasmas, fast ion confinement is anticipated to deteriorate because of a weak poloidal magnetic field in the core. In fact, in experiments at JT-6OU considerable depletion of triton burnup in reversed magnetic shear has been observed, compared with normal (positive) magnetic shear. The experimental triton burnup in the reversed shear was 11 to 39% of that expected, while that in the normal shear was 64 to 87%. Here, the expected burnup is based on one dimensional (1-D) calculations that assume classical slowing down and no diffusion of fast tritons. Orbit following Monte Carlo simulations, which treat finite orbit effects and ripple transport processes properly, match the experimental depletion and indicate that ripple loss is responsible for the enhanced triton loss in reversed shear operations. The results raise concerns about serious megaelectronvolt ion loss in reversed shear operations in steady state tokamak reactors

Neoclassical magnetic islands are observed to limit the achievable beta in COMPASS-D low collisionality single null divertor tokamak plasmas with ITER-like geometry (R₀=0.56 m, B₀=1.2 T, I_p=120-180 kA, k=1.6, epsilon =0.3). The limiting beta is typically well below that expected from ideal instabilities with maximum beta N in the range of 1.6 to 2.1. The plasma is heated with up to 1.8 MW of 60 GHz electron cyclotron resonance heating (ECRH) at the second harmonic with X mode polarization. The time history of the measured island width is compared with the predictions of neoclassical tearing mode theory, with good agreement between theory and experiment. The measured islands have a threshold width below which the mode will not grow. The density scaling of the point of onset of the measured instabilities is compared with two theories that predict a threshold island width for the onset of neoclassical tearing modes. Applied resonant helical error fields are used to induce islands in collisionality regimes wherein the neoclassical islands do not occur naturally, allowing the study of the behaviour of neoclassical tearing modes in this regime. The critical beta for the onset of neoclassical tearing modes is seen to be ~3 times higher in the naturally stable region. This observation is compared with the predictions of both threshold theories. A simple expression for the q scaling of the maximum achievable beta N in the presence of neoclassical tearing modes is derived on the basis of the assumption of a maximum allowable island width. The predicted q scaling of this beta limit is compared with data from a q scan, and the results are in good agreement

The L mode and H mode density operational window in the vicinity of the density limit has been investigated with a combination of gas puff refuelling and improved fine tuning of neutral beam injection (NBI) heating power. In this way, a novel strategy is achieved by means of a parallel increase of density and heating power. As the density limit is approached, H modes degrade into L modes independently of heating power; this is in contrast to the generally accepted L to H mode threshold scaling P_heat^L-H varies as n_eB. Furthermore, contrary to the well known heating power independent Greenwald limit, the L mode density limit increases moderately with rising heating power, n_e^DL varies as P_heat^0.3+or-0.1, if a simple power law is assumed. The power dependence becomes more obvious when analysed in terms of edge densities and powers flowing across the separatrix into the scrape-off layer, n_e^sep varies as P_sep^0.6+or-0.2. The corresponding H mode studies show that before an H mode quenches into an L mode the maximum achievable density (i.e. The H mode density limit) is practically independent of the heating power, as observed on many machines

A one dimensional (1-D) computer simulation of core particle and energy transport in the reversed field pinch (RFP) is described, which models transport driven by stochastic magnetic fluctuations along with a self-consistent calculation of the neutral atom profile. Experimental plasma parameters are reproduced using fluctuation levels in the experimental range. The positive ambipolar potential, which reduces the electron particle flux to the ion rate, provides an important additional ion heating mechanism: energy is transferred from electrons to ions as they move radially along the electric field. This, combined with classical electron-ion friction and comparatively good ion energy confinement, leads to ion temperatures matching experimental values in some cases. A simple anomalous heating model is applied, however, for simulation of high T_i values in low collisionality cases. The widely observed experimental scaling β_pe ∝ (I/N)^-2/3 is consistent with energy loss dominated by electron heat conduction with χ_e ∝ T_e^1/2, demonstrating the universality of magnetic fluctuation induced transport in RFPs

Simulation results are presented from a finite element calculation of transport in the edge and divertor of TdeV. In addition to the usual transport processes, this model also includes a calculation of the electrostatic potential and the associated E*B drifts. The inclusion of E*B drifts is shown to be essential in the calculation of the parallel and toroidal flow velocities. With the inclusion of drifts, the effect of toroidal magnetic field reversal is studied, particularly with regard to in-out asymmetries in the H_alpha radiation and power deposited on the plates. Relatively weak asymmetries are calculated on the basis of these drifts alone, and it is concluded that the much stronger asymmetries observed experimentally must be associated with other physical processes not yet included in the model. The transport equations are based on fluid conservation equations for particles, parallel (to the magnetic field) momentum, electron and ion energy. For simplicity, a single ion species is considered and, when modelling recycling, neutral transport is treated in the diffusion approximation. All transport equations are discretized with finite elements on a triangular unstructured mesh that is locally aligned with the flux surfaces. The use of such a mesh allows an accurate representation of the simulation geometry, with boundaries of arbitrary shapes and complexity. In particular, it is possible to account for divertor plates that are not orthogonal to the flux surfaces without any penalty for the validity of the transport equations. The analysis presented here concerns a low density attached ohmic TdeV discharge in a single null geometry, with the X point above the midplane

Density profile perturbations due to large rotating m/n=2/1 magnetic islands have been studied by means of a four channel pulsed radar reflectometer in TEXTOR-94. The time of flight of the microwave pulses was strongly affected by the shallow density gradients over the magnetic islands. A detailed analysis showed the existence of secondary reflection layers inside the magnetic island, which is an indication of a strong density peaking. This may be explained by assuming that the particle transport is reduced inside the magnetic island

The transient operating performance of magnetic confinement devices is often limited by one or two unstable MHD modes. The feedback stabilization of MHD instabilities is an area of research that is critical for improving the steady state performance and economic attractiveness of magnetic confinement devices. This growing realization motivated a Workshop dedicated to feedback stabilization of MHD instabilities, which was held from 11 to 13 December 1996 at Princeton Plasma Physics Laboratory. The resulting presentations, conclusions and recommendations are summarized.

Corrections to Nuclear Fusion37 (1997) 1303.