Table of contents

A substantial reduction of the L-H transition threshold power was observed under the W-shaped divertor in JT-60U, in comparison with the open divertor. Radiation power in the main region was also decreased in this case. These differences of radiation power from the main plasma between open and W-shaped divertor were not large enough to account for the apparent reduction of threshold power. In this paper, the emphasis is on the edge plasma parameters which relate to the L-H transition. It was found that edge plasma pressure just before the L-H transition in the W-shaped divertor case became smaller than that in the open divertor case. After the modification of divertor geometry, edge ion collisionality just before the L-H transition also became lower and it was established with lower auxiliary power input. It was suggested that reduction of edge ion collisionality arose from neutral particles near the X-point by the analysis of poloidal profile of neutral density. Therefore, neutral particles near the X-point prevented the threshold power for the L-H transition from more reduction at low plasma density, and caused the shift of plasma density which gave minimum threshold power.

New types of toroidally rotating fluctuations (toroidal mode numbers n = 1 and n = 2) of line-integrated electron density and H_α emission, with frequencies ranging from 10 to 100 kHz, are observed in the sustained spheromak physics experiment (SSPX). The rotating directions of these fluctuations are the same as the direction determined by E×B, while the E and B directions are determined by the gun voltage and gun magnetic flux polarities, respectively. These results take advantage of one distinctive signature of spheromaks, i.e. it is possible to observe toroidal MHD activity during decay and sustainment at any toroidal angle. A theoretical constraint on line-integrated measurement is proposed and is found to be consistent with experimental observations. Fluctuation analysis in the time and frequency domains indicates that the observed density and H_α fluctuations correlate with magnetic modes. Observation of H_α fluctuations correlating with magnetic fluctuations indicates that, at least in some cases, MHD n = 1 modes are due to the so-called `dough-hook' current paths that connect the coaxial gun to the flux conserver, rather than internal kink instabilities. These results also show that electron density and H_α emission diagnostics complement other tools for spheromak mode study.

A simple analytic fluid model based on particle and momentum conservation describes the response of the scrape-off layer to an applied radial electric field. The model explains why the variation of plenum pressure during limiter biasing in TEXTOR shows the opposite behaviour as divertor biasing in Tokamak de Varennes: the mechanical baffles that define the throats of the pump limiter are aligned along the magnetic field, thus preventing collection of the induced perpendicular flux when the drift is directed towards the scoop. Rather, the optimal pumping is expected when the drift is directed away from the scoop due to the increase of the parallel flux needed to satisfy the Bohm-Chodura criterion. Qualitative agreement is obtained with measurements of the parallel flow near the limiter throat, and with measurements of the plenum pressure during negative biasing for both directions of toroidal magnetic field. The results, extrapolated to Tore Supra, predict that the actual design with throats should provide close to optimal pumping with minor modification of the natural radial electric field, but greater performance could be obtained with negative biasing if a toroidally symmetric neutralizer geometry were adopted.

It is pointed out that close to the edge of tokamak plasmas the fluid equations can support one high frequency mode and one low frequency mode associated with drift motions. The high frequency mode is stable in H mode while the low frequency mode can remain at short wavelengths. We study different regimes of the low frequency mode including different ratios of density and temperature lengthscales. The low frequency mode can drive a particle pinch on the H mode barrier.

The radiative transfer in a system at local thermodynamic equilibrium is investigated on the basis of the solution of the (geometrical optics) transfer equation, accounting for the non-local nature of the radiative process due to both re-absorption of the emitted radiation and reflectivity of the walls of the system. The specific case of the electron cyclotron (EC) radiation in a cylindrical fusion plasma with specularly reflecting walls for which an analytical solution can be derived, is addressed and, in particular, the radial profile of the net power density radiated is evaluated by making use of an improved expression for the EC absorption coefficient. A detailed numerical analysis, carried out by varying both the wall reflection coefficient and the radial profile of the plasma temperature, reveals that a reversal of the net power density profile can occur on the plasma outboard for sufficiently high wall reflectivity. From a comparison with bremsstrahlung radiation profiles it is apparent that a local treatment of EC power emission is needed for sufficiently hot plasmas as expected, e.g. in the so-called advanced regimes of DT tokamak reactors. Furthermore, an exact approach is used to check the accuracy of approximate EC net power density profiles as calculated with the CYTRAN code showing that the latter provides a globally reasonable approximation. Evaluating the total EC radiated power from the exact local approach shows that its scaling with the reflection coefficient is very well described by a scaling following from a recently established global model for the EC radiation, which improves the well-known Trubnikov scaling. The results obtained are discussed in view of their possible relevance to affecting the plasma temperature profile.

A folded waveguide (FWG) antenna was used in the ion cyclotron range of frequency (ICRF) on the large helical device (LHD) in the National Institute for Fusion Science. The FWG antenna is a waveguide antenna folded several times in order to make its size smaller. The FWG antenna in the LHD is designed so that slow waves are excited preferentially. It was used in the fourth experimental campaign in 2000-2001 for the purpose of plasma production and plasmas with an average electron density up to 3.0×10¹⁸ m^-3 were obtained. This will be a high enough density for initial plasmas of neutral beam injection or ICRF to obtain plasmas with higher densities and temperatures. This is the first demonstration of the utility of an FWG antenna in magnetic confinement devices. Further investigations were made in order to understand the mechanism of plasma production. The maximum achievable plasma density increased with injection power and gas-puffing rate, and became saturated. The density became higher as the magnetic field strength was increased. Such experimental observations were explained by the wave accessibility conditions of a shear Alfvén wave.

The chemical sputtering yields of CH₄/CD₄ and C₂H_x/C₂D_x have been measured at the divertor plates of JT-60U. Spectroscopic measurements for CH/CD and C₂ spectral bands are applied to estimate the CH₄/CD₄ and the C₂H_x/C₂D_xflux. At the surface temperatures of 380, 440 and 560 K, the CH₄ yield is, respectively, ~0.8%, 1-2% and 2-3%, the C₂H_x yield 1-2%, 3-4% and 4-5%, and the total sputtering yield by hydrogen ions 3-4%, ~8% and ~10%. With increasing ion flux to the divertor plates (Γ_ion), the sputtering yields (Y) decrease, i.e. Y∝Γ_ion^{(-0.05 to -0.40)}. With increasing electron temperature (T_e), the sputtering yields increase, i.e. Y∝T_e^0.5. It is concluded from the result of regression analysis of Y∝T_e^0.5 that the negative dependence of the yields on the ion flux is attributed to the incident ion energies to the carbon plates. The ratio of the sputtering yields by deuterium ions and hydrogen ions is estimated to be ⩾1.5 based on the ion flux measurement by H_α/D_α intensity. The C₂H_x/C₂D_x sputtering yield accounts for ~80% of the total number of sputtered carbon atoms.

We report new measurements of turbulence during ergodic divertor (ED) operation. At low density, some de-correlation of the turbulence is observed with a decrease of the long timescale structures. It is shown that the typical time involved is compatible with a de-correlation mechanism through radial separation of the B field lines by the ED, with an associated parallel length of the order of the distance between two modules of the ED. This observation reinforces the conclusion drawn in [1] and based on computer simulations. The situation changes when the density is increased: the turbulence level is found to increase. At the highest density, the structure of the turbulent signal is modified and the bursty behaviour suppressed by the ED at low density reappears. These observations lead to the conclusion that the turbulence measured at high density is not sensitive to the ED stabilization effect. This indicates that it could be carried by the ions.

A systematic procedure to identify the plasma equilibrium response to the poloidal field coil voltages has been applied to the JT-60U tokamak. The required response was predicted with a high accuracy by a state-space model derived from first principles. The ab initio derivation of linearized plasma equilibrium response models is re-examined using an approach standard in analytical mechanics. A symmetric formulation is naturally obtained, removing a previous weakness in such models. RZIP, a rigid current distribution model, is re-derived using this approach and is compared with the new experimental plasma equilibrium response data obtained from Ohmic and neutral beam injection discharges in the JT-60U tokamak. In order to remove any bias from the comparison between modelled and measured plasma responses, the electromagnetic response model without plasma was first carefully tuned against experimental data, using a parametric approach, for which different cost functions for quantifying model agreement were explored. This approach additionally provides new indications of the accuracy to which various plasma parameters are known, and to the ordering of physical effects. Having taken these precautions when tuning the plasmaless model, an empirical estimate of the plasma self-inductance, the plasma resistance and its radial derivative could be established and compared with initial assumptions. Off-line tuning of the JT-60U controller is presented as an example of the improvements which might be obtained by using such a model of the plasma equilibrium response.

JET divertor load measurements using embedded thermocouples have indicated a large inner/outer target load asymmetry and a significant non-uniformity in the load to the outer target. Furthermore, under H-mode conditions, the load appears to be dominated by the ion component. While the inner/outer target load asymmetry can be understood in terms of the conventional fluid picture, the sharp structures in the deposition profile as well as the ion dominance of the load cannot be readily explained by the fluid approach. The guiding-centre orbit-following Monte Carlo code ASCOT is used to simulate the ion contribution to the target deposition profiles, and this is the first time a narrow structure in the divertor loads has been reproduced in simulations. The importance of the plasma edge collisionality, the atomic, molecular and ionic collisions in the scrape-off layer, and the radial electric field both inside and outside the separatrix are all studied individually in detail.

Simultaneous fast measurements of electron density and temperature, D_α emission and Mirnov probe fluctuations in JT-60U during type I ELMs have been performed. Density fluctuations, measured with two vertical chords of the far infrared interferometer (FIR) at low and high field sides, have been found instrumental in determining spatial structure of type I ELMs (fluctuations localized mainly at the low field side) and duration of its MHD phase. Two different mechanisms for the initial phase of the interaction between the ELM-induced power and particle flows into the scrape-off layer (SOL) and the divertor target have been considered. They may explain the observed difference in the behaviour of D_α emission spikes at the inner and outer targets, which is related to the value of electron collisionality at the pedestal before the ELM. In ELMs with low pedestal collisionality, sharp fronts of D_α emission spikes at the inner and outer strike points and their simultaneous occurrence, coincident with the start of large-scale oscillations observed on the FIR channels, is observed. It is concluded that in such ELMs, plasma-wall interaction is initiated by a power flow conducted to the target through electron channel. In ELMs with high pedestal collisionality (but still of type I), slower evolution of fast signals was observed, with more gradual D_α spikes and clear time delays between outer and inner D_α channels, consistent with the mechanism of convective parallel energy fluxes with the ion sound speed proposed at JET.

Recently, an experimental campaign has been launched on TCV with the aim of exploring and extending the limits of the operating space. The vertical position control system has been optimized, with the help of extensive model calculations, in order to allow operation at the lowest possible stability margin. In addition, the growth rate of the axisymmetric instability has been minimized by choosing optimum values for the plasma triangularity and squareness and by operating close to the current limit imposed by the n = 1 external kink mode. These measures have allowed us to reach record values of elongation, κ = 2.8, and normalized current, I_N = 3.6, in a tokamak with standard aspect ratio, R/a = 3.5.

Divertor plasma characteristics in the Large Helical Device (LHD) have been investigated mainly by using Langmuir probes. The three-dimensional structure of the helical divertor, which is naturally produced in the heliotron-type magnetic configuration, is clearly seen in the measured particle and power deposition profiles on the divertor plates. These observations are consistent with the numerical results of field line tracing. The particle flux to the divertor plates increases almost linearly with the line averaged density. The high-recycling regime and divertor detachment, which are observed in tokamaks, have not been observed even during high density discharges with low input power. Both electron density and temperature decrease with increasing radius in the stochastic layer with open field lines, and at the divertor plate they become fairly low compared with those at the last closed flux surface. This means the reduction of pressure along the magnetic field lines occurs in the open field line region in LHD.

Energetic tail formation by ion-cyclotron-range-of-frequencies (ICRF) heating was studied in the large helical device (LHD), by measuring high-energy charge-exchange atom spectra with a specially developed natural diamond detector, which views the LHD plasma perpendicularly. Comparison of measured effective perpendicular temperature of ICRF-driven H⁺ minority ions with the classical Stix model indicates that perpendicular ions with energy at least up to 150 keV are well confined in the centre region of the LHD. However, deviation from the classical prediction was observed for high-energy perpendicular protons in the outer region of the plasma, indicating ripple-induced transport and charge-exchange losses. With perpendicular fast atom spectrum and flux measurements, small yet notable differences were detected co-injected and counter-injected ICRF-driven beam ion confinement during combined neutral beam injection (NBI) and ICRF heating. These results can be explained by the differences in the orbit topology and ICRF-induced drift.

The large-aspect-ratio model for current-driven external kinks is applied to study control of non-axisymmetric resistive wall modes in tokamaks. Comparison with toroidal computations indicates that the cylindrical instabilities react in similar ways to feedback as the pressure-driven toroidal modes, when the feedback and sensor coils are placed on the low-field side of the torus. However, higher gain is required in the cylindrical case. The cylindrical model is used to gain insights into design issues concerning a feedback system for ITER, with a double wall and superconducting coils. Good control performance and acceptable coil voltages are found in an initial value problem, where the initial conditions correspond to expected noise levels, when sensors for the poloidal field are placed inside the first wall. This can be accomplished with a PI (proportional plus integral) controller from the voltages of the sensor loops to the voltage over the active coils. The two-wall structure of ITER makes control somewhat more demanding than for tokamaks with a single wall. Adequate control can be achieved also when poloidal sensors are placed outside a single wall, but this requires additional derivative action (PID) and the resulting voltages are significantly higher.