Table of contents

Fifth IAEA Technical Committee Meeting/US - Japan Workshop on H-Mode Physics

The Fifth IAEA Technical Committee Meeting/US - Japan Workshop on H-Mode Physics was held at the Princeton Plasma Physics Laboratory, Princeton University, Princeton, NJ 08543, USA on 18 - 20 September 1995. The meeting was attended by about 80 scientists from the US, Europe, Japan and Russia. The development of the scientific program and paper selection were carried out by the Scientific Program Committee, consisting of:

M J Greenwald S M Kaye (PPPL, Princeton University, USA (Chairman)) D J Campbell (JET, UK) V E Golant (Ioffe Institute, Russia) M J Greenwald (PFC, MIT, USA) R J Groebner (GA, USA) S-I Itoh (Kyushu University, Japan) Y Miura (JAERI, Japan) H Zohm (Max-Planck Institute, Germany)

The program of this meeting differed from those of previous ones, in that more time was devoted to topical discussion. The device-specific invited talk session was limited to five talks from those devices which only recently started doing H-mode experiments; these were TCV, Compass-D, Tuman-3 and Tuman-3M, TEXT-U and Alcator C-Mod. The following six sessions over a day period were devoted to specific H-mode topics. Each session focused on one topic, and featured a topical overview talk covering recent results (results since the last meeting), a discussion period of length approximately equal to that of the overview talk (anywhere from 25 minutes to one hour), and then a poster session for that topic. All the posters were up for the entire meeting, but were manned only during the time period relevant to the topic of interest. The six topics and their speakers were: Transition physics and scaling (T Carlstrom, GA), LH transition theories and theory of the H-mode (D J Ward, JET), Physics and role of ELMs (H Zohm, Max-Planck), Physics of the combined H-Mode (Y Kamada, JAERI), Active control of the H-mode (M Mori, JAERI) and Compatibility with reactor operations (R Goldston, PPPL). Sixty-three abstracts for posters were accepted.

The invited, overview and contributed papers are published in this special issue of Plasma Physics and Controlled Fusion. The papers to be published were all refereed by conference participants. This special issue will be distributed to all conference registrants and to regular subscribers to the journal. Additional copies can be provided by the Guest Editor or can be purchased directly from the publisher.

S M Kaye Guest Editor

H-mode confinement is observed in COMPASS-D in ohmic and ECR heated single null divertor configuration discharges over a wide parameter range. At low densities the L - H transition power threshold shows a sharp increase, in contrast to scaling law behaviour. Transition to ELM-free H-mode is better correlated with a critical value of plasma pressure than edge power density. In experiments influencing H-mode dynamics, edge current density is found to be destabilizing, suggesting the possible role of the peeling mode. ELM characteristics are discussed, and impurity rotation observations and derived radial electric fields presented.

The focus of the TUMAN-3 and TUMAN-3M tokamaks programme is on issues of improved confinement. The transition from an ordinary ohmic regime into improved confinement mode has been found in circular limiter configuration in a vessel with all-metallic walls and limiters. The signatures of the H-mode in auxiliary heated tokamaks have been observed in this regime. The crucial role of the radial electric field was found in experiments with internal probe biasing. Other techniques were demonstrated to trigger H-mode: short increase of the working gas puffing rate, minor radius magnetic compression and pellet injection.

The scaling of the energy confinement time in ohmic H-mode was obtained, which differs dramatically from the scaling for the ordinary ohmic regime. A strong dependence of on plasma current was found. The scaling for the ohmic H-mode is consistent with the scaling proposed for devices with powerful auxiliary heating (JET/DIII-D H-mode scaling). The result shows that H-mode physics is universal in tokamaks with different geometries and heating methods.

In 1994 a new vacuum vessel was installed in the TUMAN-3 tokamak. The modified device, TUMAN-3M, is able to produce higher and , up to 2 T and 0.2 MA, respectively. During the first operational period a plasma current of 0.15 MA was achieved at T, which corresponded to . The impact of the quality of wall coating on confinement was asserted. The longest energy confinement time (30 ms) was observed under the conditions of best boronization.

H-mode transitions in TEXT-U limiter plasmas have been observed at and kA ( kW) with at least 300 kW of central electron-cyclotron heating (ECH). These are dithering transitions which are induced by sawtooth crashes and display the typical signatures of H-modes ( drop, spontaneous density increase, evidence of a transport barrier). However, they show only a slight improvement over L-mode energy confinement. The vessel walls are boronized and conditioned prior to experiments to achieve low-impurity influx and particle recycling. Discharges which undergo transitions are fuelled almost entirely on residual recycling. Transitions are observed when limited on a toroidally localized top or bottom limiter and, more often, when the limiter surface is `fresh', which is achieved by alternating between top and bottom limiters on successive shots. No strong dependence upon the distance from the low-field-side limiter has been found. Transitions are not yet observed when limited on the high-field-side wall tiles or in the case of TEXT-U diverted configurations. Preliminary measurements with the 2 MeV heavy-ion beam probe (HIBP) (in the core) and Langmuir probes (in the edge) indicate that the plasma potential drops outside the q = 1 radius while only small changes are observed in the density fluctuation level.

H-modes exhibiting improved confinement above the L-mode are achieved in Alcator C-Mod with ICRF and with ohmic heating alone without boronization. Both ELM-free and ELMy H-modes are obtained with total input power from 0.75 to 4.2 MW over a range of densities (0.8 to ) and toroidal fields (3 to 8 T). Type III ELMs are often observed to have coherent, high m and n precursor oscillations with frequencies of 100 - 160 kHz. The threshold power required to achieve the H-mode increases with density and toroidal field, in rough agreement with scalings derived from other tokamaks. The power densities and density times toroidal field products are an order of magnitude larger than in other tokamaks, in the range of values expected for ITER. The L - H and H - L transitions occur at approximately the same edge electron temperature. A low density limit to the H-mode is found at about . A high midplane neutral pressure limit of about 0.6 mTorr is also observed.

The TCV tokamak has obtained ohmic H-modes in virtually all diverted plasmas with the ion drift directed towards an X-point and in several elongated limiter plasmas. Troyon factors up to 2 and line average densities up to , corresponding to Ithe Greenwald limit, have been obtained in diverted ELM-free H-modes. Quasi-stationary H-modes lasting for the entire current flat top (1.5 s) have been obtained in the presence of regular ELMs. The occurrence and magnitude of ELMs have been found to depend on configurational parameters such as the position of the `active' X-point in unbalanced double-null discharges and the plasma - wall separation in single-null discharges. These dependencies have permitted active control of ELM behaviour in TCV. A continuous spectrum of ELM amplitudes and frequencies has been observed, ranging from clearly identifiable type III ELMs to large, low-frequency ELMs which expel up to 12% of the stored energy and up to 7% of the particle content and are reminiscent of type I ELMs. A previously unknown, benign kind of ELM, with a maximum amplitude in the divertor region, has also been observed.

This paper presents an overview of recent experimental progress towards understanding H-mode transition physics and scaling. Terminology and techniques for studying H-mode are reviewed and discussed. The model of shear flow stabilization of edge fluctuations at the L - H transition is gaining wide acceptance and is further supported by observations of edge rotation on a number of new devices. Observations of poloidal asymmetries of edge fluctuations and dephasing of density and potential fluctuations after the transition pose interesting challenges for understanding H-mode physics. Dedicated scans to determine the scaling of the power threshold have now been performed on many machines. A clear dependence is universally observed but dependence on the line averaged density is complicated. Other dependences are also reported. Studies of the effect of neutrals and error fields on the power threshold are under investigation. The ITER threshold database has matured and offers guidance to the power threshold scaling issues relevant to next-step devices.

H-mode operation presents challenges for ITER and for reactor application in the areas of performance, accessibility, sustainability and divertor compatibility. Some issues are more severe for ITER, while others are more problematic for an economic fusion reactor producing 3 MW of neutron flux to its walls. The problems that are more severe for ITER include the power requirements for the H-mode, and confinement uncertainties. Problems which are more severe for a reactor include capability for high operation, the role of the Greenwald density limit, and divertor compatibility. Within the present ITER concept, it would be very helpful to have a well defined upgrade path for the auxiliary heating systems, including options for strong profile control. The world fusion research programme should maintain a major focus on improving the reactor compatibility of the H-mode, with an eye to the special advantages of H-mode plasmas with strong shaping.

Owing to the recent progress of `profile controls' utilizing the freedom of profiles of plasma parameters, various types of combined (advanced) H-modes have been obtained; PEP H-mode, high- H-mode, high- H-mode, high- H-mode, reversed shear H-mode, CH-mode, VH-mode, etc. These combined H-modes with high potential for confinement and stability are characterized by improved transport in the core region in addition to the edge or deep inward penetration of the edge confinement pedestal. Such additional improvements seem to be related to flow shear, magnetic shear, safety factor, density profile or . The improved confinement region can propagate from the core to the edge or from the edge to the core, or appear almost simultaneously over the whole radius. However, there are some critical issues to be solved concerning reactor conditions, that is, under electron heating at high edge density with small impurity accumulation in the steady-state. Most of the NB heated combined H-modes have been obtained with ion heating, where ion thermal diffusivity and particle diffusivity can be reduced to the neoclassical level. However electron thermal diffusivity has so far not been clearly reduced except in the H-mode and in the VH-mode. Therefore, improvement of electron transport is one of the main issues. The second issue is that some improved modes are accompanied by strong density peaking which may result in large impurity accumulation. The third issue is the difficulty in achieving the improved modes at a high edge density with the high particle recycling essential for a dense and cold divertor. The fourth issue is the stability at high in the steady state. It is not certain that pressure and current profiles including the bootstrap current can be sustained stably. The requirement of ELMs for heat and particle exhaust is also an important issue in stability design as concerns the extent of the second regime access for the high-n ballooning mode.

The present status of investigations concerning active control of the H-mode is discussed, including density control for steady H-mode operation, control of heat flux to lower the divertor heat load, and control of the transport barrier for further improvement of confinement and MHD stability. Control of ELM activity is extremely important for density control in the H-mode. Some examples of controlling ELMs are discussed. The remote radiative cooling of the main plasma and the divertor plasma is necessary to lower the divertor heat load. Since the heat flux across the separatrix to keep the H-mode has to be higher than the threshold power for the HL transition, the feasibility of cooling the H-mode plasma with the main radiation loss depends on the ratio of the threshold power to the total heating power. Control of the edge transport barrier with plasma shaping has been demonstrated. The toroidal field dependence of the threshold power for an internal transport barrier (ITB) formation of the high- mode is significantly different from that for an edge transport barrier formation in the H-mode, indicating the transition physics might be different. Formation of an ITB was successfully demonstrated by ion Bernstein wave resonant heating and by negative magnetic shear.

Recent developments in H-mode theory are discussed with earlier work described to put new theories in context. Much of the recent work concerns the development of the radial electric field near the plasma edge and its impact on transport driven by fluctuations, and is the main topic discussed.

In this paper, various types of so-called edge localized mode (ELM), an MHD instability occurring in the edge of H-mode plasmas, are discussed. A phenomenological classification based on experimental observations from several experiments is given. ELMs affect the confinement of both energy and particles, with a stronger degradation of particle confinement. Thus, ELMs provide particle control in the H-mode, allowing steady-state operation at only modest confinement degradation. However, large ELMs may lead to unacceptable power loads on the divertor plates of future large fusion experiments and, thus, active control of ELM type and characteristics is required. At present, three control schemes have been demonstrated, namely dynamic variation of the plasma shape, application of resonant magnetic perturbations and control of the energy flux through the separatrix by additional radiation.

H-mode operation is projected to be needed in ITER in order to achieve its objectives of sustained burn for 1000 s with 1.5 GW of fusion power. However, experiments have shown that a minimum level of input power is required to access and sustain the H-mode regime. This paper studies the impact on ITER operation of the presently available scalings of this minimum input or threshold power for both the direct L- to H- and reverse H- to L-mode transitions. We present scenarios for access into the H-mode, for maintaining the plasma in H-mode and for the fusion power shutdown with the reverse transition into L-mode. The commonly used empirical scaling for the L- to H-mode transition with a reduction of a factor of two for the reverse transition has severe consequences on ITER operation and performance and various methods to quantify this impact are presented. We conclude, however, that the ITER mission and physics objectives remain compatible with such a scaling.

Previous scaling results indicate that the H-mode power threshold increases nearly linearly with the line-averaged density, , and the toroidal field, . The power threshold was measured in similar, ITER-like, discharges in JET and DIII-D, at the same and in order to determine the size scaling of the power threshold. The results indicate a size scaling proportional to the surface area, , which is weaker than the linear surface area dependence previously assumed.

The results of a series of ITER simulation experiments on JET are described. A series of H-mode threshold experiments are shown to reproduce one of the standard power threshold scaling expressions that is being used to predict the power threshold in ITER. Then, from a series of experiments in which the Larmor radius scaling of ELMy H-modes is examined, it is concluded that the scaling of the confinement is gyro-Bohm-like provided the power levels are well above the threshold. Finally, we show that at high there is a dramatic reduction in the confinement at .

Experiments have been carried out on the DIII - D tokamak to investigate whether off-axis NBI can: (a) drive significant perpendicular flow to lead to increased suppression of turbulence and improved confinement, and (b) be used to control the radial electric field profile. Measurements of both impurity ion poloidal and toroidal rotation profiles were made using charge exchange recombination spectroscopy. These experiments used a low current, low elongation ( MA, ) plasma whose magnetic axis was shifted 36 cm vertically upward from the vessel midplane and then shifted downward to be centred on the midplane later in the discharge. 10.7 MW of beam power was applied to maximize the NBI effect whilst operating at low target densities and high temperature to minimize poloidal damping. Results from these experiments show a slight increase in impurity ion poloidal rotation velocity during the vertical shifted phase of off-axis NBI discharge. The toroidal rotation profile is more peaked during off-axis NBI. Both these effects lead to a change in the contribution to the radial electric field during off-axis NBI.

Measurements of local electron density , electron temperature and ion temperature have been made at the very edge of the plasma just prior to the transition into H-mode for four different single parameter scans in the DIII-D tokamak. The means and standard deviations of , and under these conditions for a value of the normalized toroidal flux of 0.98 are, respectively, , keV, and keV. The threshold condition for the transition is more sensitive to temperature than to density. The data indicate that the dependence is not as simple as a requirement for a fixed value of the ion collisionality.

A pair of probe pins whose lengths differ by 5 mm are inserted into the separatrix of H-mode plasmas. When the long pin is biased positively against the short one, the H - L transition takes place and the current through the resistor making a connection between these pins increases for 10 and the potential of the pin increases for 100 . The energy dissipated by the resistor during the H - L transition is comparable to the energy of the radial electric field in H-mode plasma. When the long pin is biased negatively, the H - L transition also occurs; however, it takes 1 ms to change the potential of the pin and the plasma returns to H-mode after the extraction of the probe.

Measurements of edge impurity ion parameters have been made in JET neutral beam heated H-modes, using charge exchange spectroscopy. In general there is no evidence of any change in the impurity poloidal velocity at the L to H transition (to within 5 km in the best measurements). The upper limit of the changes in at transition due to changes in poloidal flow is significantly smaller than those seen in smaller devices. Evidence is seen of the rapid formation of a particle transport barrier at the plasma edge and this leads to a modification of the density profile (and also therefore the profile) at the separatrix. In established H-modes an extended region of step density gradient and a gradient in toroidal rotation give a negative well-like structure to .

The temporal behaviour of turbulence associated with the L - H transition is described. In the ASDEX tokamak the level of density fluctuations in a narrow region inside the separatrix and in the scrape-off layer is quenched on a 100 timescale. After the L - H transition fluctuations of a different nature are destabilized which, however, show no marked influence on transport across the barrier. The time development of magnetic turbulence is closely correlated with density turbulence. A comparison is made with edge turbulence in the W7-AS stellarator H-mode where edge turbulence during the transition evolves on a similar timescale.

Nonlinear time series analysis and wavelet bicoherence are used to characterize the changes to microturbulence in the radial electric field shear layer across the L to H transition in DIII-D. Time series of fluctuating floating potential and ion saturation current from the reciprocating probe array have been analysed for skewness and kurtosis as indications of intermittency in the turbulence. Provided the time series are carefully selected for stationarity, the skewness S and kurtosis K are consistent with Gaussian probability distributions in both L- and H-modes. Wavelet bicoherence analysis reveals the presence of intermittency in the coupling due to sum frequencies above 400 kHz in both the L- and H-modes. The intermittency is most pronounced in the H-mode, with the intervals of strong coupling correlated with changes in local edge plasma conditions. These results appear consistent with predictions from self-organized criticality models.

The ITER H-mode threshold database, which currently includes data from nine divertor tokamaks is presented. The results obtained for single devices and for the combined data set are given. The similarities and differences between the different devices are shown. Possible expressions for the H-mode threshold in agreement with the data are given and the extrapolation to ITER is discussed.

The following scaling of threshold power for the H-mode transition is obtained for ; where is around unity for and for and . The threshold power increases with decreasing density in the range . For there is no dependence of . However, for the threshold power increases with increasing .

The H-mode initiated by the control of the rotational transform profile in CHS exhibits a very rapid transition (in 100 s or less) in NBI heated plasmas. There is no obvious difference in transition behaviour between hydrogen and deuterium. Langmuir probe measurement has revealed that the rapid change in floating potential occurs at the transition, but the change follows the formation of an edge transport barrier. The presence of an surface just inside the last closed flux surface and a sawtooth crash triggered by internal modes play an important role in the H-mode transition of CHS.

We have experimentally studied the effect of neutrals on the H-mode transition in JT-60U. The neutral density near the separatrix was evaluated with the DEGAS code. The effective edge ion collisionality right before the H-mode transition tends to decrease with the increase of . L- and H-mode phases are clearly separated in the - plane. This result suggests that the edge neutrals have a substantial influence on the condition of the H-mode transition.

Test particle radial flux surface excursions are reduced during the H-mode. Particle transport is reduced by a factor of 10 in the H-mode, but energy confinement increases are small. In the H-mode the evolution of poloidally resolved turbulent statistics are not explained by published theory. Turbulent momentum transport leads to a concentration of poloidal momentum within the transport barrier, and compressibility leads to poloidal shock-like phenomena. The electron distribution functions may be modified by this shock, leading to kinetic instabilities. A physics-based understanding of the H-mode must therefore include toroidal effects combined with an adequate treatment of particle orbits, plasma compressibility and associated kinetic effects, and at least a two-species model of turbulent transport. The results suggest that rapid poloidal core-plasma rotation could form core transport barriers without reliance on fluid shear or reversed magnetic shear effects.

The potential formation near the edge, X-points, and on the divertor footprints are studied using Langmuir probe arrays with respect to the density pump-out due to ECRH. It is found that a poloidal electric field is generated in the interior region, and this is considered to cause the enhanced particle flux across the LCFS. A poloidal structure of the edge potential distribution seems to be an m = 2 type and is discussed based on the orbit topology of warm electrons in the l = 2 heliotron configuration.

Results of experiments performed on DIII-D since the last H-mode workshop impose a number of constraints on theories. First, measurements demonstrate that both the main ion pressure gradient, , and poloidal rotation, , are important in determining in the H-mode, with the term being the trigger for the transition. Accordingly, theories must include both and effects. Since the term appears to be the trigger, the physics of the power threshold must involve the physics of the perpendicular rotation. Second, the time sequence of changes in and the density and potential fluctuations are consistent with shear causing the change in the fluctuations. Third, the sharp onset of the fluctuation suppression during this same time sequence strongly suggests a critical shear for fluctuation suppression. Fourth, Langmuir probe measurements show the need for a turbulence stabilization theory including multiple fluctuating fields and their relative phases. Finally, theories based on predicted by standard neoclassical theory disagree with experiment.

The mechanism of L - H transition and transport barrier formation is studied by the use of a one-dimensional transport code including plasma rotation and radial electric field. Particle transport simulation with a fixed temperature profile has indicated that the density profile is quite sensitive to the edge plasma temperature. When the edge temperature exceeds a critical value of about 100 eV, the radial electric field is built up near the plasma edge and the rotation shear reduces the particle transport. The large density gradient due to the reduction of transport enhances the bootstrap current which also contributes to the reduction through less magnetic shear.

The high-beta poloidal discharge number 17110 in JT-60U that develops an internal transport barrier is analysed for the transport of ion energy and momentum. First the classical ion temperature gradient stability properties are calculated. Then the evolving toroidal and poloidal velocity profiles reported by Koide et al (1994 Phys. Rev. Lett. 72 3662) are used to show how the sheared flows control the stability and transport. Coupled momentum-energy transport equations predict the creation of a transport barrier.

Dynamics of the L - H and H - L transitions are addressed numerically. The focus is on the power level slightly above the threshold. Therefore, at the edge the electric field is assumed to be neoclassical and the toroidal rotation to be damped by anomalous viscosity and inertia. The diffusion coefficient depends on the shear of the electric field. For a given set of boundary conditions, the phenomenon of `dithering' emerges. In contrast, if the power on the separatrix significantly exceeds the threshold the L - H transition occurs directly, without falling into the dithering regime.

H-mode theory (Shaing and Crume Jr 1989 Phys. Rev. Lett. 63 2369) is extended to include realistic effects such as charge-exchange momentum loss and the magnetic stresses. Local critical plasma parameters for the onset of the bifurcation are obtained and can be compared quantitatively with experimental measurements. It is concluded that local critical parameters are more meaningful in quantifying H-mode transition than the global power threshold scaling law.

To study L/H transition physics, the hysteresis, generated by the ambipolar condition, is examined. Three mechanisms for the bipolar losses, i.e. the loss cone loss, collisional bulk viscosity loss of ions, and the anomalous loss, are simultaneously retained, and the competition between them is investigated. It is found that fivefold multiple branches exist in the gradient - flux relation, when the effective ion collision frequency is close to unity. Multiple bifurcations in the hysteresis curve appear. The new type of the limit cycle oscillation, compound dither, is predicted to occur.

As a model example of the effect of shear flow on the cross phase between electrostatic potential and pressure fluctuations, a nonlinear theory of resistive pressure gradient driven turbulence (RPGDT) in a shear flow is presented. This work builds on numerical studies of RPGDT, which have shown that both flow shear and curvature can affect the cross phase as well as the fluctuation levels. In this work, we show that the effect of shear flow on transport can be expressed through the temporal response of pressure to potential. It is shown heuristically that even in the case where the fluctuation levels are not modified, the flow shear still acts to reduce the phase angle between potential and pressure fluctuations, thereby suppressing transport. The scaling of the cross phase with flow shear and flow curvature is presented.

The theory of the fluctuation-driven radial electric field in toroidal geometry is presented. It is shown that the gyro-viscosity as well as the poloidal pressure gradient and parallel inertia provide significant contributions to the generation of the radial electric field. The neoclassical parallel viscosity modified by the inertia terms and fluctuation-driven transport is evaluated. The equations for the rotation of a turbulent tokamak plasma are derived.

Circular limiter DD and DT H-modes have been obtained through transitions in supershot and high-poloidal-beta plasmas in TFTR. High tritium concentrations have been realized, with tritium fuelling mostly through the heating beams, varying from all to all beams. Density fluctuations were reduced by a factor of two across the plasma and and profiles have features consistent with simultaneous transport barriers in the plasma core and edge. values from TRANSP decrease by a factor of 2 to 5 in these regions at the transition and remain low until the onset of ELMs.

A description of the various ELM types observed in single null divertor plasmas in COMPASS-D is given, including type I and type III ELMs in both ECRH and ohmically heated plasmas. Precursor oscillations to both type I and type III ELMs have been observed. The large variation in the precursor mode structure is discussed, as well as models of the effect on the plasma.

A model of the giant ELMs (type-I ELM) is developed. Theory of the self-sustained turbulence of the current-diffusive ballooning mode (CDBM) is developed, and multifold states for the L-mode, H-mode and the third state with magnetic braiding, M-mode, are obtained. The transition to M-mode occurs if the pressure gradient reaches a critical value. The nonlinear excitation of the magnetic perturbation takes place, the growth time of which is of the order of the poloidal-Alfvén transit time, followed by catastrophic enhancement of the transport coefficient. Avalanche of the transport catastrophe is also analysed, showing a very rapid radial propagation velocity. The M-state terminates if the pressure gradient becomes small, leading to the back-transition to the H-(L-)mode. Under the constant power supply, these processes can repeat themselves, causing periodic bursts. The period becomes shorter as the average power flux increases.

Large ELM events in TFTR are often accompanied by short, intense bursts of electron cyclotron emission. A unique combination of two fast grating polychromator instruments located at different toroidal positions is used to measure the emission and characterize these bursts, which exhibit strong toroidal asymmetries. Bursts are compared to those which occur early in the thermal quench phase of high beta disruptions, the only other time that bursts occur with comparable robustness. In these disruptions, bursts are localized to the vicinity of the ballooning mode, a medium toroidal mode number (n = 10 - 20) precursor, localized toroidally, poloidally, and radially, which triggers the disruption. Rapid loss of particles and sudden rapid cooling occurs. In both cases, bursting can be explained not in terms of excitation of enhanced emission but rather in the reduction of absorption of thermal emission. Bursting is consistent with a modification of the electron distribution function due to a rapid energy or particle exchange between hot electrons and cold electrons from the edge, momentarily reducing the velocity gradient of in the thermal region. This model is qualitatively and quantitatively consistent with experiment, predicting for example emission enhancement factors of .

We have studied the evolution of the edge plasma in VH-mode discharges in DIII-D, by following the changes in the radial profiles of the density and temperature, in the core plasma near the separatrix and in the scrape-off-layer (SOL) plasma outside the separatrix. The electron density and temperature profiles in the SOL do not show any significant difference between the ELM-free H-mode and VH-mode phases of the discharge. In the ELMing phase, the density profile broadens during an ELM, forming a high-density plateau that extends out into the SOL to the limit of the measurement. This plateau persists between the ELMs, although the density in the SOL does relax somewhat between the ELMs, with a characteristic time that can be greater than 10 ms, much longer than the sonic particle flow time to the divertor plates. The density scale length increases with the ELM background, as measured by the photodiode viewing the divertor floor nearest to, but outside, the outer strike point. The electron temperature profile in the SOL also broadens during an ELM, but the broad profile does not persist between ELMs.

For the standard shape of JT-60U at low triangularity , the onset condition for giant ELMs is clearly correlated with the high-n ballooning limit in the first stability regime over wide ranges of plasma parameters (, ...). The limit of the normalized edge pressure gradient (-parameter) increases with elongation (1.5 - 1.8) and internal inductance . Recently, a new connection of the poloidal field coils enabled us to scan triangularity from the original value of up to (currently limited at MA). The -scan showed that the limit of edge density, edge pressure and the -parameter increase with . When both and are high (, ), minute-grassy ELMs appear. The edge -parameter during the minute-grassy ELMs can be higher than that for the onset of giant ELMs.

This paper presents a linear analysis on an instability that is driven by the electron temperature gradient and is localized by the density pedestal at the edge. We find that the instability appears only in H-mode plasmas and can survive the observed levels of sheared plasma rotation because of its narrow radial width. For realistic plasma profiles, the mode frequency, growth time, poloidal mode number, and direction of rotation are all comparable to those of the precursors for Type III (small) edge-localized modes (ELMs) observed in DIII-D and ASDEX. We conclude that the instability is a promising candidate for the precursors.

A dynamical model of edge-localized mode (ELM) phenomena is presented. It combines the paradigms of the L - H transition mediated by electric field shear induced suppression and generation with a simplified description of MHD instability and with a transport analysis of the plasma edge. In the parameter regime characteristic of an H-mode plasma, the model exhibits a transition to stationary relaxation oscillations (i.e. stable limit cycle behaviour) corresponding to ELMs. The dependence of ELM frequency, amplitude etc on the heating power and other control parameters is studied.

An addition of ion Bernstein wave power of only a tenth of the neutral beam power into an H-mode causes the formation of a core transport barrier. At the location of the barrier, reduced ELM losses in the soft x-ray profile are observed. During the IBW-modified ELM, one can observe enhanced fluctuations in the soft x-ray fluctuation profile over the whole of the observed frequency range. In the region of the core transport barrier, the enhancement of fluctuations seems to be strongly reduced. The ELM loss propagates to the core with a velocity greater than that which would be consistent with a normal energy diffusion process.

Characteristics of the various types of edge localized modes (ELMs) observed during the H-mode of the ASDEX Upgrade tokamak are presented. Magnetic high-mode-number ELM precursor activity is found for type I ELMs and type III ELMs, the former only during neutral beam injection in the counter direction. The effect of ELMs on electron temperature and density profiles is discussed.

The nonlinear interchange mode shows an intermittent oscillation and generates a zonal counter-streaming flow when transport due to vortex flow is suppressed. A model of the ELM (edge localized mode) observed in the H- (high-confinement) mode plasma is discussed based on the period of intermittent oscillation depending upon the Rayleigh number.

The effect of ELMs on confinement in ohmic H-modes has been investigated by analysing the rates of change of global parameters such as the electron content and the stored energy as a function of ELM frequency. In double-null H-modes, ELMs are found to expel on average 2% of the electron content and 2.5% of the stored energy. In single-null discharges which are separated by more than 2.2 cm from the inner wall, larger ELMs are observed which expel on average 3 - 7% of the electron content and 3 - 12% of the stored energy. When the plasma - wall distance is reduced, ELM frequencies increase and ELM amplitudes decrease. Quasi-stationary H-modes are obtained for Hz in double-null plasmas and for in the range 50 - 300 Hz in single-null plasmas, depending upon the plasma - wall distance and the ELM amplitude. The reduction in particle confinement due to the ELMs is sufficient to bring the ratio of particle to energy confinement times below 8 - 12 depending upon the ELM size and frequency.

An empirical model for anomalous transport which contains both Bohm and gyroBohm terms is proposed and tested on the set of JET discharges which include L-, H- and hot-ion mode.

The importance of the flow shear in various enhanced confinement regimes is discussed in terms of the turbulence suppression criterion in general toroidal geometry. This criterion is then further generalized to include the poloidal angle dependence of the equilibrium electrostatic potential. The implication of the recently observed in - out asymmetry in the fluctuation behaviour in DIII-D VH-mode is discussed.

The effect of radial electric field and velocity shear on thermal transport is studied in Heliotron-E. When neutral beams are injected into the target plasma produced by electron cyclotron heating (high mode), or an ice pellet is injected into an NBI plasma (pellet injection mode), the central ion temperature increases in time up to 0.7 - 0.8 keV. These high plasmas are characterized by a peaked ion temperature profile and are associated with a peaked electron density profile produced by neutral beam fuelling with low wall recycling, or by pellet fuelling. The global energy confinement is improved compared with L-mode plasmas by a factor of 1.4 for the same line-averaged electron density. The observed improvement in ion heat transport is related to the radial electric field shear rather than to the rotation velocity shear in the bulk plasma.

Negative central magnetic shear (NCS) discharges with and up to 80% of the current non-inductively driven are reproducibly produced in the DIII-D tokamak. Strong peaking of , plasma rotation and, in some cases, are observed inside the NCS region. Transport analysis shows that the core ion thermal diffusivity is substantially reduced and near the neoclassical value after the formation of the internal transport barrier. The negative central shear is necessary but not sufficient for the formation of this transport barrier. The power required for the formation appears to increase with the toroidal magnetic field. The high performance phase of H-mode NCS discharges often ends with an ELM-like collapse initiated from the edge whereas the L-mode discharges which have a more peaked pressure profile tend to end with a more global n = 1 MHD event.

Experimental results obtained in PBX-M suggest that IBW can be used to induce locally a transport barrier and improve core confinement. Using similar neoclassical-like particle and ion energy diffusivity values as inferred inside the IBW-induced core barrier, the CH-mode performance is assessed for TFTR high performance D - T plasmas. By controlling the barrier location, Q-factors in the range can be obtained with a total stored energy of MJ.

The high performance H-mode regime on DIII-D has been extended to both low q and high q (high ) operation. In high current operation, VH-mode discharges were obtained for the first time with . These discharges had , H = 2.9, , and %-sec. was improved by approximately 50% over previous results. These discharges were obtained with neutral beam injection during the plasma current ramp up which maintained the axial q above 1. In low current operation, neutral beam heated discharges with 100% of the plasma current from non-inductive sources were obtained at high , with , H = 3.1 and . These discharges represent an extension of the high performance regime to , which was made possible by reduction in the locked mode low density limit, as the result of improvements in the magnetic field error correcting coils. These low current discharges do not exhibit some of the standard signatures of VH-mode, but appear to represent a new regime of improved H-mode confinement. Similar, non-VH-mode, high energy confinement discharges were obtained at low density and moderate q.

The time evolution of confinement and transport properties of ICC H-mode plasmas in JT-60U has been studied. We found the change of confinement character on a timescale much smaller than the thermal energy confinement time, , at the L - H and H - L transitions. A slow change on a timescale of is also found during the H-mode phase.

An operating mode with a very high confinement core like the VH-mode but a very low power flow to the divertor plates and low edge particle confinement like an L-mode would be beneficial. For a large tokamak like the proposed ITER, the power density at the separatrix is not that far above the scaled H-mode power threshold so not much of the power can be radiated inside the separatrix without causing a return to L-mode. The thicker scrape-off layer of an L-mode increases the radiating volume of the scrape-off layer and helps shield impurities from the core. This is especially important if the first wall is metallic. In this paper an H-mode transport model based on velocity shear suppression of turbulence will be used to show that it is possible to have a strongly radiating mantle near the separatrix, which keeps the edge in L-mode, while having a VH-mode core with a broad region of suppressed turbulence. The existing results of enhanced L-mode confinement during impurity injection on a number of tokamaks will be surveyed. We will identify the operating conditions which will most likely result in further improvement of the core confinement by control of the heating, fuelling, and torque profiles.

The tokamak database indicates a trend that the electron heat conductivity in H-mode plasmas is less variable than the ion heat conductivity. This can be explained theoretically if we assume that the ion channel is laminarized on magnetic surfaces due to poloidal rotation but that the parallel electron dynamics remains turbulent due to asymmetries along the flux tube. Core H-mode density asymmetry data indicate that a poloidal shock is present with . This type of magnetic configuration leads to persistent anomalous electron heat transport as indicated earlier [1]. The magnetically trapped particle turbulence may be masked due to the existence of the above poloidal asymmetry along the flux tubes. In L-mode these flux tubes further deteriorate confinement due to associated electric braiding which increases the radial correlation scale. An H-mode regime electron heat conductivity formula is given for a flux tube turbulence limited plasma channel. Thus an H-mode factor may be computed for tokamaks relative to popular L-mode scaling laws.

In this paper we examine the evolution of the gradients in toroidal velocity, , and ion temperature, , during the CH-mode in PBX-M. This enhanced confinement mode develops from the H-mode following the application of IBW power and is characterized by strong density peaking and the formation of significant gradients in both toroidal velocity and ion temperature. The radial profiles of both the toroidal velocity and the ion temperature are obtained from the PBX-M charge exchange recombination spectroscopy (CHERS) diagnostic. A significant IBW power threshold is observed in the formation of the gradients. A power step of only 25 kW - from 200 to 225 kW - marks the onset. Details of the evolution of the gradients are given and, in particular, the variation with IBW power is described.

A steady-state ELMy H-mode profile data set of 68 DIII-D discharges and 74 JET discharges is fitted with an error of 7 - 8%. The advantages of a parametrization of the plasma profiles in terms of a semi-parametric representation, , are described. The shape of the temperature profile depends almost exclusively upon the size, R and , with a secondary dependence on the heating power. The density profile depends primarily upon with a secondary dependence on . The line-average temperature scales as instead of . The predicted ITER temperature is keV.

The idea of controlling the radial electric field by means of a radial current resulting from ion orbit loss caused by counter neutral beam injection has been theoretically and experimentally investigated. A large fraction (%) of the 75 keV deuterium ions counter-injected into a low- plasma ( MA) suffers prompt orbit loss, which forces an inward ion current to maintain charge neutrality. Monte Carlo guiding-centre orbit calculations predict a radial current of 80 A at the last closed flux surface. In these discharges, is negative everywhere, owing to the counter-going toroidal rotation, and exhibits a double-bump shape, in contrast to the usual positive parabolic shape for the co-injection case. The measured carbon impurity ion toroidal rotation profile shows a pedestal over the outer region where fast ions are lost, possibly due to the effect of torque. The momentum diffusion process tends to slow down and to spatially spread the torque effect. The L - H transition did not occur more quickly in these discharges than in similar co-injected discharges.

Experiments that decoupled H-mode plasma current, density and temperature have been performed on the DIII-D tokamak by the application of divertor cryopumping. Deuterium ELMing H-mode steady state discharges were operated in the single null configuration with neutral beam heating. A power-law dependence of the ELMing thermal confinement was assumed with the result that the thermal energy confinement depends weakly on density and strongly on plasma current. A local power balance analysis generally found that the ion and electron diffusivity was unchanged with an approximate factor of two change in density at constant temperature. In contrast, both the electron and ion diffusivities increased with increasing temperature at constant density. These plasmas were simulated with the Rebut - Lallia - Watkins critical temperature gradient model. Our results indicate that the temperature dependence of the model does not fit the DIII-D data; the model was too optimistic at the highest power. However, the density dependence of the model agrees well with DIII-D data.

ITER or any other future fusion device will operate in H-mode to take advantage of the improved confinement. Stationary operation in H-mode, however, requires the occurrence of ELMs, and the resulting power load on the divertor target plates is not acceptable from the engineering point of view. Beyond this, the present divertor concepts rely on a cold and dense (possibly detached) divertor plasma. Large type-I ELMs, however, carry too large a power and possibly burn through such cold divertor plasmas.

The completely detached H-mode (CDH-mode), recently established in ASDEX Upgrade shows detachment at both divertor target plates and small ELMs (so-called type-III ELMs), that do not burn through the divertor plasma. The extrapolation of this scenario to ITER, however, is an important question.

An extensive experimental investigation of reactor-relevant features of H-mode plasmas has been carried out using the JET-pumped divertor. Steady-state H-modes have been established with high confinement and with durations of up to 20 s. At the highest densities, these plasmas return to the L-mode when the radiated power fraction is %. Detached H-modes have been produced by gas-puffing deuterium plus nitrogen, and their properties studied at total input powers of up to 32 MW. Helium transport and exhaust have also been investigated, and a systematic comparison of H-modes using carbon fibre composite and beryllium divertor targets has been carried out.

8 - 18 August 1996 International Summer School on Plasma Physics and Technology La Jolla, CA, USA Contact: Mr V Stefan, Institute for Advanced Physics Studies, PO Box 2964, La Jolla, CA 92038, USA. Tel +1-619-456-5737.

26 - 30 August 1996 Joint Varenna - Lausanne International Workshop on Theory of Fusion Plasmas Villa Monastero, Varenna, Italy Contact: Centro di Cultura Villa Monastero, 1 Piazza Venini, 22050 Varenna (Lecco), Italy. Tel +39-341-831261, Fax +39-341-831281. Application and abstract deadline: 15 June 1996.

2 - 5 September 1996 EU - US Workshop on Transport in Fusion Plasmas Villa Monastero, Varenna, Italy Further information: G Gorini, ISPP, 16 Via Celoria, I-20133 Milano, Italy. Tel +39-2-2392637, Fax +39-2-2392205, E-mail ggorini@mi.infn.it. Administrative contact: Centro di Cultura Villa Monastero, 1 Piazza Venini, 22050 Varenna (Lecco), Italy. Tel +39-341-831261, Fax +39-341-831281. Application and abstract deadline: 15 June 1996.

9 - 13 September 1996 International Conference on Plasma Physics Nagoya, Japan Contact: Conference Secretariat, c/o Prof. Hiromu Momota, National Institute for Fusion Science, Nagoya 464-01, Japan. Tel +81-52-789-4260, Fax +81-52-789-1037, E-mail icpp96@nifs.ac.jp. Abstract deadline: 31 March 1996.

16 - 20 September 1996 19th Symposium on Fusion Technology Lisbon, Portugal Contact: Professor Carlos Varandas, Centro de Fusão Nuclear, 1096 Lisboa Codex, Portugal. Fax +351-1-8417819, E-mail cvarandas@cfn.ist.utl.pt. General information will be available via WWW with URL http://www.cfn.ist.utl.pt.

25 - 29 September 1996 Summer University of Plasma Physics Garching, Germany Contact: Ms Ch Stahlberg, Max-Planck-Institut für PlasmaPhysik, Boltzmannstr 2, D-85748 Garching, Germany. Tel +49-89-3299-2232, Fax +49-89-3299-1001.

11 - 15 November 1996 38th Annual Meeting of the Division of Plasma Physics, APS Denver, CO, USA Contact: Dr Richard Hazeltine, University of Texas, Institute for Fusion Studies, RLM 11.314, Austin, TX 78712. Tel +1-512-471-1322, E-mail stewart@hagar.ph.utexas.edu.

17 - 18 February 1997 Plasma '97: 21st Australian Institute of Nuclear Science and Engineering Plasma Science and Technology Conference Sydney, Australia Contact: Margaret Lanigan, Conference Manager, PMB 1, MENAI NSW 2234, Australia. Fax +61-(0)2-439-6561, E-mail ainse@ansto.gov.au.

6 - 11 April 1997 10th Joint Workshop on Electron Cyclotron Emission and Electron Cyclotron Resonance Heating Ameland, The Netherlands Contact: J Hamers-Smit, FOM - Instituut voor Plasmafysica 'Rijnhuizen', Postbus 1207, 3430 BE Nieuwegein. Tel +31-30-6096999, Fax +31-30-6031204, E-mail ec10@rijnh.nl. Application and abstract deadline: 17 January 1997.

8 - 12 September 1997 12th International Conference on Gas Discharges and their Applications Greifswald, Germany Contact: Dr G Babucke, Inst. f. Niedertemperatur-Plasmaphysik, Robert-Blum-Str. 8 - 10, 17489 Greifswald, Germany. Tel +49-3834-554411, Fax +49-3834-554301, E-mail gd97@public.inp.uni-greifswald.de.