Progress and Scientific Results in the TCV Tokamak

The TCV tokamak has the dual mission of supporting ITER and exploring alternative paths to a fusion reactor. Its most unique tools are a 4.5 MW electron cyclotron resonance heating system with seven real-time controllable launchers and a plasma control system with 16 independent shaping coils. Recent upgrades in temperature, density and rotation diagnostics are being followed by new turbulence and suprathermal electron diagnostics, and a new digital real-time network has been commissioned. The shape control flexibility of TCV has enabled the generation and control of the first 'snowflake' divertor, characterized by a null point in which both the poloidal field and its gradient vanish. The predicted increases in flux expansion and edge magnetic shear have been verified experimentally, and stable EC-heated snowflake ELMy H-modes have been obtained and characterized. ECCD modulation techniques have been used to study the role of the current profile in energy transport, and simulations reproduce the results robustly. The relation between impurity and electron density gradients in L-mode is explained in terms of neoclassical and turbulent drives. Studies of torqueless plasma rotation have continued, highlighting the important role of MHD and sawtooth relaxations in determining the rotation profiles. A newly predicted mechanism for turbulent momentum transport associated with up-down plasma asymmetry has been verified in TCV. Sawtooth period control, neoclassical tearing mode control and soft x-ray emission profile control have been demonstrated in TCV using the new digital control hardware, as a step on the way to more complex applications.

On benchmarking of simulations of particle transport in ITER

Federico Alberto Alfredo Felici, Javier García Hernández, Francesca Maria Poli, Olivier Sauter, Anna Teplukhina

We report results of benchmarking of core particle transport simulations by a collection of codes widely used in transport modelling of tokamak plasmas. Our analysis includes formulation of transport equations, difference between electron and ion solvers, comparison of modules of the pellet and edge gas fuelling on the ITER baseline scenario. During the first phase of benchmarking we address the particle transport effects in the stationary phase. Firstly, simulations are performed with identical sources, sinks, transport coefficients, and boundary conditions prescribed in the flattop H-mode phase. The transformation of ion particle transport equations is introduced so to directly compare their results to electron transport solvers. Secondly, the pellet fuelling models are benchmarked in various conditions to evaluate the dependency of the pellet deposition on the pellet volume, injection side, pedestal, and separatrix parameters. Thirdly, edge gas fuelling is benchmarked to assess sensitivities of source profile predictions to uncertainties in plasma conditions and detailed model assumptions. At the second phase, we address particle transport effects in the time- evolving plasma including the current ramp-up to the ramp-down phase. The ion and the electron solvers are benchmarked together. Differences between the simulation results of the solvers are investigated in terms of equilibrium, grid resolution, radial coordinate, radial grid distribution, and plasma volume evolution term. We found that the selection of the radial coordinate can yield prominent differences between the solvers mainly due to differences in the edge grid distribution. The simulations reveal that electron and ion solvers predict noticeably different density peaking for the same diffusion and pinch velocity while with the peaked profile of helium, expected in fusion reactors. The fuelling benchmarking shows that gas puffing is not efficient for core fuelling in H-modes and density control should be done by the high field side pellet injection in contrast to present machines.

2019

Étude du transport d'énergie thermique dans les plasmas du tokamak à configuration variable au moyen de chauffage électronique cyclotronique

Yann Camenen

The development of controlled thermonuclear fusion, a quasi-unlimited energy source suitable for large scale electricity production, is one of the main goals of plasma physics research. Among the directions explored to date, the use of toroidal devices called tokamaks to create and confine hot plasmas using strong magnetic fields is particularly promising. The energy used to heat the plasma must remain well confined in order to achieve plasma temperatures higher than one hundred millon degrees for a sufficiently long period to obtain numerous fusion reactions. In the frame of efficient electricity production, maximising the energy confinement is also essential to achieve the required temperature with the lowest heating power. In tokamak plasmas, energy losses are mainly due to radiation and radial energy transport from the plasma core to the edge. A significant fraction of plasma physics research is therefore dedicated to the study of radial transport in tokamaks and the exploration of new operation regimes characterised by a low transport level. The development and optimisation of diagnostics used to observe plasmas is also part of this work. This thesis work, performed on the Tokamak à Configuration Variable (TCV) in Lausanne, covers the implementation and exploitation of a multi-channel soft X-ray detector with high spatial and temporal resolution, together with the development of the tomographic inversion routines used for data analysis. The detector, comprised of two superposed wire chambers, has been tested and calibrated using an X-ray source and then installed onto the tokamak. The position of the detector was chosen such as to observe the whole plasma cross-section with maximum spatial resolution leading to high quality tomographic inversions. A mobile absorber holder was installed between the plasma and the wire chambers. The energy range of the soft X-ray emission observed by the detector was thus chosen by selecting the appropriate absorber. These various features have made possible the use of the detector for numerous studies and in particular for the spatial and temporal characterisation of the plasma internal transport barrier formation. Plasma shaping abilities covering a wide range of plasma elongations and triangularities, including negative values, are one of the strengths of the TCV tokamak. For instance, plasmas with elongated cross-sections offer higher energy confinement as well as higher plasma current and pressure limits. However, the increase of the plasma vertical instability growth rate with elongation makes the vertical control of elongated plasmas difficult, in particular if the plasma current profile is too peaked. As the current profile is usually peaked for low plasma currents, current profile broadening is required there to achieve high elongation. During this thesis, a current profile broadening method based on temperature profile modification by localised EC heating has been studied in detail. The mechanism of this method has been documented and the optimal conditions for the EC power deposition determined. Using these conditions, the TCV operational space has been extended towards higher elongation at low current. The highest elongation obtained at low current has been increased by over 25% permitting the exploration of the plasma transport properties in this regime. The flexibility of the TCV EC heating system has also been used to investigate radial electron heat transport in L-mode plasmas. For the first time, the normalised temperature gradient has been varied by a factor of four and its influence on electron heat transport has been separated from that of the electron temperature. Electron heat transport increases strongly with the normalised temperature gradient, for values between 6 and 10, and then becomes independent of this parameter. In addition, electron heat transport increases with increasing electron temperature, decreasing density and increasing effective charge. The electron heat transport dependence on these three parameters can be cast as a single dependence on the plasma collisionality. TCV shaping abilities have then been used to test the influence of plasma triangularity. The main variations of the level of electron heat transport are described by a decrease of the electron heat diffusivity towards negative triangularity and high collisionality. At constant collisionality, electron heat transport is two times lower at a negative triangularity of –0.4 than at a positive triangularity of +0.4. Concerning micro-instabilities, gyro-fluid and gyro-kinetic simulations indicate that TEM and ITG instabilities are at play in these plasmas. The good qualitative agreement between the observed experimental dependencies and the predictions of simulations suggests strongly that the TEM instabilities are involved in the transport of electron heat. The experimental study provides dependable scaling of the electron heat transport on plasma parameters that can now be used to test the prediction of transport simulations. New elements such as the saturation of electron heat transport at high values of the normalised temperature gradient and the decrease of electron heat transport towards negative triangularities have been demonstrated.

EPFL2006

SOLPS modelling of ELMing H-mode

Barbora Gulejová

Numerical simulation of the tokamak scrape-off layer (SOL) is an essential tool for the prediction of the conditions to be expected in future fusion reactors such as the ITER project, now under construction in Southern France. One particularly important issue regards the estimation of the expected transient power loads on plasma-facing components (PFC) due to magnetohydrodynamic plasma relaxations, known as Edge Localised Modes (ELMs). These loads are a major cause of concern for ITER owing to the very severe restrictions on PFC lifetime (especially the divertor targets) that they will impose if their amplitude is not maintained below a given size. Even though SOL plasma modelling has reached a comparatively high level of sophistication (the ITER divertor is being designed in part with complex edge plasma codes), the majority of simulations are performed for steady state conditions, necessarily excluding the description of transient events. This thesis explores the utility and validity of the fluid plasma, Monte-Carlo neutrals approach to SOL and divertor modelling in the presence of time dependent ELM phenomena. It aims to test the most complex tool of this type currently available, the fluid (B2.5)-neutral Monte-Carlo (EIRENE) code package SOLPS5, against a variety of ELM sizes in two very different tokamaks, TCV and JET. Although the SOLPS package has been the modelling tool of choice for ITER design, it has not yet been systematically used for the study of ELM transients. A key element throughout is rigorous benchmarking – seeking the best possible agreement between both experiment and simulation and between different codes for the same experiment, using as many different measurements as possible to constrain the model. Such benchmarking attempts are still, unfortunately, comparatively rare on today's machines. Fully time-dependent simulations (2-D plasma, 3-D neutrals) have been performed of four H-mode plasmas, two each on TCV and JET, covering Type III and Type I ELMs over a range of pedestal collisionality and energy expelled per ELM from ΔWELM ∼ 0.005 → 0.7 MJ. The high end of this limit corresponds to the current maximum ΔWELM which is thought to be tolerable on ITER for acceptable divertor target lifetime. The two tokamaks differ radically in size, input power and divertor geometry, but share carbon as the main PFC material. The SOLPS5 simulations have thus been performed with all carbon charge states included but do not feature activated poloidal drift terms. The approach is first to seek the closest match to experimental upstream, pedestal/SOL and downstream target profiles during the inter-ELM phase. This is achieved through the specification of radially varying perpendicular particle and heat diffusivities and/or convective radial velocity in order to account for the different transport levels in the edge and SOL regions. Poloidal variation of these transport coefficients is also applied to distinguish between main chamber SOL and divertor regions. This is important in a device like TCV with rather unconventional divertor geometry. Similar reasoning applies even more to the ELM itself, which is known to burst into the SOL in the outboard, unfavourable curvature region and is thus extremely poloidally localized. This has also been accounted for especially in the attempts to simulate the TCV ELM events. The complexity of the ELM instability continues to prevent a complete theoretical description of the evolution of transport during the event. In SOLPS5, the simplest and currently only method by which the ELM can be simulated is to increase the anomalous transport coefficients used to simulate the pre-ELM state during a brief interval corresponding to the ELM duration, such that the total energy expelled during this time is compatible with that measured experimentally. In the case of TCV Type III ELM, where reasonable upstream and downstream data are available and for which the largest number of sensitivity studies have been performed in this thesis with SOLPS, agreement is good in the pre-ELM phase and reasonable, but less satisfactory during the ELM. This ELM is a largely convective event in terms of pedestal temperature collapse and is, by virtue of its low ΔWELM, the "least kinetic" of the four events studied. Nevertheless, comparison of the SOLPS5 simulation results at the divertor target with those from dedicated Particle-in-Cell kinetic transport code calculations for the same ELM, demonstrate that kinetic effects are important and must be properly accounted for (by appropriate adjustment of kinetic coefficients in the fluid simulations). This presumably becomes even more important as the ELM size increases, but can only be tested to the extent that the appropriate experimental data is available. As a consequence, the tentative conclusion from the work presented here is that the use of SOLPS in a predictive sense for ITER would at best provide indicative results. In addition to the code-experiment benchmark, a code-code comparison has also been performed, checking SOLPS5 against published and well known time dependent Type I ELM simulations obtained with the dedicated JET code suite EDGE2D-Nimbus. A benchmark of this complexity has not previously been attempted and has been reassuringly somewhat successful, albeit with some unresolved discrepancies. A key feature of ELM boundary physics occupying much current research are the energy deposition asymmetries observed at the targets, which favour the inner target during the ELM for forward toroidal field direction and which appear to reverse when the field direction is inverted. These trends are opposite to the behaviour seen in inter-ELM phases, behaviour which is conventionally understood to result from toroidal geometry and the contribution of poloidal drift physics. Added complexity comes from magnetic geometry, a prominent feature of the TCV-JET comparisons described in this thesis, the results of which seem to influence the simulation results (which do not include drift effects). A recent development has been the suggestion that the ELM, in convecting plasma from pedestal to SOL regions, carries with it memory of the high toroidal rotation velocity known to characterise the H-mode pedestal on all devices. This hypothesis has been tested here in a preliminary manner, and for the first time in this kind of simulation, by imposing a toroidal velocity inside the magnetic separatrix in the simulations and studying the radial transport of this toroidal momentum into the SOL. Applied in the first instance to the TCV Type III ELM, the indications are that transfer of this rotation into the SOL can drive target asymmetries in the direction seen experimentally, though there are significant negative consequences for the resulting target profiles in other parameters for which a potential resolution would require protracted further study which has not been possible here.

EPFL2010