In the early twentieth century, scientists found out what process keeps the Sun burning: nuclear fusion. At that time, they almost instantly realized that controlling this process could also be a huge potential energy source on Earth. In order to make the fusion reaction (deuterium-tritium) happen, the hydrogen isotopes must be within the mutual attraction range of their nuclear forces. This requires an environment of extremely high temperature (more than 100 million K), in which all atoms are fully ionized and form a plasma. The ionization of the particles offers an opportunity to magnetically confine such a plasma. To date, this is most successfully done with the tokamak concept. A tokamak is a device in the shape of a torus, where magnetic field lines follow helical paths around the torus. In an efficient fusion reactor, the plasma must not only be contained at high temperature, but also at high density. Moreover, the energy confinement time should be sufficiently long to prevent the generated fusion energy from being released too fast. When temperature and density are increased, the plasma pressure rises. At a certain threshold, a bifurcation occurs that brings the plasma from a low- to a high-confinement state: the so-called H-mode. In the H-mode, however, the confinement suffers from an instability known as the edge localized mode (ELM). During a cyclic collapse of the plasma edge, the ELM instability ejects hot parts of the confined plasma. For large tokamaks such as ITER (which is currently being built to demonstrate the viability of nuclear fusion as an energy source) the ejected filaments could lead to intolerable power loads on the plasma facing components of the device. ELMs are a complex phenomenon and their survey has developed into an important field of contemporary nuclear fusion research. Hitherto, various types of ELMs have been classified: apart from the most common, but also most violent, type-I ELM regime, there are also conditions under which the impact of ELMs is more benign (e.g. in the small ELM regimes of type-II and/or -III). In order to predict ELM behaviour in larger tokamaks, understanding ELM physics and developing a consistent model of ELM losses is crucial. Simultaneously, intensive research is still ongoing in the areas of small or no ELM regimes and active ELM control. The work described in this thesis has been conducted at ASDEX Upgrade (AUG), a tokamak well suited to address several of the topics mentioned above. At AUG, plasmas in H-mode are routinely made and different kinds of ELM regimes can be obtained. The main tool used for the survey of ELMs is the Electron Cyclotron Emission imaging (ECEI) diagnostic, installed at AUG as a part of this project. ECEI is a diagnostic that provides a 2D measurement of the electron temperature (Te) at the high temporal and spatial resolution required for capturing ELM dynamics. The 2D capability allows following Te fluctuations in the vertical direction, in the poloidal cross section where the measurements are acquired. This was previously not possible with a standard 1D ECE system, which measures only radially along a single line of sight. For the first time, it became possible to observe a variety of Te fluctuations associated with different ELM types with the ECEI diagnostic. Describing the characteristics and dynamics of these various modes and their roles in the ELM cycle forms the major part of this work. In the type-I ELM cycle, three distinct types of Te fluctuations have been observed. First, in the last few tens of µs before the ELM crash, a mode is observed that rotates in the electron diamagnetic drift direction. During its short life-time, the poloidal mode number has been seen to increase, simultaneously with an increase of the mode’s poloidal velocity. This is followed by the actual ELM crash, i.e. where the temperature of the plasma edge rapidly decreases. During this crash phase, secondly, multiple filamentary structures are seen just outside the confined plasma. Most of these filaments rotate in the same direction as the mode observed just before the crash, although sometimes the first occurring filaments are seen to rotate in the opposite direction. A third kind of Te fluctuation that has been observed is associated with the difference between long and short ELM periods (also known as ‘slow’ and ‘fast’ ELMs). This mode is actually only found before slow ELMs and it displays a distinct poloidal asymmetry of its amplitude: the amplitude of the Te fluctuations has a minimum on the plasma mid-plane and maxima above and below. Although the presence of this mode does not cover the full delay of the slow ELM cycle compared to the fast one, it does suggest that this fluctuation regulates the plasma edge in such a way that a stable situation is prolonged a few ms longer until the Te crash comes. The type-II ELM regime is a regime of small ELMs that combines both good confinement and small energy losses. It is obtained by strong plasma shaping at high edge density and is regularly accessed at AUG. Measurements with ECEI revealed a broadband Te fluctuation, characteristic for type-II ELMs. This mode is situated just inside the top of the edge temperature barrier and completely flattens the Te profile at this location, whilst leaving the pedestal gradient unaffected. After averaging over a longer time, the 2D distribution of this fluctuation’s amplitude also shows a distinguished minimum around the mid-plane. The observation of a beat wave with a low beat frequency suggests that the broadband feature is actually caused by the coexistence of multiple modes with slightly different mode numbers or frequencies. Since the installation of a set of magnetic perturbation coils close to the plasma (during the course of this project), it has become possible at AUG to achieve conditions under which the large type-I ELMs are mitigated. The exact mechanism that causes this mitigation is not fully understood yet (especially the role of the edge electron density threshold seems critical). Although some of the features of the small ELMs (remaining after mitigation of the type-I ELMs) are shared with type-I and/or type-II ELMs, these small ELMs are clearly distinguishable from either type. Compared to type-I ELMs, the small ELMs show opposite behaviour with respect to the variation of the edge density: while the type-I ELM frequency decreases with increasing density, the frequency of the small ELMs increases. The main difference between these small ELMs and the type-II ELMs is that the latter display continuous Te fluctuations, whereas the Te fluctuations during the small ELMs are only seen for very short times. It is furthermore not straightforward to directly relate these small ELMs to the type-III ELMs observed directly after the transition to H-mode. If, despite some differences, the small ELMs are nevertheless most closely related to type-III ELMs, this could help explain the fact that sometimes suppression of type-I ELMs already occurs before activation of the magnetic perturbation coils. Although the different types of modes measured with ECEI in the various ELM regimes display features that clearly distinguish them from each other, there are also similarities that closely link the different observations together. For example, a comparable poloidal asymmetry is found in the amplitudes of both the mode that is only observed in the slow type-I ELM cycle and the broadband Te fluctuation that characterizes type-II ELMs. Moreover, whereas the one mode appears to delay the occurrence of the next Te crash (in the type-I cycle), the other mode seems to cause the complete absence of Te crashes (in the type-II case). Both modes also display equally high poloidal mode numbers. This, once more, is also a property of the small ELMs that remain after the mitigation of type-I ELMs with the magnetic perturbations. Furthermore, in both the transitions from type-I to type-II and from type-I to these latter small ELMs, it is seen that the substituting type is already present before the type-I ELMs have stopped occurring. Finally, all observed modes (the filaments excluded) rotate in the same direction, are located in the vicinity of the top of the edge temperature barrier, and consist of (multiple) coherent oscillations in the same frequency range and with medium-high to high poloidal mode numbers. These properties suggest that the observed modes are in fact (peeling-) ballooning modes, the most widely accepted theoretical model that describes the linear onset to the ELM crash.
|Qualification||Doctor of Philosophy|
|Award date||7 Nov 2012|
|Place of Publication||Eindhoven|
|Publication status||Published - 2012|