Thermonuclear fusion reactors are one of the mid to long term solutions to transit towards a world dominated by carbon-free energy. Extreme temperatures are required for fusion reactions and the plasma of hydrogen isotopes must be magnetically confined in a torus shape. Sustaining such high level of particle and energy confinements is a key issue. Reactors are expected to operate in a high confinement regime - the H-mode - in which turbulent transport is reduced by the presence of a transport barrier in the edge plasma. This regime is observed in all current devices but remains largely miss-understood. In this thesis, we investigate several mechanisms involved in the transition towards H-mode. For that purpose, we use a range of numerical simulation tools of increasing complexity. Using simple models, we first highlight and analyze basic mechanisms likely to play a role in the on-set of transport barriers and in their impact on turbulence. Moving progressively to more complex models, we discuss the relevance of these physics in explaining experimental observations. The magnetic geometry and especially the magnetic shear are pointed out as key players.

Results from recent experiments on the DIII-D tokamak have revealed many important details on transport barriers at the plasma edge and in the plasma core. These experiments include: (a) the formation of the H-mode edge barrier directly by pellet injection; (b) the formation of a quiescent H-mode edge barrier (QH-mode) which is free from edge localized modes (ELMs), but which still exhibits good density and radiative power control; (c) the formation of multiple transport barriers, such as the quiescent double barrier (QDB) which combines a internal transport barrier with the quiescent H-mode edge barrier. Results from the pellet-induced H-mode experiments indicate that: (a) the edge temperature (electron or ion) is not a critical parameter for the formation of the H-mode barrier, (b) pellet injection leads to an increased gradient in the radial electric field, E{sub r}, at the plasma edge; (c) the experimentally determined edge parameters at barrier transition are well below the predictions of several theories on the formation of the H-mode barrier, (d) pellet injection can lower the threshold power required to form the H-mode barrier. The quiescent H-mode barrier exhibits good density control as the result of continuous magnetohydrodynamic (MHD) activity at the plasma edge called the edge harmonic oscillation (EHO). The EHO enhances the edge particle transport while maintaining a good energy transport barrier. The ability to produce multiple barriers in the QDB regime has led to long duration, high performance plasmas with [beta]{sub NH{sub 8}9} values of 7 for up to 10 times the confinement time. Density profile control in the plasma core of QDB plasmas has been demonstrated using on-axis ECH.

Edge transport barriers (ETBs) in tokamak plasmas accompany transitions from low confinement (L-mode) to high confinement (H-mode) and exhibit large density and temperature gradients in a narrow pedestal region near the last closed flux surface (LCFS). Because tokamak energy confinement depends strongly on the boundary condition imposed by the edge plasma pressure, one desires a predictive capability for the pedestal on a future tokamak. On Alcator C-Mod, significant contributions to ETB studies were made possible with edge Thomson scattering (ETS), which measures profiles of electron temperature (20 [leq] Te[eV] [leq] 800) and density (0.3 [leq] ne[10^20m^-3] [leq] 5) with 1.3-mm spatial resolution near the LCFS. Profiles of Te, ne, and pe = neTe are fitted with a parameterized function, revealing typical pedestal widths [delta] of 2-6mm, with [delta]Te [geq] [delta]ne , on average. Pedestals are examined to determine existence criteria for the enhanced D[alpha] (EDA) H-mode. A feature that distinguishes this regime is a quasi-coherent mode (QCM) near the LCFS. The presence or absence of the QCM is related to edge conditions, in particular density, temperature and safety factor q. Results are consistent with higher values of both q and collisionality [nu]* giving the EDA regime. Further evidence suggests that increased abs([nabla]pe) may favor the QCM; thus EDA may have relevance to low-[nu]* reactor regimes, should sufficient edge pressure gradient exist.

The E x B shear stabilization paradigm explains much of the phenomenology of ion thermal transport in tokamaks. Behavior in the electron channel, however, has continued to challenge our understanding. Recent experiments in DIII-D and elsewhere produce regions where electron thermal transport is almost completely eliminated with intense, localized, direct electron heating. Simulations of DIII-D discharges identify [alpha]-stabilization, local magnetic shear stabilization due to the Shafranov shift, as the dominant turbulence reduction mechanism in these experiments and may point the way toward regimes with simultaneous electron and ion internal transport barriers.

In this study, high spatial resolution Doppler backscattering measurements in JET have enabled new insights into the development of the edge Er. We observe fine-scale spatial structures in the edge Er well with a wave number kr[rho]i ≈ 0.4-0.8, consistent with stationary zonal flows, the characteristics of which vary with density. The zonal flow amplitude and wavelength both decrease with local collisionality, such that the zonal flow E x B shear increases. Above the minimum of the L-H transition power threshold dependence on density, the zonal flows are present during L mode and disappear following the H-mode transition, while below the minimum they are reduced below measurable amplitude during L mode, before the L-H transition.