The penetration of deuterium pellets into ohmically heated TORE SUPRA plasmas has been studied for various plasma conditions (ne = 1 × 1019-6 × 1019 m3, Te = 1-4 keV) and pellet characteristics (Np = 2 × 1020-1.3 × 1021 atoms/pellet, Vp = 0.6-2.4 km/s - i.e. an increase of about 1 km/s from the velocity range covered by the available data for pellet studies). The measured penetration depths compare well with the predictions of the NGS model. A refined NGPS model is presented, in which the plasma channel radius is computed self-consistently and the heating of the neutral cloud by the 'cold' plasma sheath is taken into account explicitly. When the value of ne is sufficiently high, it fits the experimental penetration values well, and the computed matter deposition profile compares well with the measured Hα signal. The same model has been shown to fit the experimental scaling laws deduced from the JET and ASDEX data

Details of the increase of the plasma density following a pellet injection have been measured with high temporal resolution (16 μs) and the parallel expansion of ablated matter modeled with a four-fluid hydrodynamic code. The driving force of the expansion is the parallel pressure gradient, progressively balanced by the compression of the background plasma. It is shown that the ablated material experiences a strong transient poloidal motion (≊5×103 ms−1, ≊100 μs) as it expands along the field lines. This motion, which is induced by the pellet itself, results from the conservation of kinetic momentum: the convective motion in the sheet of ablated material (due to its positive potential with respect to the plasma) is compensated by a global drift of the whole magnetic surface. This model reproduces the main observations concerning the parallel propagation of the ablatant in the discharge. In particular, it shows that the stretching of the sheet of ablated material by the magnetic shear and the poloidal rotation is responsible for the homogenization of the density in a characteristic time of ≊1 ms. The plasma rotation measured immediately after a pellet injection is therefore not, in general, simply linked to the background radial electric field.

The ultimate aim of pellet ahlation studies is to predict what the plasma pro6lesarejust afterapelletinjection. Thisrequiresdescriptionofthepelletablation process, the parallel expansion of the ablatant and the fast outward motion of the deposited material since these three phenomena successively occur from the time of pelletinjectionmthemomentwhennewaxisynnnehicpmfilesarereached. Onlythe two first points have been quantitatively modelled. If the most important processes of ablation physics are identified and although current models reproduce both m e a s d penetrationsand averagedcharacterisficsofablationclouds, somedebatable points remain, mainly hearing on the drifts associated with the pellet motion and, consequently, on the effective shielding efficiency of the ionizedpart of the ablation cloud. During its pardel expansion, the ablated material experiences a strong poloidal rotation which depends on the ratio of the pellet and plasma masses and is due to the total kinetic momentum conservation on each magnetic surface. The fact that this rotation occurs on the same hescale as the outward motion suggests that bothphenomenacanbelinkedandthatacomprehensivemodelofthe whole fuelling process may emerge from considering the peller and the plasma as a unique system.