Search

A centrifuge injector that repetitively fires 1.3 mm deuterium pellets (1 torrL per pellet) at a rate of 32 pellets per second was used to build up and maintain a Doublet III 2.4 MW neutral-beam-heated limiter discharge at a line-averaged density of 1 × 1014cm?3. When compared to a conventional gas-fuelled plasma at similar density, the pellet-fuelled plasma was characterized by a factor-of-three reduction in edge neutral density and limiter recycling, a centrally peaked profile, a 70% increase in global energy confinement, and a tenfold increase in the fusion reaction rate.

Results are presented for multiple deuterium pellet injection into both the Ohmic phase and the beam heated phase of Doublet III discharges. Pellet injection experiments were carried out using an injector, built at Oak Ridge National Laboratory, which has the capability of firing up to 40 hydrogen or deuterium pellets per second at a speed of 800 m·s?1. For the deuterium pellets used in this experiment, one pellet raises the line average density to 1.3 × 1013cm?3. In the Ohmic phase, the pellets penetrate deep into the plasma, but with 2.2 MW of neutral beam heating they penetrate only half as deep. This difference is mostly due to the increased ablation in the outer 25% of the plasma during beam heating. Ablation calculations using the neutral shielding model agree well with experimental results during Ohmic heating, but they predict a shallower pellet penetration than is actually observed during neutral beam heating. About 90-100% of the pellet mass can be accounted for after injection. Electron temperature profile relaxation accompanies pellet injection, but on a time-scale that is shorter than the pellet lifetime. A transient increase in the energy confinement time ?Eis possible, provided pellet penetration is sufficient to sustain a peaked density profile. In discharges with one or more multifractured pellets, which ablate rapidly at the plasma edge and thus broaden the density profile, no increase in ?Eis seen. These results indicate that continuous ?Eimprovement may be possible if a peaked density profile can be maintained by sustained deep pellet penetration.

Solid hydrogen spheres were injected into the ORMAK tokamak as a test of pellet refuelling for tokamak fusion reactors. Pellets 70 ?m and 210 ?m in diameter were injected with speeds of 91 m/s and 100 m/s, respectively. Each of the 210-?m pellets added about 1% to the number of particles contained in the plasma. Excited neutrals, ablated from these hydrogen spheres, emitted light, which was monitored either by a photomultiplier or by a high-speed framing camera. From these light signals it was possible to measure pellet lifetimes, ablation rates, and the spatial distribution of hydrogen atoms in the ablation clouds. The average measured lifetime of the 70-?m pellets was 422 ?s, and the 210-?m spheres lasted 880 ?s under bombardment by the plasma. These lifetimes and measured ablation rates are in good agreement with a theoretical model which takes into account shielding of plasma electrons by hydrogen atoms ablated from spherical hydrogen ice.

It is shown that the ablation of a solid hydrogen pellet subject to a plasma is likely to produce a quasi-steady dense neutral gas cloud. The total integrated density of the cloud is such that the plasma electrons lose essentially all their energy in the cloud. The electron energy flux is degraded by inelastic collisions and elastic backscattering with the neutral molecules, providing local heating and acceleration of the neutral gas. Only a small fraction of the energy flux reaches the surface of the pellet, raising the pellet's surface temperature to a point where the energy flux at the pellet's surface is in balance with the energy lost through vaporization. The vaporization rate, in turn, determines the total integrated neutral gas cloud density. The scaling laws derived from the model indicate that the pellet lifetime varies as: where ?pis the lifetime of the pellet and Te, ne, and rp0are the electron temperature, density of the plasma, and initial pellet radius, respectively. A good agreement is found between this model and the ORMAK pellet injection experiment.