Search

Controlled runaway electron (RE) plateau-wall strikes with different initial impurity levels are used to study the effect of background plasma ion charge Z (resistivity) on RE-wall loss dynamics. It is found that Joule heating (magnetic to kinetic energy conversion) during the final loss does not go up monotonically with increasing Z but peaks at intermediate Z similar to 6. Joule heating and overall time scales of the RE final loss are found to be reasonably well-described by a basic 0D coupled-circuit model, with only the loss time as a free parameter. This loss time is found to be fairly well correlated with the avalanche time, possibly suggesting that the RE final loss rate is limited by the avalanche rate. First attempts at measuring total energy deposition to the vessel walls by REs during the final loss are made. At higher plasma impurity levels Z > 5, energy deposition to the wall appears to be consistent with modeling, at least within the large uncertainties of the measurement. At low impurity levels Z < 5, however, local energy deposition appears around 5-20x less than expected, suggesting that the RE energy dissipation at low Z is not fully understood. Published by AIP Publishing. This work was supported in part by the U.S. Department of Energy under Nos. DE-FG02-07ER54917, DE-FC02-04ER54698, DE-AC05-00OR22725, DE-AC52-07NA27344, and DE-AC05-06OR23100 and in part by the Spanish Direccion General de Investigacion Cientifica y Tecnica under Projects ENE2012–31753 and ENE2015-66444R (MINECO/FEDERE, UE). DIII-D data shown in this paper can be obtained in digital format by following the links at https://fusion.gat.com/global/D3D_DMP Publicado

Small (2–3 mm, 0.9–2 Pa · m3) argon pellets are used in the DIII-D tokamak to cause rapid shutdown (disruption) of discharges. The Ar pellet ablation is typically found to be much larger than expected from the thermal plasma electron temperature alone; the additional ablation is interpreted as being due to non-thermal runaway electrons (REs) formed during the pellet-induced temperature collapse. Simple estimates of the RE seed current using the enhanced ablation rate give values of order 1–10 kA, roughly consistent with estimates based on avalanche theory. Analytic estimates of the RE seed current based on the Dreicer formula tend to significantly underestimate it, while estimates based on the hot tail model significantly overestimate it.

Locked mode disruptions with a controlled toroidal phase are produced in the DIII-D tokamak by locking to large non-axisymmetric applied magnetic perturbations with different toroidal phases. The disruption conducted heat loads are found to reach almost completely the divertor region, possibly due to not only strong inner leg detachment but also plasma motion and limiting on the outer divertor leg shelf. The outer leg conducted heat loads are found to have a significant toroidal variation of order ±30%, with a dominant n = 1 structure. The heat load phase is shifted from the initial locked mode phase in a way that is approximately consistent with heat loss into the scrape-off layer being enhanced at the mode island O-point outer midplane crossing. These measurements suggest that pre-existing locked modes can affect the conducted heat load structure during the thermal quench by affecting the thermal quench MHD phase. This is consistent with previous MHD simulations which indicated that pre-disruption locke...

We report on the first demonstration of dissipation of fully avalanched post-disruption runaway electron (RE) beams by shattered pellet injection in the DIII-D tokamak. Variation of the injected species shows that dissipation depends strongly on the species mixture, while comparisons with massive gas injection do not show a significant difference between dissipation by pellets or by gas, suggesting that the shattered pellet is rapidly ablated by the relativistic electrons before significant radial penetration into the runaway beam can occur. Pure or dominantly neon injection increases the RE current dissipation through pitch-angle scattering due to collisions with impurity ions. Deuterium injection is observed to have the opposite effect from neon, reducing the high-Z impurity content and thus decreasing the dissipation, and causing the background thermal plasma to completely recombine. When injecting mixtures of the two species, deuterium levels as low as ~10% of the total injected atoms are observed to adversely affect the resulting dissipation, suggesting that complete elimination of deuterium from the injection may be important for optimizing RE mitigation schemes.

Injection of massive quantities of noble gases or D2 has proven to be effective at mitigating some of the deleterious effects of disruptions in tokamaks. Two alternative methods that might offer some advantages over the present technique for massive gas injection are ?shattering? massive pellets and employing close-coupled rupture disks. Laboratory testing has been carried out to evaluate their feasibility. For the study of massive pellets, a pipe-gun pellet injector cooled with a cryogenic refrigerator was fitted with a relatively large barrel (16.5-mm bore), and D2 and Ne pellets were made and were accelerated to speeds of ~ 600 and 300 m/s, respectively. Based on the successful proof-of-principle testing with the injector and a special double-impact target to shatter pellets, a similar system has been prepared and installed on DIII-D, with preliminary experiments already carried out. To study the applicability of rupture disks for disruption mitigation, a simple test apparatus was assembled in the laboratory. Commercially available rupture disks of 1-in nominal diameter were tested at conditions relevant for the application on tokamaks, including tests with Ar and He gases and rupture pressures of ~ 54 bar. Some technical and practical issues of implementing this technique on a tokamak are discussed.

Disruptions on ITER present challenges to handle the intense heat flux, the large forces from halo currents, and the potential first wall damage from energetic runaway electrons. Injecting large quantities of material into the plasma during the disruption can reduce the plasma energy and increase its resistivity to mitigate these effects. Assessments of the amount of various mixtures and quantities of the material required have been made to provide collision mitigation of runaway-electron conversion, which is the most difficult challenge. The quantities of the material required (~0.5 MPa·m3 for deuterium or helium gas) are large enough to have implications on the design and operation of the vacuum system and tokamak exhaust processing system.

A comparison of radiated power poloidal peaking during disruption mitigation using massive gas injection at multiple poloidal positions on the DIII-D tokamak is presented. The two injectors are located poloidally above and below the low field side midplane and toroidally located within the quadrants to either side of the fast bolometry diagnostic used to measure the radiated power. Differing quantities of injected neon are compared. A strong dependence of impurity poloidal flows upon the injector location is observed. Injection from the upper half of the vessel results in strong poloidal flows over the top of the plasma to the high field side midplane, while lower injection exhibits far less pronounced poloidal flow that is oriented in the opposite direction. The poloidal location of both pre-thermal quench and thermal quench emissivity peaking shows a strong dependence upon the injector location, although the poloidal flow in the upper injection case results in a much broader distribution. The wall radia...

Edge localized mode (ELM) triggering by pellet injection in the DIII-D tokamak has been simulated with the non-linear MHD code JOREK with a view to validating its physics models. JOREK has been subsequently applied to evaluate the requirements for ELM control by pellet injection in ITER. JOREK modelling results for DIII-D show that the key parameter for the triggering of ELMs by pellets is the value of the localized pressure perturbation caused by pellet injection which leads to a threshold minimum pellet size for a given injection velocity, injection geometry and H-mode plasma characteristics. The minimum pellet size for ELM triggering is found to depend on injection geometry with the largest value being required for injection at the outer midplane, intermediate for injection near the X-point and the smallest one for injection at the high-field side. The first results of studies for ELM triggering by pellet injection in ITER 15 MA Q = 10 plasmas with the foreseen injection geometry in ITER are presented.

Injection of large shattered pellets composed of variable quantities of the main ion species (deuterium) and high-Z impurities (neon) in the DIII-D tokamak demonstrates control of thermal quench (TQ) and current quench (CQ) properties in mitigated disruptions. As the pellet composition is varied, TQ radiation fractions increase continuously with the quantity of radiating impurity in the pellet, with a corresponding decrease in divertor heating. Post-TQ plasma resistivities increase as a result of the higher radiation fraction, allowing control of current decay timescales based on the pellet composition. Magnetic reconstructions during the CQ show that control of the current decay rate allows continuous variation of the minimum safety factor during the vertically unstable disruption, reducing the halo current fraction and resulting vessel displacement. Both TQ and CQ characteristics are observed to saturate at relatively low quantities of neon, indicating that effective mitigation of disruption loads by shattered pellet injection (SPI) can be achieved with modest impurity quantities, within injection quantities anticipated for ITER. This mixed species SPI technique provides a possible approach for tuning disruption properties to remain within the limited ranges allowed in the ITER design.

A newly installed Lithium Granule Injector (LGI) was used to pace edge localized modes (ELM) in DIII-D. ELM pacing efficiency was studied injecting lithium granules of nominal diameter 0.3–0.9 mm, speed of 50–120 m s−1 and average injection rates up to 100 Hz for 0.9 mm granules and up to 700 Hz for 0.3 mm granules. The efficiency of ELM triggering was found to depend strongly on size of the injected granules, with triggering efficiency close to 100% obtained with 0.9 mm diameter granules, lower with smaller sizes, and weakly depending on granule velocity. Robust ELM pacing was demonstrated in ITER-like plasmas for the entire shot length, at ELM frequencies 3–5 times larger than the ‘natural’ ELM frequency observed in reference discharges. Within the range of ELM frequencies obtained, the peak ELM heat flux at the outer strike point was reduced with increasing pacing frequency. The peak heat flux reduction at the inner strike point appears to saturate at high pacing frequency. Lithium was found in the plasma core, with a concurrent reduction of metallic impurities and carbon. Overall, high frequency ELM pacing using the lithium granule injection appears to be compatible with both H-mode energy confinement and attractive H-mode pedestal characteristics, but further assessment is needed to determine whether the projected heat flux reduction required for ITER can be met.

Shattered pellet injection (SPI) is one of the prime candidates for the ITER disruption mitigation system because of its deeper penetration and larger particle flux than massive gas injection (MGI) (Taylor et al 1999 Phys. Plasmas 6 1872) using deuterium (Commaux et al 2010 Nucl. Fusion 50 112001, Combs et al 2010 IEEE Trans. Plasma Sci. 38 400, Baylor et al 2009 Nucl. Fusion 49 085013). The ITER disruption mitigation system will likely use mostly high Z species such as neon because of more effective thermal mitigation and pumping constraints on the maximum amount of deuterium or helium that could be injected. An upgrade of the SPI on DIII-D enables ITER relevant injection characteristics in terms of quantities and gas species. This upgraded SPI system was used on DIII-D for the first time in 2014 for a direct comparison with MGI using identical quantities of neon. This comparison enabled the measurements of density perturbations during the thermal quench (TQ) and radiated power and heat loads to the divertor. It showed that SPI using similar quantities of neon provided a faster and stronger density perturbation and neon assimilation, which resulted in a lower conducted energy to the divertor and a faster TQ onset. Radiated power data analysis shows that this was probably due to the much deeper penetration of the neon in the plasma inducing a higher core radiation than in the MGI case. This experiment shows also that the MHD activity during an SPI shutdown (especially during the TQ) is quite different compared to MGI. This favorable TQ energy dissipation was obtained while keeping the current quench (CQ) duration within acceptable limits when scaled to ITER.

Publisher s Note: Changes in particle transport as a result of resonant magnetic perturbations in DIII-D [Phys. Plasmas 19, 056503 (2012)]a) In the Invited Papers from the 53rd Annual Meeting of the APS Division of Plasma Physics of the May 2012 issue of the journal, this article was originally published online and in print in the incorrect section; it was published within Ionospheric, Solar-System and Astrophysical Plasmas (Sec. 65) instead of Magnetically Confined Plasmas, Heating, Confinement (Sec. 61). AIP apologizes for this error. a)

Intentionally triggered upward and downward vertical displacement events (VDEs) leading to disruptions were pre-emptively mitigated with neon massive gas injection (MGI) coming from either above or below the plasma. Global indicators of disruption mitigation effectiveness (conducted heat loads, radiated power, and vessel motion) do not show a clear improvement when mitigating with the gas jet located closer to the VDE impact area. A clear trend of improved mitigation is observed for earlier MGI timing relative to the VDE impact time. The plasma edge magnetic perturbation is seen to lock to a preferential phase during the VDE thermal quench, but this phase is not clearly matched by preliminary attempts to fit to the conducted heat load phase. Clear indications of plasma infra-red (IR) emission are observed both before and during the disruptions. This IR emission can affect calculation of disruption heat loads; here, the time decay of post-disruption IR signals is used to correct for this effect.

Measurements from the DIII-D tokamak show that toroidal radiation asymmetries during fast shutdown by massive gas injection (MGI) are largely driven by n = 1 magnetohydrodynamic modes during the thermal quench. The phenomenology of these modes, which are driven unstable by pro le changes as the thermal energy is quenched, is described based on detailed magnetic measurements. Here, the toroidal evolution of the dominantly n = 1 perturbation is understood to be a function of three parameters: the location of the MGI port, pre-MGI plasma rotation, and n = 1 error elds. Here, the resulting level of radiation asymmetry in these DIII-D plasmas is modest, with a toroidal peaking factor (TPF) of 1:2 ± 0:1 for the total thermal quench energy and 1:4 ± 0:3 for the peak radiated power, both of which are below the estimated limit for ITER (TPF ≈ 2).

Simulations of massive gas injection for disruption mitigation in DIII-D are carried out to compare the toroidal peaking of radiated power for the cases of one and two gas jets. The radiation toroidal peaking factor (TPF) results from a combination of the distribution of impurities and the distribution of heat flux associated with the mode. When ignoring the effects of strong uni-directional neutral beam injection and rotation present in the experiment, the injected impurities are found to spread helically along field lines preferentially toward the high-field-side, which is explained in terms of a nozzle equation. Therefore when considering the plasma rest frame, reversing the current direction also reverses the toroidal direction of impurity spreading. During the pre-thermal quench phase of the disruption, the toroidal peaking of radiated power is reduced in a straightforward manner by increasing from one to two gas jets. However, during the thermal quench phase, reduction in the TPF is achieved only for a particular arrangement of the two gas valves with respect to the field line pitch. In particular, the relationship between the two valve locations and the 1/1 mode phase is critical, where gas valve spacing that is coherent with 1/1 symmetry effectively reduces TPF.

One of the major challenges that the ITER tokamak will have to face during its operations are disruptions. During the last few years, it has been proven that the global consequences of a disruption can be mitigated by the injection of large quantities of impurities. But one aspect that has been difficult to study was the possibility of local effects inside the torus during such injection that could damage a portion of the device despite the global heat losses and generated currents remaining below design parameter. 3D MHD simulations show that there is a potential for large toroidal asymmetries of the radiated power during impurity injection due to the interaction between the particle injection plume and a large n = 1 mode. Another aspect of 3D effects is the potential occurrence of Vertical Displacement Events (VDE), which could induce large poloidal heat load asymmetries. This potential deleterious effect of 3D phenomena has been studied on the DIII-D tokamak, thanks to the implementation of a multi-location massive gas injection (MGI) system as well as new diagnostic capabilities. This study showed the existence of a correlation between the location of the n = 1 mode and the local heat load on the plasma facing components but shows also that this effect is much smaller than anticipated (peaking factor of ∼1.1 vs 3-4 according to the simulations). There seems to be no observable heat load on the first wall of DIII-D at the location of the impurity injection port as well as no significant radiation asymmetries whether one or 2 valves are fired. This study enabled the first attempt of mitigation of a VDE using impurity injection at different poloidal locations. The results showed a more favorable heat deposition when the VDE is mitigated early (right at the onset) by impurity injection. No significant improvement of the heat load mitigation efficiency has been observed for late particle injection whether the injection is done “in the way” of the VDE (upward VDE mitigated by injection from the upper part of the vessel vs the lower part) or not.

The injection of small deuterium pellets at high repetition rates up to 12× the natural edge localized mode (ELM) frequency has been used to trigger high-frequency ELMs in otherwise low natural ELM frequency H-mode deuterium discharges in the DIII-D tokamak [J. L. Luxon and L. G. Davis, Fusion Technol. 8, 441 (1985)]. The resulting pellet-triggered ELMs result in up to 12× lower energy and particle fluxes to the divertor than the natural ELMs. The plasma global energy confinement and density are not strongly affected by the pellet perturbations. The plasma core impurity density is strongly reduced with the application of the pellets. These experiments were performed with pellets injected from the low field side pellet in plasmas designed to match the ITER baseline configuration in shape and normalized β operation with input heating power just above the H-mode power threshold. Nonlinear MHD simulations of the injected pellets show that destabilization of ballooning modes by a local pressure perturbation is responsible for the pellet ELM triggering. This strongly reduced ELM intensity shows promise for exploitation in ITER to control ELM size while maintaining high plasma purity and performance.

DIII-D experiments on rapid shutdown runaway electron (RE) beams have improved the understanding of the processes involved in RE beam control and dissipation. Improvements in RE beam feedback control have enabled stable confinement of RE beams out to the volt-second limit of the ohmic coil, as well as enabling a ramp down to zero current. Spectroscopic studies of the RE beam have shown that neutrals tend to be excluded from the RE beam centre. Measurements of the RE energy distribution function indicate a broad distribution with mean energy of order several MeV and peak energies of order 30?40?MeV. The distribution function appears more skewed towards low energies than expected from avalanche theory. The RE pitch angle appears fairly directed (????0.2) at high energies and more isotropic at lower energies (??100?keV). Collisional dissipation of RE beam current has been studied by massive gas injection of different impurities into RE beams; the equilibrium assimilation of these injected impurities appears to be reasonably well described by radial pressure balance between neutrals and ions. RE current dissipation following massive impurity injection is shown to be more rapid than expected from avalanche theory?this anomalous dissipation may be linked to enhanced radial diffusion caused by the significant quantity of high-Z impurities (typically argon) in the plasma. The final loss of RE beams to the wall has been studied: it was found that conversion of magnetic to kinetic energy is small for RE loss times smaller than the background plasma ohmic decay time of order 1?2?ms.

Experiments have been conducted on the DIII-D tokamak to study the distribution and repeatability of heat loads and vessel currents resulting from vertical displacement events (VDEs). For unmitigated VDEs, the radiated power fraction appears to be of order 50%, with the remaining power dominantly conducted to the vessel walls. Shot-to-shot scatter in heat loads measured at one toroidal location is not large (

The first visible light images of synchrotron emission from disruption runaway electrons are presented. The forward-detected continuum radiation from runaways is identified as synchrotron emission by comparing two survey spectrometers and two visible fast cameras viewing in opposite toroidal directions. Analysis of the elongation of 2D synchrotron images of oval-shaped runaway beams indicates that the velocity pitch angle v⊥/v|| ranges from 0.1 to 0.2 for the detected electrons, with energies above 25 MeV. Analysis of synchrotron intensity from a camera indicates that the tail of the runaway energy distribution reaches energies up to 60 MeV, which agrees with 0D modeling of electron acceleration in the toroidal electric field generated during the current quench. A visible spectrometer provides an independent estimate of the upper limit of runaway electron energy which is roughly consistent with energy determined from camera data. Synchrotron spectra reveal that approximately 1% of the total post-thermal q...

Recent experiments in the DIII-D tokamak have demonstrated real-time control and dissipation of post-disruption runaway electron (RE) beams. In the event that disruption avoidance, control, and mitigation schemes fail to avoid or suppress RE generation, active control of the RE beam may be an important line of defense to prevent the rapid, localized deposition of RE beam energy onto vulnerable vessel sections. During and immediately after the current quench, excessive radial compression of the runaway beams is avoided by a combination of techniques, improving the likelihood of the beams surviving this dynamic period without a fast termination. Once stabilized, the runaway beams are held in a steady state (out to the ohmic flux limit) with the application of active plasma current and position controls. Beam interaction with the vessel wall is minimized by avoiding distinct thresholds for enhanced wall interaction at small and large radii, corresponding to inner wall and outer limiter interaction, respectively. Staying within the “safe zone” between those radial thresholds allows for the sustainment of long-lived, quiescent runaway beams. The total beam energy and runaway electron population are then dissipated gradually by a controlled ramp-down of the runaway current.

In this paper, we introduce the first direct perturbed particle transport measurements in resonant magnetic perturbation (RMP) H-mode plasmas. The perturbed particle transport increases as a result of application of RMP deep into the core. In the core, a large reduction in E × B shear to a value below the linear growth rate, in conjunction with increasing density fluctuations, is consistent with an increase in turbulent particle transport. In the edge, the changes in turbulent particle transport are less obvious. There is a clear correlation between the linear growth rates and the density fluctuations measured at different scales, but it is uncertain which is the cause and which is the consequence. © 2012 American Institute of Physics.

The spatial distribution of runaway electron (RE) strikes to the wall during argon pellet-initiated rapid shutdown of diverted and limited plasma shapes in DIII-D is studied using a new array of hard x-ray (HXR) scintillators. Two plasma configurations were investigated: an elongated diverted H-mode and a low-elongation limited L-mode. HXR emission from MeV level REs generated during the argon pellet injection is observed during the thermal quench (TQ) in diverted discharges from REs lost into the divertor. In limiter discharges, this prompt TQ loss is reduced, suggesting improved TQ confinement of REs in this configuration. During the plateau phase when the plasma current is carried by REs, toroidally symmetric HXR emission from remaining confined REs is seen. Transient HXR bursts during this RE current plateau suggest the presence of a small level of wall losses due to the presence of an unidentified instability. Eventually, an abrupt final loss of the remaining RE current occurs. This final loss HXR emission shows a strong toroidal peaking and a consistent spatiotemporal evolution that suggests the development of a kink instability.

Large relativistic runaway electron currents (0.1?0.5?MA) persisting for ~100?ms are created in the DIII-D tokamak during rapid discharge shut down caused by argon pellet injection. Slow upward and downward ramps in runaway currents were found in response to externally applied loop voltages. Comparison between the observed current growth/decay rate and the rate expected from the knock-on avalanche mechanism suggests that classical collisional dissipation of runaways alone cannot account for the measured growth/damping rates. It appears that a fairly constant anomalous dissipation rate of order 10?s?1 exists, possibly stemming from radial transport or direct orbit losses to the vessel walls, although the possibility of an apparent loss due to current profile shrinking cannot be ruled out at present.

New rapid shutdown strategies have been recently tested in the DIII-D tokamak to mitigate runaway electrons (REs). Disruptions in ITER are predicted to generate multi-MeV REs that could damage the machine. The RE population in large tokamaks is expected to be dominated by avalanche amplification which can be mitigated at high density levels by collisional drag. Particle injection schemes for collisional suppression of RE have been developed and tested in ITER-relevant scenarios: massive gas injection, shattered pellet injection (SPI) and shell pellet injection. The results show an improved penetration of particles injected with the SPI. Another strategy has been developed to harmlessly deconfine REs by applying a non-axisymmetric magnetic perturbation to worsen their confinement. This technique appeared to deconfine seed RE before the avalanche process could amplify the RE beam. The last method tested was to use the plasma position control system on the RE beam to prevent it from contacting the wall. This proved effective in preventing high current RE beam from touching the wall and providing more time to mitigate an existing RE beam but a successful ‘soft landing’ (without fast final losses) of the RE has not been observed yet.

A severe consequence of a disruption on large tokamaks such as ITER could be the generation of multi-megaelectronvolt electron beams that could damage the vacuum vessel and the structures of the machine if they hit the wall unmitigated. The mitigation of runaway electron beams is thus a key requirement for reliable operation of ITER. In order to achieve reliable disruption mitigation, a new fast shutdown technique has been developed: the injection of a large shattered cryogenic pellet in the plasma, which is expected to increase the electron density up to levels where the beam generation processes are mitigated by collisional losses. This technique has been implemented and tested for the first time ever on DIII-D. The first tests show evidence of an almost instantaneous deposition of more than 260 Pa m3 of deuterium deep in the core. Record local densities during the thermal quench were observed for each injection with a very high reliability. Pellet mass and plasma energy content scans show an improvement of the assimilation of the particles for higher plasma energy and larger pellet mass.

Experiments have been performed in the DIII-D tokamak [J. L. Luxon, Nucl. Fusion 42, 614 (2002)] toward understanding runaway electron formation and amplification during rapid discharge shutdown, as well as toward achieving complete collisional suppression of these runaway electrons via massive delivery of impurities. Runaway acceleration and amplification appear to be well explained using the zero-dimensional (0D) current quench toroidal electric field. 0D or even one-dimensional modeling using a Dreicer seed term, however, appears to be too small to explain the initial runaway seed formation. Up to 15% of the line-average electron density required for complete runaway suppression has been achieved in the middle of the current quench using optimized massive gas injection with multiple small gas valves firing simultaneously. The novel rapid shutdown techniques of massive shattered pellet injection and shell pellet injection have been demonstrated for the first time. Experiments using external magnetic perturbations to deconfine runaways have shown promising preliminary results.

Pellet injection is planned to be the main fuelling method on ITER. The high temperature of the plasma during a fusion burn will limit the penetration of the pellet to the outer third of the minor radius. This limited penetration is expected to be compensated by a polarization drift, which will deposit the particles deeper in the plasma for the pellets injected from the high field side. In order to evaluate the expected depth of the fuelling on ITER, a good understanding of this drift effect is important. Experimental data acquired on the DIII-D (USA) and Tore Supra (France) tokamaks show that the polarization drift is influenced by the low order rational q surfaces. These surfaces appear to attenuate the polarization mechanism as the drifting particles cross them. In this paper, a correlation between the maximum of the pellet mass deposition profile and the positions of the q = 2 and q = 3 surfaces on DIII-D and Tore Supra is shown for high field side and low field side injection. A model is proposed to explain this effect and compared with the experimental results. To conclude, the possible consequences of this phenomenon on the fuelling in ITER are described.