Session PO06: Magnetic Confinement: DIII-D Tokamak

Live

DIII-D seeks to discover the new science and solutions for fusion energy -- to make the plasma burn in ITER and pioneer the path to a fusion pilot. Through a combination of new techniques and performance upgrades, it is possible to transform its capabilities, accessing reactor-like high power and particle densities to develop high energy confinement solutions, crucial for a compact pilot, compatible with a dissipative divertor. Upgrades are already underway to explore this exciting territory. Doubling off-axis beam injection broadened profiles to raise stability and prevent energetic particle modes. Top launch EC injection doubled current drive efficiency -- 2 more are planned. New helicon fast wave and inboard-launch LHCD are being installed to pioneer efficient reactor current drive technology. Projections indicate higher field and shaping will reach high, opaque, reactor-like pedestals, combining with new slot divertors, observed to lower target temperatures, and reactor-relevant materials tests to explore integrated core-edge solutions. An exciting negative triangularity path will also be assessed. These combine with new 3D flexibility and particle injectors to eliminate the disruption problem, enabling DIII-D to confront the challenges of the next generation of fusion reactors.   

Live

Recent DIII-D experiments contributed to the ITER physics basis and to physics understanding for future devices. A new plasma response model quantitatively predicts the narrow isolated q95 windows of ELM suppression for n$=$1, 2 and 3 RMPs in multiple devices. ExB shear from high toroidal rotation plays a key role in very high energy confinement in SH-mode plasmas. High power X-point diverted negative triangularity discharges are characterized by high confinement, significant $\beta $N and robustly ELM-free L-mode edge. Fast Er transients triggering the L-H transition are quantitatively consistent with radial polarization currents due to Reynolds stress, thermal ion orbit loss, and ion viscosity. First main ion CER inferred pedestal ion heat flux shows transition from neoclassical at high $\nu $*, to strongly anomalous at low ITER $\nu $*. Alfv\'{e}n eigenmodes close to the ion cyclotron frequency are stabilized via a controlled energetic ion density ramp. Increased off-axis NB power reduces AE drive giving improved fast ion confinement in high-qmin plasmas. NIMROD simulations of shell-pellet injection show outer flux surfaces maintained as core thermal energy is radiated. SOLPS-ITER analysis shows importance of SOL ExB drifts in SAS simulations. SOL radial heat flux width expands at high power and plasma density, exceeding empirical scaling but consistent with MHD stability limits.   

Live

The inter-ELM intrinsic carbon impurity fraction in DIII-D is measured to be 4.5$+$/-1.0{\%} in attached H-mode conditions, falling to 0.4$+$/-0.1{\%} in detached conditions, a 10X drop and a significant departure from a fixed fraction assumption. Intensity calibrated, vertically-viewing EUV/VUV spectroscopy provides line data for dominant radiation emissions in the divertor. Spectroscopic data is interpreted using Divertor Thomson scattering (DTS) for direct measurement of electron temperature and density and ADAS for simulation of emission intensities. The EUV/VUV spectrum suggests that \textasciitilde 20{\%} of the measured spectrum is unaccounted-for by line emissions alone, suggesting that molecular emissions (D$_{\mathrm{2}}$ Lyman-Werner bands) may be present. These results provide critical benchmarks for code validation and detachment scalings, insight into divertor/scrape-off-layer (SOL) impurity transport, and reveal how efficiently the intrinsic impurity can be complemented with extrinsic sources.   

Live

Experiments in DIII-D and SOLPS-ITER simulations with full E\texttimes B drifts show a strong interplay between drifts and divertor geometry on divertor detachment. For the ion \textbf{B}\texttimes $\nabla $B drift away from the X-point (`unfavorable-B$_{\mathrm{\varphi }}$'), cold plasma with T$_{\mathrm{e}}$ \textless 10 eV across the entire SAS divertor target can be achieved at very low main plasma densities. In contrast, for the ion \textbf{B}\texttimes $\nabla $B drift toward the X-point (`favourable-B$_{\mathrm{\varphi }}$'), the divertor plasma remains hot and attached across the entire target plate until the eventual onset of detachment, at a much higher density, similar to other divertors in DIII-D. The analysis shows that for unfavorable-B$_{\mathrm{\varphi }}$, the E\texttimes B drifts carry particles from the inner divertor into the outer common flux region (CFR) via the private flux region (PFR), reinforcing the anticipated geometry effect of SAS, facilitating divertor detachment. While for favorable-B$_{\mathrm{\varphi }}$, the E\texttimes B drifts drive particles out of the outer divertor via PFR, offsetting the anticipated geometric effects. The modelling also predicts a higher neutral density near the target at lower main plasma densities with unfavorable-B$_{\mathrm{\varphi }}$, compared with unfavorable-B$_{\mathrm{\varphi }}$. These results indicate that at least for DIII-D scale devices, divertor geometry and drifts have comparable magnitude effects on the divertor plasma.   

Live

Silicon carbide coatings on ATJ graphite exposed to 25 H-mode plasma discharges with edge localized modes (ELMs) in the DIII-D lower divertor demonstrated minimal changes to the macroscopic or microscopic surface morphology. Post-exposure compositional analysis reveals Si enrichment of about 10{\%}. To interpret these results, an analytic surface model is developed using calculated physical and temperature-dependent chemical sputtering yields from Si, SiC, and C (graphite). The Si content in the plasma-facing surface layer is predicted to increase with both surface temperature and divertor electron temperature, T$_{\mathrm{e,div}}$, due to more efficient physical and chemical sputtering yields of C relative to Si. The total gross erosion of Si from this mixed Si-C-SiC layer increases strongly with T$_{\mathrm{e,div}}$ but erosion of C stays relatively constant. These trends are reproduced by measurements obtained via spectroscopic inference using the S/XB method. Quantitatively, the model slightly under-predicts the measured erosion rates; this is attributed to ELMs, which have minimal (but non-zero) impact on SiC erosion. Extrapolating to a case with all-SiC walls in DIII-D, an order-of-magnitude decrease of the overall C source is predicted, motivating further investigation of SiC as a non-metallic plasma-facing material with favorable erosion properties.   

Live

DIII-D experiments in upper-single-null ELMy H-mode plasmas demonstrate that injection of low-Z materials in particulate form can effectively enhance dissipation in a closed divertor configuration. A rapid reduction of the downstream electron temperature, ion flux and heat flux was measured in the small-angle slot divertor as a result of local injection of lithium, boron, and boron nitride powder, in 2-s intervals at constant rates of 1-50 mg/s. BN injection led to a substantial increase of near-target neutral pressure and a transition to strike-point detachment with only a 2-15\% degradation in global energy confinement. 3D modeling with EMC3-EIRENE supported by spectral divertor imaging measurements is used to analyze the radiative losses within the divertor plasma. While Li radiation is concentrated near the separatrix, N radiative losses occur in the main scrape-off layer, and the B peak emission front is located in the far SOL. B and N are about an order of magnitude more efficient radiators than Li.   

Live

Validation of gyrokinetic particle transport models is crucial for predictions of low-Z impurity profiles in fusion reactors and helps to advance research of high-Z impurities which cannot be diagnosed with sufficient accuracy. The peaking of carbon density is investigated in a database of 143 H-mode discharges. Experimental $R/L_{n_C}$ values are contrasted with quasilinear gyrokinetic modeling by CGYRO. Linear multi regression analysis identified $\omega_r$ and parallel compressibility $k_{||}^2=(\hat{s}/qR)^2$ as the major parameters, explaining 93\% of non-random variance in the measured $R/L_{n_C}$. In ITG cases, CGYRO reproduced 95\% of dependence on $\omega_r$, however only 36\% of variation connected to $k_{||}^2$. Such discrepancy points to a significantly under-predicted parallel compressibility drift in the model. The remaining discrepancy indicates a too high sum of inward curvature drift together with outward thermo-diffusion. The rotodiffusion contribution in the rotation scans was near zero. Finally, we have compared the database with nonlinear ion scale CGYRO runs, and gyrofluid TGLF runs. While quasilinear runs reproduce $ R/L_{n_C}$ from nonlinear runs remarkable well, a significant discrepancy is found in TGLF thermodiffusion.   

Live

Wide pedestal QH-mode discovered on DIII-D in recent years is characterized by a stationary and quiescent H-mode with a pedestal width exceeding EPED prediction by 25{\%}. Simulations carried out by BOUT$++$ six-fields reduced MHD model demonstrate that two fluid effects may be key to understanding the physics of the wide pedestal QH-mode, which drive two kinds of MHD-scale instabilities in different radial locations: one is a peeling-ballooning mode modified by two fluid effects at the peak pedestal gradient position; the other is a drift Alfveìn wave (DAW) at the pedestal top which is driven unstable when electron dynamics is included, and therefore imposes a limit on the pedestal height. Detailed turbulence $\omega $-k power spectrum analyses in different radial locations show a reasonable agreement among BES/MIR experimental measurements on directions of multiple-modes rotation, frequency range and wave number. In order to study micro-scale turbulence transport dynamics, simulations using gyro-kinetic code CGYRO find trapped-electron mode (TEM) unstable inside the pedestal region, which may be regulating the density and temperature gradients of wide pedestal QH-mode. This work presents improved physics understanding of the pedestal stability and turbulence dynamics for wide pedestal QH-mode.   

Live

The L-H transition power threshold (P$_{\mathrm{LH}})$ in favorable magnetic geometry (ion $\nabla $B drift direction towards X-point) is up to a factor of 2 lower than in the unfavorable magnetic geometry (ion grad-B drift direction away from X-point) on multiple tokamaks. In a systematic experiment on DIII-D, the ion $\nabla $B drift direction was changed continuously from unfavorable to favorable configurations during the plasma shot. In the process the input power was kept constant at the value above the P$_{\mathrm{LH}}$ for favorable configuration, but lower than the P$_{\mathrm{LH}}$ for unfavorable configuration. Toroidal field and plasma current were also kept constant and there is little change in the edge n$_{\mathrm{e}}$ and T$_{\mathrm{e}}$ profiles. The density fluctuation amplitude, measured with BES decreases approaching the transition, while a large increase of turbulence Reynolds stress and turbulence flow shear are simultaneously observed during the slow varying of the ion $\nabla $B drift direction approaching the transition. These measurements demonstrate an important correlation between turbulence and turbulence driven flow and a lowering of P$_{\mathrm{LH}}$, provide insights into the underlying physics behind the hidden parameters and inform a more complete physics-based model of L-H transition power threshold.   

Live

We report empirical observations of magnetic island heteroclinic bifurcation for the first time. This new class of bifurcations is predicted to occur in tokamaks when multiple, rotationally coupled tearing modes (TMs) of the same helicity grow simultaneously [1]. In this process a second island forms within the original island, resulting in a composite structure of two islands with disjoint O-lines. This behavior is observed in coupled 2/1 TMs in DIII-D. Poincare maps fully constrained by magnetic data show bifurcation from heteroclinic to homoclinic topology in the 2/1 island as the 4/2 relative amplitude (R42) decreases. Initially, the local electron temperature (Te) peak in the 2/1 island splits, consistent with 2 O-points. As R42 decreases a single Te peak forms, consistent with 1 O-point. Electron cyclotron current drive (ECCD) can be an effective method for controlling homoclinic 2/1 island growth. However, heteroclinic bifurcation splits the ECCD between the O-points which can complicate or prevent active stabilization. This imposes challenges on the EC wave launch geometry which is not accounted for in present tokamaks nor in the ITER research plan. These observations call for developing tearing stability theory and control solutions for heteroclinic islands in tokamaks.   

Live

Experiments in the DIII-D tokamak show that a broadened fast-ion pressure profile enables better control of Alfv\'{e}n Eigenmodes (AEs), improves fast-ion confinement, and allows access to new regimes with 15{\%} higher normalized plasma beta ($\beta_{\mathrm{N}})$ than previously achieved in high-field, steady-state scenarios with negative central shear and q$_{\mathrm{min}}$\textgreater 2. Reversed Shear Alfv\'{e}n Eigenmodes (RSAEs) were reduced in the current ramp by increasing the off-axis neutral beam power fraction, resulting in $\sim $24{\%} higher ratio of measured neutrons to calculated classical neutrons. The neutron fraction was further improved using Electron Cyclotron Current Drive aimed on-axis, which suppressed RSAEs by moving the q$_{\mathrm{min}}$ location inward toward reduced beam pressure gradient and higher plasma pressure, resulting in a $\sim $36{\%} higher neutron ratio than the reference shot. In flattop, fast-ion confinement improved by $\sim $25{\%} after reducing beam pressure gradient (thus AE drive) by increasing the off-axis beam power fraction from 30{\%} to 70{\%}. Record parameters were achieved by increasing the relative density, reaching $\beta_{\mathrm{N}}\sim $3.1 and H$_{\mathrm{89}}\sim $2.3 at B$_{\mathrm{T}}=$2.0 T and q$_{\mathrm{95}}=$6.0. These experiments mark significant progress in understanding potential optimized regimes for steady-state advanced tokamaks that can avoid AE-induced fast-ion redistribution, loss, reduced heating efficiency, and limits to the achievable $\beta_{\mathrm{N}}$.   

Live

Tokamak fusion reactors will require predictive, integrated models to optimize performance while maintaining robustness against disruptions. The STEP (Stability, Transport, Equilibrium, \& Pedestal) module, developed in OMFIT, predicts stable equilibria self-consistently with core-transport and pedestal calculations by coupling together the following codes: ONETWO, TGYRO, EFIT, CHEASE, EPED, DCON, GATO, and CHEF (a current-drive, heating, \& fueling module). Each code reads and writes data from a centralized IMAS data structure, allowing them to be run in arbitrary order and enabling open-loop, feedback, and optimization workflows. Core-pedestal calculations with STEP have been validated against DIII-D, and used to assess performance in ITER and the suppression of turbulence in DIII-D negative-triangularity plasmas. We use STEP to optimize heating and current drive to maximize plasma pressure while maintaining MHD stability. Stability maps are generated and validated against stability limits in DIII-D. Predictive optimization for potential DIII-D upgrades and other next-step devices are performed to assess their capability to explore sustained, high-power-density scenarios.   

Live

It has been found that massive (500 Torr-L) deuterium injection rapidly (\textless 5 ms) expels existing argon from the runaway electron (RE) plateau current channel in DIII-D, creating a low-dissipation RE plateau regime which could give reduced RE-wall energy deposition in ITER. The Ar expulsion has been found to result from rapid cooling of the background thermal plasma due to D and D$_{\mathrm{2}}$ neutral cooling. In the resulting neutral-dominated plasma, radial transport of argon changes from slow cross-field ion transport to more rapid neutral transport, resulting in a hollow total density profile and the bulk of the argon found outside the RE current channel. The Ar-purged RE plateaus appear to result in a very rapid final loss instability, resulting in reduced RE energy deposition to the wall. This reduced energy deposition is consistent with coupled-circuit modeling of the RE plateau-wall interaction, which predicts low energy deposition if the final loss is rapid compared with the plasma resistive timescale.   

Live

Experiments injecting boron-filled diamond shell pellets into DIII-D are building the physics basis for disruption mitigation through core impurity deposition [1]. This technique is being developed as a possible alternative to shattered pellet injection (SPI) as a mitigation technique on ITER and/or CFPP. As predicted [2], comparison of the measured neutron rate evolution during shell transit with 0-D calculations derived from measured line-integrated density were consistent with plasma cooling dominated by dilution, indicating that measured radiated power spikes were highly localized near the shell (with minimal energy loss) and that the injection process represents a reasonable approximation of ideal core deposition. Total impurity assimilation fraction was fairly insensitive to pellet injection velocity, and hence final pellet deposition depth. Energetic particles from the beamline intersecting the pellet trajectory were found to alter pellet penetration, indicating they should be accounted in ablation models. These results form an early basis for determining if mitigation by core impurity deposition can provide qualitatively better mitigation than existing methods. [1] E.M. Hollmann et al., Phys. Rev Lett. 122, 065001 (2019) [2] V. A. Izzo, Nucl. Fusion 60, 066023 (2020)   

Live

New DIII-D results provide a candidate explanation for why achieving ELM suppression by resonant magnetic fields (RMPs) remains elusive in double null (DN) diverted configurations: the lack of ELM suppression in DN correlates with a damped high-field side response of field-aligned structures that could be indicative of a missing resonant tearing needed to stop inward growth of pedestal. The lack of ELM suppression in DN is found despite matching favorable conditions for RMP suppression identified in lower single null (LSN). Here, low $\Omega_{E\times{B}}$ is aligned with a resonant surface at the pedestal top over a range of $q_{95}$ from 3.4 to 4.1 with $n_{e,ped}<2.5\times10^{19}m^{-3}$. The 3D plasma response measured on the high-field side (HFS) drops in plasma shapes transitioning from LSN to DN and recovers in upper single null (USN), while the low field side (LFS) response remains relatively constant from LSN to USN. The reduced HFS response is found across a range of $\left|dR_{sep}\right|<1cm$ indicating it is not restricted to balanced DN. Linearized MHD modeling similarly shows a reduction in HFS response in double null configurations. Conceptually, the additional null adds radial shear to externally driven field-aligned modes on the LFS and may inhibit HFS coupling.