The L–H transition in tokamaks: power threshold, density minimum and toroidal-field asymmetry

B. De Lucca, P. Ricci, B. Labit, D. Mancini, L. N. Stenger, Z. Tecchiolli · 2026

Read The L–H transition in tokamaks: power threshold, density minimum and toroidal-field asymmetry by B. De Lucca, P. Ricci, B. Labit, D. Mancini, L. N. Stenger, Z. Tecchiolli (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-00624.txt.

Introduction

Brenno De Lucca and 5 co-authors model the L-H transition in tokamaks, the abrupt jump from low to high confinement that fusion reactors depend on. No one has explained it in over forty years. Their 3D, flux-driven simulations use a diverted geometry, the shape of real machines. The jump appears at lower power in the preferred field direction, matching the known asymmetry.

In the runs, electromagnetic drift-wave turbulence on its own creates a sheared flow that suppresses transport, the accepted signature of the high-confinement state. The field-direction asymmetry is traced to time-reversal symmetry breaking by finite collisionality, shown through a study of the momentum flux of the churning flow. From basic theory the authors derive scaling laws for the power threshold, the density minimum, and the minimum power. The laws match or beat the measured scalings engineers use today.

Catalog record

CREATORB. De Lucca, P. Ricci, B. Labit, D. Mancini, L. N. Stenger, Z. Tecchiolli
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE25 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.00624

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

The L–H transition in tokamaks: power threshold, density minimum and toroidal-field asymmetry B. De Lucca, P. Ricci, B. Labit, D. Mancini, L. N. Stenger, Z. Tecchiolli Abstract Flux-driven two-fluid simulations in a diverted tokamak exhibit a confinement transition when a critical heating threshold is exceeded, as in experimental observations of the L–H transition. The simulations show that electromagnetic drift-wave turbulence spontaneously generates a sheared 𝑬 × 𝑩 \bm{E}\times\bm{B} flow responsible for transport suppression in the edge. The toroidal-field asymmetry effect arises from time-reversal symmetry breaking by finite collisionality, as demonstrated by a quasilinear calculation of the turbulent momentum flux. First-principles scaling laws are derived for the L–H power threshold in the high- and low-density branches, for the density minimum, and for the minimum power, all matching or surpassing existing empirical scalings. Above a threshold input power, turbulence in the edge region of fusion devices undergoes a rapid suppression, leading to a factor \sim 2 increase in the energy confinement time. Known as the L–H transition 40 , this mechanism is considered essential to achieve burning plasma conditions in future devices, including ITER 9 and SPARC 6 . Despite numerous theoretical proposals 28 ; 33 ; 18 ; 21 ; 14 ; 35 ; 11 , the physics underlying the L–H transition remains debated, as no existing model simultaneously accounts for several well-established experimental observations. These include the non-monotonic dependence of the threshold power, P_{\mathrm{LH}} , on plasma density 12 and the observation that the threshold power is generally lower when the ion \nabla B drift points from the core region toward the X -point in a diverted geometry ( favourable configuration) 1 ; 37 . For densities above a critical value, called the density minimum, P_{\mathrm{LH}} is well-described by the empirical scaling law 24 : P_{\mathrm{LH}}^{\rm ITPA}=0.0488\,\overline{n}_{e20}^{0.717}B^{0.803}S^{0.941}\,[\text{MW}], (1) with \overline{n}_{e20} the line-averaged density in 10^{20}\,\mathrm{m}^{-3} , B the magnetic field strength in T, and S the plasma surface area in \mathrm{m}^{2} . Though edge-transport bifurcations have been observed several times in global flux-driven turbulence simulations 5 ; 16 ; 41 , their relevance to the experimental L–H transition observations remains unclear, and first-principles-based predictive capabilities are lacking. Here, we present power-ramp simulations, starting from L–mode turbulence, carried out with the global flux-driven code GBS 29 ; 15 , which solves the drift-reduced Braginskii equations. These simulations exhibit a confinement transition occurring at lower power in the favourable toroidal-field configuration, and based on these results, we derive a theory of the L–H transition with predictive capabilities. The physical model solved by GBS is reported in detail in 15 . Using a fluid model is justified by the large edge collisionality in existing and future tokamaks. For typical L-mode parameters in TCV 10 , the ratio of electron mean-free-path to connection length in the edge is small \lambda_{e}/L_{\parallel}\lesssim 0.1 , with comparable values in AUG 33 , and similar values expected in ITER and SPARC. The simulation parameters at the separatrix are R/\rho_{s}=500 , \nu=5\times 10^{-2}c_{s}/R , a/R\simeq 0.3 , m_{i}/m_{e}=800 , Z=1 , and \beta=10^{-4} , where R is the tokamak major radius, \rho_{s}=c_{s}/\Omega_{i} is the sound Larmor radius, with c_{s}=\sqrt{T_{e}/m_{i}} the ion sound speed and \Omega_{i}=ZeB/m_{i} the cyclotron frequency, \nu=e^{2}n/(m_{i}\sigma_{\|}) is the collisional frequency, a is the minor radius, m_{i}/m_{e} is the ion-to-electron mass ratio, Z is the ion charge number, and \beta is the electron plasma beta. The magnetic field \bm{B}=\sigma_{T}R|B_{\varphi}|\nabla\varphi+\sigma_{p}\nabla\varphi\times\nabla\psi is that of a diverted lower-single-null tokamak, with edge safety factor q_{a}\simeq 4 and magnetic shear \hat{s}_{a}\simeq 2 , and previously used in Ref. 16 . The plasma current direction is encoded in \sigma_{p}\equiv\mathrm{sign}(\bm{I}_{p}\cdot\nabla\varphi) and the favourable configuration corresponds to \sigma_{T}\equiv\mathrm{sign}(\bm{B}\cdot\nabla\varphi)=-1 . Figure 1 : Time traces of injected power P_{\rm in} (top), energy confinement time \tau_{E} (middle), minima of radial electric field E_{r}^{\rm min} and ion pressure gradient (\partial_{r}p_{i}/en)^{\rm min} (bottom), for simulations in the favourable magnetic field direction ( fav , red), unfavourable ( unfav , blue) and unfavourable at higher power ( unfav-high , dark-blue). Starting from L–mode simulations in quasi-steady-state, the input power is suddenly increased by a factor \simeq 1.4 at time t=0 . This is achieved by raising the heating sources in the core, while keeping the density source constant, with equal ion and electron heating. Figure 1 shows time traces of input power, P_{\rm in} , energy confinement time, \tau_{E} , and the radial electric field minimum, E_{r}^{\rm min} . In the favourable configuration, \sim 10R/c_{s} after additional heating is injected, a fast transport bifurcation occurs on a timescale of \sim 5R/c_{s} , with \tau_{E} increasing by a factor \sim 2 . After an initial overshoot, the confinement time stabilises at a factor \sim 1.5 larger than its initial value. The flux-surface-averaged pressure and radial electric field profiles are shown in Fig. 2 as a function of distance from the separatrix, r-r_{\rm sep} . During the transition, the electric-field well deepens, suppressing turbulent transport and steepening the edge pressure profile (Figs. 1 and 2 ). Conversely, for the same input power, the unfavourable simulation shows neither improved confinement nor significant changes in the electric field. Repeating the ramp at roughly twice the L-mode input power, the unfavourable case also transitions to a high confinement mode, with a similar deepening of the electric-field well as in the favourable case. Figure 2 : Radial profiles of flux-surface-averaged total pressure p (normalised to the separatrix value p_{\rm sep} ) and radial electric field E_{r} , at t=0 and at t=80\,R/c_{s} . Favourable (top) and unfavourable (bottom), with the latter shown at both matched input power (blue squares), and higher power (dark-blue triangles). Edge turbulence prior to the transition arises from a combination of resistive ballooning and electron drift-wave (DW) modes. Removing the curvature drive from the simulations substantially suppresses turbulence, confirming the role of interchange modes, and spatio-temporal Fourier analysis reveals DW signatures: \omega\simeq\omega_{*} and e\widetilde{\phi}/\langle T_{e}\rangle\sim\widetilde{n}/\langle n\rangle\sim\widetilde{p}_{e}/\langle p_{e}\rangle (fluctuations are denoted by \widetilde{f}=f-\langle f\rangle , with \langle f\rangle the time- and flux-surface average). Simulations show that a transport barrier only forms when the edge plasma– \beta exceeds a critical value, consistent with previous flux-tube results 33 . Two electromagnetic mechanisms drive the suppression: linear DW stabilisation via enhanced electron adiabatic response 32 ; 25 , and nonlinear \bm{E}\times\bm{B} shear generation by finite– \beta DWs 18 . In fact, artificially removing the zonal component of the \bm{E}\times\bm{B} drift from the simulations immediately restores high transport levels, confirming its role in barrier formation. The rise of \bm{E}\times\bm{B} shear observed in the simulations originates from turbulent generation of momentum flux, as revealed by analysing the radial electric field evolution equation. Indeed, in the drift-reduced Braginskii model implemented in GBS , the flux-surface-averaged radial electric field evolves according to \partial_{t}\langle\mathcal{E}_{r}\rangle=\mathcal{F} , where \mathcal{E}_{r}=-(dr/d\psi)\,nm_{i}|\nabla\psi|^{2}E_{r}/B^{2} and \mathcal{F}=\langle J_{\parallel}\delta\bm{B}\cdot\nabla\psi/B\rangle-\langle\bm{v}_{E}\cdot\nabla\bm{\varpi}\cdot\nabla\psi\rangle encodes the nonlinear momentum flux (Maxwell and Reynolds stress). Here, \bm{\varpi}=(nm_{i}/B^{2})(\nabla_{\perp}\phi+\nabla_{\perp}p_{i}/en) is the generalised vorticity, \delta\bm{B}=\nabla\times(A_{\parallel}\bm{b}) the perturbed magnetic field, and \bm{v}_{E} the \bm{E}\times\bm{B} drift velocity. Gyro-viscosity, parallel vorticity advection, radial diamagnetic current flux, and \partial_{t}(\partial_{r}p_{i}/en) are neglected in the expression for \mathcal{F} , yielding small contributions to \partial_{t}E_{r} . Figure 3 a shows the nonlinear momentum flux \mathcal{F} at the position of the electric-field-well minimum, r=r_{\rm min} , as a function of time during the transition phase, t\in(0,20R/c_{s}) . The

Read the complete text