Skip to content

Drift Waves, ITG and Zonal Flows

Source: PHY653B Ch. 6

Intuition

Tokamak transport is not collisional. Neoclassical theory — collisions plus toroidal geometry — underpredicts observed heat transport by one to two orders of magnitude. The missing mechanism is turbulence driven by the gradients the plasma is confined to have. And the thing that stops that turbulence from being far worse is a structure the turbulence generates itself.

The ITG mode

The ion-temperature-gradient mode is the dominant instability in the tokamak core. The mechanism is an interchange-like feedback in bad curvature:

  1. A perturbation displaces hot plasma outward on the low-field side.
  2. Because the ∇B and curvature drifts are energy-dependent, hot and cold ions drift at different rates, separating charge.
  3. The resulting \(\mathbf{E}\times\mathbf{B}\) drift convects more hot plasma outward.

Positive feedback. It switches on above a threshold in \(\eta_i = L_n/L_{T_i}\) — and because transport rises so steeply above threshold, the profile is pinned near it. This is profile stiffness: push harder on the heating and the temperature gradient barely moves, because the turbulence absorbs the extra drive. It is one of the most consequential facts about tokamak confinement, and it comes out of gyrokinetic simulation rather than analytic theory.

Zonal flows

The turbulence generates, through nonlinear coupling, a mode with \(k_\theta = k_\parallel = 0\) and finite \(k_r\): a sheared, axisymmetric \(\mathbf{E}\times\mathbf{B}\) flow varying only in radius.

Zonal flows are special because they are linearly stable and nearly undamped — they cannot transport heat themselves (no poloidal or parallel structure to drive radial flux), but they shear apart the turbulent eddies that created them. Energy that would have gone into transport goes into flows that suppress transport.

The result is a predator–prey system: turbulence feeds zonal flows, zonal flows eat turbulence. Simulations show the characteristic limit-cycle oscillation, and getting zonal-flow damping right is the single most important factor in whether a gyrokinetic code predicts the correct heat flux. Early simulations that damped them incorrectly overpredicted transport by large factors.

Why this is a computational result

None of this is analytically tractable. The ITG threshold, the saturated flux, the zonal-flow damping, the resulting stiffness — all come from nonlinear gyrokinetic simulation. This is one of the clearest cases in physics of a phenomenon that was discovered and is understood through computation, and it is why gyrokinetic codes (GENE, GS2, CGYRO, GYRO) are central to the ITER programme.

It also connects to a broader idea: zonal flows are the plasma cousin of the banded jets on Jupiter, generated by the same inverse-cascade mechanism in a rotating fluid. The energy cascade in two dimensions runs to large scales, and both systems exploit it.

Common mistakes

  • Treating zonal flows as a nuisance mode. They are the main saturation mechanism; a code that damps them wrongly gets the transport wrong by orders of magnitude.
  • Expecting linear theory to predict the flux. The linear growth rate says whether the plasma is unstable, not how much heat it transports. Saturation is entirely nonlinear.
  • Ignoring collisions. Zonal flows are damped collisionally, and this weak damping sets the saturated amplitude — one of the few places where a small collision frequency dominates the answer.

Knowledge graph position

Prerequisites: gyrokinetic ordering, drift motions, interchange instability. Leads to: transport prediction for ITER, profile stiffness, transport-barrier physics.

Quiz

Q1 (conceptual). Why do zonal flows suppress transport rather than causing it?

Answer

They have \(k_\theta = k_\parallel = 0\), so they have no structure in the directions needed to drive a radial heat flux — they cannot transport anything themselves. But they vary in radius, so they shear the turbulent eddies apart, reducing the radial correlation length and hence the flux the turbulence would otherwise produce.

Q2 (conceptual). What is profile stiffness and why does it matter for reactor design?

Answer

Above the ITG threshold, transport rises so steeply with \(\nabla T\) that the gradient is effectively pinned near the critical value. Extra heating produces more turbulence rather than a steeper profile. It matters because core temperature is then set largely by the edge boundary condition — which is why pedestal physics dominates reactor performance projections.

Q3 (MCQ). Getting zonal-flow damping wrong in a gyrokinetic code primarily affects:

  • (a) the linear growth rate of the ITG mode
  • (b) the saturated turbulence level and hence the predicted heat flux
  • (c) the gyro-average
  • (d) the divergence constraint
Answer

(b). Linear growth is unaffected — zonal flows are a nonlinear saturation mechanism. Their damping sets the flow amplitude, which sets the shearing rate, which sets the saturated flux. Early codes that got this wrong overpredicted transport substantially.