Part 4, Chapter 8

Plasma Turbulence

Energy cascades, drift-wave turbulence, and anomalous transport

8.1 Kolmogorov Cascade in Plasma

The foundation of turbulence theory is the concept of an energy cascade. Energy is injected at large scales (low wavenumber $k$), cascades through an inertial range via nonlinear interactions, and is dissipated at small scales (high $k$).

In three-dimensional hydrodynamic turbulence, dimensional analysis (following Kolmogorov 1941) gives the famous energy spectrum in the inertial range:

$$\boxed{E(k) = C_K \varepsilon^{2/3} k^{-5/3}}$$

where $\varepsilon$ is the energy dissipation rate (equal to the energy injection rate in steady state), and $C_K \approx 1.5$ is the Kolmogorov constant.

In two-dimensional turbulence (relevant to magnetized plasmas where the magnetic field suppresses parallel dynamics), there is a dual cascade:

Inverse energy cascade: Energy flows to large scales with spectrum $E(k) \propto k^{-5/3}$
Forward enstrophy cascade: Enstrophy flows to small scales with the Kraichnan spectrum:$E(k) \propto k^{-3}$

8.2 Drift-Wave Turbulence and ITG Modes

In fusion plasmas, the dominant turbulence is driven by micro-instabilities, primarily:

Ion temperature gradient (ITG) modes: Driven by$\nabla T_i$, dominant at ion scales ($k_\perp \rho_i \sim 0.1\text{-}1$)
Trapped electron modes (TEM): Driven by$\nabla n$ and $\nabla T_e$, involving trapped particles in magnetic mirrors
Electron temperature gradient (ETG) modes: Driven by$\nabla T_e$, dominant at electron scales ($k_\perp \rho_e \sim 0.1\text{-}1$)

The mixing length estimate provides a simple scaling for the turbulent diffusivity:

$$\boxed{D \sim \frac{\gamma}{k_\perp^2}}$$

where $\gamma$ is the linear growth rate and$k_\perp$ is the perpendicular wavenumber at the maximum growth rate. For ITG turbulence, this gives the gyro-Bohm diffusivity:

$$D_{gB} \sim \frac{\rho_i}{L_T}\frac{T_e}{eB} = \frac{\rho_i}{L_T}D_B$$

where $D_B = T_e/(eB)$ is the Bohm diffusivity and $L_T$ is the temperature gradient scale length. The gyro-Bohm scaling ($\propto \rho_i/L_T$) implies that transport improves with machine size, a favorable scaling for reactor-sized tokamaks.

8.3 Hasegawa-Wakatani Model

The Hasegawa-Wakatani model extends the Hasegawa-Mima equation by including resistive parallel electron dynamics, which provides the free energy drive for drift-wave instability:

$$\frac{\partial}{\partial t}\nabla_\perp^2\phi + [\phi, \nabla_\perp^2\phi] = \alpha(\phi - n)$$$$\frac{\partial n}{\partial t} + v_{de}\frac{\partial \phi}{\partial y} + [\phi, n] = \alpha(\phi - n)$$

where $\alpha = k_\parallel^2 / (\nu_{ei} \mu_0)$ is the adiabaticity parameter:

$\alpha \to \infty$: Adiabatic limit -- electrons respond instantly, $n \approx \phi$, reduces to Hasegawa-Mima equation (no instability drive)
$\alpha \to 0$: Hydrodynamic limit -- density and potential decouple, producing 2D fluid turbulence
$\alpha \sim 1$: Drift-wave regime -- phase shift between $n$ and $\phi$ drives instability and turbulent transport

The Hasegawa-Wakatani model is the simplest system that self-consistently generates drift-wave turbulence, zonal flows, and anomalous cross-field transport.

8.4 Computational Exploration

The code below generates a synthetic 1D turbulent energy spectrum and compares the Kolmogorov ($k^{-5/3}$) and Kraichnan ($k^{-3}$) scalings, illustrating the different cascade regimes.

Turbulence Simulation

Python

script.py76 lines

import numpy as np
import matplotlib.pyplot as plt

np.random.seed(42)

# Generate synthetic turbulent spectrum
N = 4096
k = np.arange(1, N//2 + 1, dtype=float)

# Injection scale
k_inj = 5.0
# Dissipation scale
k_diss = 500.0

# Kolmogorov spectrum with injection and dissipation
E_kolmogorov = k**(-5/3) * np.exp(-k/k_diss) * (1 - np.exp(-(k/k_inj)**2))
# Add fluctuations (intermittency)
E_kolmogorov *= np.exp(0.3 * np.random.randn(len(k)))

# Kraichnan spectrum (2D enstrophy cascade)
E_kraichnan = k**(-3) * np.exp(-k/k_diss) * (1 - np.exp(-(k/k_inj)**2))
E_kraichnan *= np.exp(0.3 * np.random.randn(len(k)))

# Synthetic plasma turbulence spectrum (ITG-like)
# Ion scale break at k*rho_i ~ 1
k_rho_i = 50.0  # k where k*rho_i = 1
E_plasma = np.where(k < k_rho_i,
                     k**(-5/3) * np.exp(-k/k_diss),
                     k_rho_i**(-5/3) * (k/k_rho_i)**(-3) * np.exp(-k/k_diss))
E_plasma *= (1 - np.exp(-(k/k_inj)**2))
E_plasma *= np.exp(0.2 * np.random.randn(len(k)))

fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 5))

# Left: Kolmogorov vs Kraichnan
ax1.loglog(k, E_kolmogorov / E_kolmogorov.max(), 'orange', alpha=0.7, linewidth=0.8, label='Synthetic (3D)')
ax1.loglog(k, E_kraichnan / E_kraichnan.max(), 'cyan', alpha=0.7, linewidth=0.8, label='Synthetic (2D)')

# Reference slopes
k_ref = np.logspace(0.7, 2.3, 50)
ax1.loglog(k_ref, 2.0 * (k_ref)**(-5/3) / k_ref[0]**(-5/3), 'w--', linewidth=2, label='$k^{-5/3}$')
ax1.loglog(k_ref, 0.5 * (k_ref)**(-3) / k_ref[0]**(-3), 'r--', linewidth=2, label='$k^{-3}$')

ax1.set_xlabel('Wavenumber $k$', fontsize=13)
ax1.set_ylabel('$E(k)$ (normalized)', fontsize=13)
ax1.set_title('Kolmogorov vs Kraichnan Spectra', fontsize=14)
ax1.set_xlim(1, 1000)
ax1.set_ylim(1e-8, 10)
ax1.legend(fontsize=10)
ax1.grid(True, alpha=0.3, which='both')

# Right: Plasma turbulence spectrum with scale break
ax2.loglog(k, E_plasma / E_plasma.max(), 'orange', alpha=0.8, linewidth=1.0, label='Plasma spectrum')

k_ion = np.logspace(0.7, np.log10(k_rho_i), 30)
k_elec = np.logspace(np.log10(k_rho_i), 2.5, 30)
scale_i = E_plasma[int(k_ion[0])] / E_plasma.max()
scale_e = E_plasma[int(k_rho_i)] / E_plasma.max()

ax2.loglog(k_ion, scale_i * (k_ion/k_ion[0])**(-5/3), 'w--', linewidth=2, label='$k^{-5/3}$ (ion scale)')
ax2.loglog(k_elec, scale_e * (k_elec/k_rho_i)**(-3), 'r--', linewidth=2, label='$k^{-3}$ (electron scale)')
ax2.axvline(x=k_rho_i, color='yellow', linestyle=':', alpha=0.5, label=r'$k\rho_i = 1$')

ax2.set_xlabel('Wavenumber $k$', fontsize=13)
ax2.set_ylabel('$E(k)$ (normalized)', fontsize=13)
ax2.set_title('Plasma Turbulence Spectrum', fontsize=14)
ax2.set_xlim(1, 1000)
ax2.set_ylim(1e-8, 10)
ax2.legend(fontsize=9, loc='lower left')
ax2.grid(True, alpha=0.3, which='both')

plt.tight_layout()
plt.savefig('output.png', dpi=100, bbox_inches='tight', facecolor='#0a0a1a')
plt.close()
print("Turbulent spectra: Kolmogorov, Kraichnan, and plasma spectrum plotted.")

Click Run to execute the Python code

Code will be executed with Python 3 on the server

8.5 Zonal Flows and Transport Regulation

A remarkable feature of drift-wave turbulence is the self-generation of zonal flows -- azimuthally symmetric$\mathbf{E}\times\mathbf{B}$ flows with$k_y = 0$ and finite $k_x$. These flows are generated by the inverse energy cascade and act as a turbulence self-regulation mechanism:

Predator-prey dynamics: Zonal flows (predator) feed on drift-wave turbulence (prey). Strong turbulence generates strong zonal flows, which then shear apart the turbulent eddies, reducing transport.
L-H transition: The transition from low-confinement (L-mode) to high-confinement (H-mode) in tokamaks is believed to involve a bifurcation where edge$\mathbf{E}\times\mathbf{B}$ shear suppresses turbulence, forming a transport barrier.
Geodesic acoustic modes (GAMs): In toroidal geometry, zonal flows couple to an oscillatory branch -- the GAM -- with frequency$\omega_{GAM} \approx \sqrt{2} c_s/R$, observed in many tokamak experiments.

← Previous Next →