betaFeedback

WorksPhysics and Astronomy

Analytic regimes of tokamak plasma

Abstract

The paper suggests certain regimes of fusion plasma with magnetic confinement, when the Larmor radius λλ and the angle αα (determining the orientation of the local coordinates) form an analytic function w=λeiαw=λe^{iα} of complex variable z=x1+ix2z=x_1+i x_2 (x1x_1 and x2x_2 are coordinates of tokamak cross-section). It is known that an efficient nuclear fusion requires transport barriers, which might be due to zonal/poloidal flow; and the emergence of such flow is implied by the adiabatic extra conservation (established in a local slub). The suggested analytic regimes are based on the global extension of this extra conservation; transport barriers could spread to the entire tokamak plasma. It is interesting that in these regimes, plasma temperature reaches maximum often not at a single location (of tokamak cross-section), but at several points separated by regions of lower temperature.

Extra invariant

Suppose a significant mode of plasma dynamics is represented by the drift waves with dispersion relation ωk=q/(1+p2+q2)\omega_k = q/(1 + p^2 + q^2). The interaction of these waves can take different forms. The dispersion is written in dimensionless form for a local slub: The local wave number components pp, qq of the wave vector k\mathbf{k} are non-dimensionalized by the Larmor radius λ\lambda.

The radial transport in a tokamak is reduced when the wave system organizes itself to produce strong alternating zonal/poloidal flow [3]. This is because the ballistic transfer of heat and particles in the radial direction — by any plasma mode — is interrupted (radial velocity field is de-correlated) when there is strong randomly alternating poloidal current. At the same time, the poloidal flow does not contribute to radial transport.

It is interesting that the emergence of zonal/poloidal flow follows from the extra conservation. In addition to the energy and enstrophy, a system of drift waves is known [4, 5] to conserve one more quadratic integral I=∫ηkNk dkI = \int\eta_k \mathcal{N}_k\,dk. This is the extra invariant with spectral density

ηk=arctan⁡p+q3p2+q2−arctan⁡p−q3p2+q2.(1)\eta_k = \arctan\frac{p + q\sqrt{3}}{p^2 + q^2} - \arctan\frac{p - q\sqrt{3}}{p^2 + q^2}. \tag*{(1)}

The function Nk=εk/ωk\mathcal{N}_k = \varepsilon_k/\omega_k is the wave action spectrum, εk\varepsilon_k is the energy spectrum. The extra conservation is adiabatic: It holds approximately over long time. The function η(p,q)\eta(p,q) is shown in Figure 1.

Image not extracted

Figure 1. Contour plot of the function (1). Color represents the values of η(p,q)\eta(p,q). The graph is shown only for positive p,qp,q, due to the symmetries η(p,q)=η(−p,q)=−η(p,−q)\eta(p,q)=\eta(-p,q)=-\eta(p,-q). All contour lines pass through the origin with 30∘30^\circ slope.

As well known, the conservation of the energy and enstrophy leads to the inverse energy cascade (local or non-local). The conservation of the extra invariant ensures [6] that the energy is not just transferred towards small wave numbers, but concentrates around pp-axis (∣q∣≪∣p∣|q| \ll|p|), which corresponds to zonal/poloidal flow. (In the last three references, pp and qq are interchanged for geophysical context.)

The waves similar to the plasma drift waves occur in different problems. As well known, the waves with the same dispersion relation are the Rossby waves in atmospheres and oceans. Much slower waves (slower than the relatively slow Rossby waves) also appear in astrophysical magnetohydrodynamics [7, 8, 9]. In all these situations, the extra invariant has important physical implications.

Global conservation

Let (x1,x2,x3)(x_1,x_2,x_3) be global dimensional coordinates; x3x_3 is the toroidal coordinate. In all cross-sections x3=constx_3 = const, plasma properties are as-

sumed to be the same. In the present paper, we assume that the confining magnetic field B\mathbf{B} is purely toroidal, B=Bx^3\mathbf{B}=B\hat{x}_3; and so, it is orthogonal to the cross-section.

The drift waves have the dispersion relation Ω=ν(x1,x2)ω(p,q)\Omega=\nu(x_1,x_2)\omega(p,q), where pp and qq are local dimensionless wave numbers; they are related to the global dimensional wave vector k=(k1,k2)\mathbf{k}=(k^1,k^2) by a rotation through angle α(x1,x2)\alpha(x_1,x_2) [c≡cos⁡α, s≡sin⁡α][c\equiv\cos\alpha,\ s\equiv\sin\alpha]

p=λ(ck1+sk2),q=λ(−sk1+ck2).(2)p=\lambda(ck^1+sk^2),\qquad q=\lambda(-sk^1+ck^2). \tag*{(2)}

In the plane x=(x1,x2)\mathbf{x}=(x_1,x_2), the drift wave field is described by the inhomogeneous wave kinetic equation

∂N∂t+∂ω∂k⋅∂N∂x−∂ω∂x⋅∂N∂k=St⁡[N](3)\frac{\partial\mathcal{N}}{\partial t}+\frac{\partial\omega}{\partial\mathbf{k}}\cdot\frac{\partial\mathcal{N}}{\partial\mathbf{x}}-\frac{\partial\omega}{\partial\mathbf{x}}\cdot\frac{\partial\mathcal{N}}{\partial\mathbf{k}}=\operatorname{St}[\mathcal{N}] \tag*{(3)}

for the dynamics of the wave action spectrum N(k,x,t)\mathcal{N}(\mathbf{k},\mathbf{x},t); tt is dimensionless time (nondimensionalized by the frequency ν\nu). (The wave kinetic equation has been utilized for a long time in physics and applied mathematics; recently, kinetic equation approach was used [10, 11] to study a related problem: interaction between zonal flow and turbulence.) The l.h.s. in equation (3) describes the slow refraction of waves, while the r.h.s. (stoss-term or collision integral) describes changes in N\mathcal{N} due to 3-wave resonance interactions

k3=k1+k2,ω(k3)=ω(k1)+ω(k2).(4)\mathbf{k}_3=\mathbf{k}_1+\mathbf{k}_2,\qquad\omega(\mathbf{k}_3)=\omega(\mathbf{k}_1)+\omega(\mathbf{k}_2). \tag*{(4)}

We assume that the interactions occur locally in physical space, and so, the three interacting waves have the same plasma parameters.

Note: We can consider only linear equations but study nonlinear interactions. This is because — in general — the conservation of a quadratic integral is determined by the dispersion relation. This fact was noted by Zakharov and Schulman: They called degenerative [12] a dispersion relation that allows for an additional conservation law (besides the energy, momentum, and wave action). They showed [13] that conservation in the homogeneous wave kinetic equation is equivalent to cancellation of the “small divisors”, and therefore, to approximate extra conservation in the dynamic equation. The extra conservation holds for any nonlinearity, provided the dynamic equation is Hamiltonian (which is usually the case for a physical system). I believe the conservation by the inhomogenous part, on the left of the wave kinetic equation (3), can also be studied in terms of the dynamic equation.

Let us check the global conservation of an arbitrary integral

I(x,t)=∫F(k,x)N(k,x,t) dk,(5)I(\mathbf{x},t)=\int F(\mathbf{k},\mathbf{x})\mathcal{N}(\mathbf{k},\mathbf{x},t)\,d\mathbf{k}, \tag*{(5)}

determined by an arbitrary spectral density FF. According to the equation (3),

∂I∂t+∂∂x⋅∫∂ω∂kFN dk+∫∂(F,ω)∂(k,x)N dk=∫St⁡[N]F dk,(6)\frac{\partial I}{\partial t}+\frac{\partial}{\partial\mathbf{x}}\cdot\int\frac{\partial\omega}{\partial\mathbf{k}}F\mathcal{N}\,d\mathbf{k}+\int\frac{\partial(F,\omega)}{\partial(\mathbf{k},\mathbf{x})}\mathcal{N}\,d\mathbf{k}=\int\operatorname{St}[\mathcal{N}]F\,d\mathbf{k}, \tag*{(6)}

where we use notation

∂(F,G)∂(u,v)=∂F∂u⋅∂G∂v−∂F∂v⋅∂G∂u\frac{\partial(F,G)}{\partial(u,v)}=\frac{\partial F}{\partial u}\cdot\frac{\partial G}{\partial v}-\frac{\partial F}{\partial v}\cdot\frac{\partial G}{\partial u}

for arbitrary functions FF, GG of arbitrary variables uu, vv (vector or scalar). The quantity II is conserved if in the equation (6), the last two integrals vanish.

The wave kinetic equation (3) always conserves the energy, when F=ωF=\omega.

3. Multiplicity. Now let FF be the function (1), F=ηF=\eta, with p,qp,q and II in (2). Then the r.h.s. in (6) vanishes. Indeed by (2), pp and qq are linear combinations of the global components of the wave vector k\mathbf{k}, and so, the equations (4) lead to p3=p1+p2p_3=p_1+p_2, q3=q1+q2q_3=q_1+q_2, ω(p3,q3)=ω(p1,q1)+ω(p2,q2)\omega(p_3,q_3)=\omega(p_1,q_1)+\omega(p_2,q_2). These equations imply that

η(p3,q3)=η(p1,q1)+η(p2,q2),(7)\eta(p_3,q_3)=\eta(p_1,q_1)+\eta(p_2,q_2), \tag*{(7)}

i.e. the function η\eta is conserved in 3-wave resonance interactions. Actually, η\eta is the only independent function conserved in the 3-wave interactions. This fact is valid even if the equation (7) holds only in an arbitrary small neighborhood anywhere on the resonance manifold [14, 15]. All other conserved functions are linear combinations of pp, qq, ω(p,q)\omega(p,q), and η(p,q)\eta(p,q).

In the second integral of equation (6),

∂(F,ω)∂(k,x)=∂(η,ω)∂(p,q)∂(p,q)∂(k,x).(8)\frac{\partial(F,\omega)}{\partial(\mathbf{k},\mathbf{x})}=\frac{\partial(\eta,\omega)}{\partial(p,q)}\frac{\partial(p,q)}{\partial(\mathbf{k},\mathbf{x})}. \tag*{(8)}

According to the equations (2),

∂(p,q)∂(k,x)=λk⋅(−∂λ∂x2−λ∂α∂x1,∂λ∂x1−λ∂α∂x2).(9)\frac{\partial(p,q)}{\partial(\mathbf{k},\mathbf{x})}=\lambda\mathbf{k}\cdot\left(-\frac{\partial\lambda}{\partial x_2}-\lambda\frac{\partial\alpha}{\partial x_1},\frac{\partial\lambda}{\partial x_1}-\lambda\frac{\partial\alpha}{\partial x_2}\right). \tag*{(9)}

So, the expressions (9) and (8) vanish for all wave vectors k\mathbf{k} if functions μ=ln⁡λ\mu=\ln\lambda and α\alpha satisfy the Cauchy-Riemann conditions:

1λ∂λ∂x1=∂α∂x2,1λ∂λ∂x2=−∂α∂x1,(10)\frac{1}{\lambda}\frac{\partial\lambda}{\partial x_1}=\frac{\partial\alpha}{\partial x_2},\qquad\frac{1}{\lambda}\frac{\partial\lambda}{\partial x_2}=-\frac{\partial\alpha}{\partial x_1}, \tag*{(10)}

saying that μ+iα\mu+i\alpha is an analytic function of z=x1+ix2z=x_1+ix_2. In other words, w=λeiαw=\lambda e^{i\alpha} should be an analytic function of zz.

The drift waves are determined by the gradient of some macroscopic plasma quantity. When this quantity reaches maximum, the frequency ν\nu vanishes. There are no drift waves at such locations, the entire reasoning based on the wave kinetic equation (3) fails, and the analyticity is not required. The analytic function w(z)w(z) could have singularities at such points.

4. Examples. Since λ\lambda is related to the plasma temperature (e.g. λ∝T\lambda\propto\sqrt{T}, with TT being the ion temperature), we would like to have smaller λ\lambda at the tokamak boundary and larger λ\lambda in the central part of the tokamak. The figures below show the Larmor radius λ=∣w(z)∣\lambda=|w(z)| in the logarithmic scale: The color bar numbers are the values of ln⁡λ(x1,x2)\ln\lambda(x_1,x_2); the white area corresponds to the values exceeding the top color.

Example 1.

w=A1z−a1+A2z−a2(11)w=\frac{A_1}{z-a_1}+\frac{A_2}{z-a_2} \tag*{(11)}

with four complex parameters a1,a2,A1,A2a_1,a_2,A_1,A_2. To ensure fastest decrease of λ=∣w∣\lambda=|w| as z→∞z\to\infty, we set A1+A2=0A_1+A_2=0 so that w=O(z−2)w=O(z^{-2}). Using rotation and stretching in the ww-plane, we can assume A1=−A2=1A_1=-A_2=1. Choosing the axis x1,x2x_1,x_2, we can have a1=−a2=iba_1=-a_2=ib (bb is a real number). Re-scaling length, we set b=1b=1. See Figure 2. The boundary of the tokamak cross-section should pass through the region with smaller values of λ\lambda.

Image not extracted

Figure 2. w=1z−i−1z+iw=\frac{1}{z-i}-\frac{1}{z+i}.

Example 2.

w=A1z−a1+A2z−a2+A3z−a3(12)w=\frac{A_1}{z-a_1}+\frac{A_2}{z-a_2}+\frac{A_3}{z-a_3} \tag*{(12)}

with six complex parameters Aj,ajA_j,a_j (j=1,2,3j=1,2,3). Similar to example 1, we require w=O(z−3)w=O(z^{-3}), which gives us the two linear equations on A1,A2,A3A_1,A_2,A_3. Using rotation and stretching in the ww-plane, we can set

A1=a3−a2,A2=a1−a3,A3=a2−a1.(13)A_1=a_3-a_2,\quad A_2=a_1-a_3,\quad A_3=a_2-a_1. \tag*{(13)}

Finally, using rotation, shifting, and stretching in the zz-plane, we can exclude 4 real parameters, leaving us with two parameters. See Figure 3.

Image not extracted

Figure 3. The analytic function w(z)w(z) is defined in (12)-(13) with a1=ia_1=i, a2=−ia_2=-i, a3=1/2a_3=1/2.

References

References

  1. [3]H. Biglari, P. Diamond, and P. Terry, Physics of Fluids B: Plasma Physics 2, 1 (1990).
  2. [4]A. M. Balk, S. V. Nazarenko, and V. E. Zakharov, Phys. Lett. A 152, 276 (1991).
  3. [5]A. M. Balk, Phys. Lett. A 155, 20 (1991).
  4. [6]A. M. Balk, Phys. Lett. A 345, 154 (2005).
  5. [7]R. Hide, Proc. Roy. Soc. London 259, 615 (1966).
  6. [8]T. V. Zharzhavashvili, R. Oliver, J. L. Ballester, and B. M. Shergelashvili, A&A 470, 815 (2007).
  7. [9]A. M. Balk, The Astrophysical Journal 926, 2 (2022).
  8. [10]J. B. Parker, Journal of Plasma Physics 82, 59580602 (2016).
  9. [11]J. Marston and S. Tobias, arXiv preprint arXiv:2205.05513 (2022).DOI
  10. [12]V. E. Zakharov and E. I. Schulman, Physica D 1, 192 (1980).
  11. [13]V. E. Zakharov and E. I. Schulman, Physica D 29, 283 (1988).
  12. [14]E. V. Ferapontov, “Web geometry and mathematical physics,” (Kluwer Academic Publishers, Norwell, MA, 1992) pp. 310–323, appendix in the book “Geometry and Algebra of Multidimensional Three-Webs” by M. A. Akivis and A. M. Shelekhov.
  13. [15]A. M. Balk and E. V. Ferapontov, in Nonlinear waves and weak turbulence, edited by V. E. Zakharov (Amer. Math. Soc. Trans. Ser. 2, vol. 182, 1998) pp. 1–30.

Paper details

vlnc:W_w35hrdac
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents