PhysPlasmas 16 082504

PHYSICS OF PLASMAS 16, 082504 2009
Toroidal ow and radial particle ux in tokamak plasmas

J. D. Callen,a A. J. Cole, and C. C. Hegna
University of Wisconsin, Madison, Wisconsin 53706-1609, USA
Received 16 April 2009; accepted 29 July 2009; published online 17 August 2009 Many effects inuence toroidal ow evolution in tokamak plasmas. Momentum sources and radial plasma transport due to collisional processes and microturbulence-induced anomalous transport are usually considered. In addition, toroidal ow can be affected by nonaxisymmetric magnetic elds; resonant components cause localized electromagnetic toroidal torques near rational surfaces in owing plasmas and nonresonant components induce global toroidal ow damping torque throughout the plasma. Also, poloidal magnetic eld transients on the magnetic eld diffusion time scale can inuence plasma transport. Many of these processes can also produce momentum pinch and intrinsic ow effects. This paper presents a comprehensive and self-consistent description of all these effects within a uid moment context. Plasma processes on successive time scales and constraints they impose are considered sequentially: compressional Alfvn waves GradShafranov equilibrium and ion radial force balance, sound waves pressure constant along a eld line and incompressible ows within a ux surface, and ion collisions damping of poloidal ow. Finally, plasma transport across magnetic ux surfaces is induced by the many second order in the small gyroradius expansion toroidal torque effects indicated above. Nonambipolar components of the induced particle transport uxes produce radial plasma currents. Setting the ux surface average of the net radial current induced by all these effects to zero yields the transport-time-scale equation for evolution of the plasma toroidal ow. It includes a combination of global toroidal ow damping and resonant torques induced by nonaxisymmetric magnetic eld components, poloidal magnetic eld transients, and momentum source effects, as well as the usual collision- and microturbulence-induced transport. On the transport time scale, the plasma toroidal rotation determines the radial electric eld for net ambipolar particle transport. The ultimate radial particle transport is composed of intrinsically ambipolar uxes plus nonambipolar uxes evaluated at this toroidal-rotation-determined radial electric eld. 2009 American Institute of Physics. DOI: 10.1063/1.3206976
I. INTRODUCTION
Determining the magnitude, radial prole, and evolution of toroidal ow in tokamak plasmas is an important issue for both present tokamak plasmas and ITER1for E B ow shear control of anomalous transport, prevention of locked modes, control of edge localized modes via resonant magnetic perturbations, etc. Many effects2 inuence the evolution of toroidal ow in tokamak plasmas. Radial transport of toroidal ow due to collision-induced37 and microturbulence-induced anomalous processes obtained using mean-eld theory813 is usually considered within the context of an equilibrium axisymmetric magnetic eld model. In addition, the toroidal ow is constrained by faster time scale processes and can be affected by threedimensional 3D nonaxisymmetric NA magnetic eld components,1421 magnetic eld transients,2225 and externally supplied toroidal momentum sources e.g., from neutral beam injection26 NBI. Heretofore these diverse effects on the plasma toroidal ow have been mostly considered separately and in an ad hoc fashion. In this paper a detailed framework is developed for describing toroidal ow evolution in tokamaks that
a
Electronic mail: callen@engr.wisc.edu. Website: http://www.cae.wisc.edu/ ~callen.
accounts for all these physical effects simultaneously and self-consistently. A neoclassical equilibrium is obtained for time scales longer than the collision time scales. It is briey revisited to add self-consistently possible effects due to microturbulence and momentum sources e.g., noninductive current drives CDs on the lowest order plasma ows and parallel Ohms law in tokamak experiments. The overall toroidal ow equation is developed using an approach originally developed27,28 for a neoclassical description of ows in NA e.g., stellarator toroidal magnetic connement systems, but here extended to include microturbulence, magnetic eld transients, and momentum source and sink effects. NA magnetic eld components in tokamaks arise from coil irregularities, active control coils, and plasma collective magnetic distortions e.g., neoclassical tearing modes and resistive wall modes. The NA components are typically very small compared to the equilibrium magnetic eld: B / B0 103. Thus, they will be considered to be of rst order or smaller in the small gyroradius expansion, which will greatly facilitate analysis of the 3D effects. This paper uses a uid moment approach to develop a comprehensive description of the various effects on the particle uxes, toroidal ow, radial electric eld, and resultant net radial particle ux in a tokamak plasma. An earlier and more limited and abridged version of this work has been
2009 American Institute of Physics
1070-664X/2009/168/082504/18/$25.00
16, 082504-1
Downloaded 18 Aug 2009 to 128.104.1.219. Redistribution subject to AIP license or copyright; see http://pop.aip.org/pop/copyright.jsp
082504-2
Callen, Cole, and Hegna
Phys. Plasmas 16, 082504 2009
presented elsewhere.2 The magnetic eld B, electric potential , and plasma parameters will be expanded in terms of their axisymmetric equilibrium plus gyroradius-small NA perturbations. Both externally induced perturbations and collective plasma uctuations are taken into account. Effects on successively longer time scales will be considered sequentially: on the compressional Alfvn wave time scale, the radial ion force balance yields a relation between the poloidal and toroidal ows within a ux surface and the radial electric eld; on Coulomb collision time scales, the density, temperature, and pressure of each plasma species become constant along a magnetic eld line and ows within a magnetic ux surface become incompressible; and then, for times longer than the ion collision time, the poloidal ion ow is usually damped to a diamagnetic-type ow dependent on the ion temperature gradient. Finally, there are many second order in a small gyroradius expansion and the 3D NA magnetic eld magnitudes radial particle transport uxes induced by toroidal torques on a plasma species, both collision-induced intrinsically ambipolar ones and possibly nonambipolar ones. The collisioninduced ones include classical,3 PrschSchlter PS, banana-plateau,5,6 and paleoclassical29 particle uxes plus ones induced by parallel noninductive and dynamo CDs, and the EA B / B2 pinch. Possibly nonambipolar uxes are caused by NA neoclassical toroidal viscosity NTV ow damping effects,1417,27,28 perpendicular viscosities,3,4,7 toroidal torques on resonant surfaces due to nonideal effects e.g., due to resistivity18 and two-uid diamagnetic ows19 on NA resonant magnetic eld components combined with NTV effects,20 polarization ows, microturbulence,813 magnetic eld transients,2225 and momentum sources. Nonambipolar components of the particle uxes cause radial plasma currents. Setting the ux surface average of the total radial current to zero so, from charge conservation and the time derivative of Gausss law, the radial electric eld does not increase monotonically in time yields2,15,27,28 a transporttime-scale toroidal ow evolution equation. The radial electric eld is determined from the toroidal ow. The net radial particle ux is the sum of the collision-induced intrinsically ambipolar particle uxes and the nonambipolar particle uxes evaluated at the toroidal-ow-determined radial electric eld. We make a number of assumptions to facilitate determining comprehensive equations for the particle uxes and evolution of toroidal rotation in tokamak plasmas. 1 Small gyroradius expansion, which to zeroth order yields magnetohydrodynamic MHD force balance equilibrium, ows within ux surfaces at rst order, and second order radial transport uxes 2 Axisymmetric lowest order magnetic eld structure with nested toroidal ux surfaces i.e., no magnetic islands in region of interest 3 Gyroradius-small magnetic eld nonaxisymmetries such that toroidal nonaxisymmetries NAs in the magnetic eld B are rst order or smaller in the gyroradius expansion 4 Banana-plateau collisionality regime, where electron
and ion collision lengths are long compared to the poloidal periodicity length 2Rq and hence plasma properties are constant on magnetic ux surfaces 5 Gyroradius-small plasma uctuations, which lead to second order microturbulence-induced anomalous radial plasma transport 6 Slow poloidal magnetic eld transients, which at their fastest occur on the transport time scale 7 Poloidal damping of electron heat ow is neglected in obtaining the parallel neoclassical Ohms law No explicit inverse aspect ratio r / R0 expansion will be made, except in estimating the magnitude of some terms and in neglect of some minor O2 terms. While the analysis will be valid over most of a tokamak plasma, because of assumptions 4 and 5, it might not apply to its very edge. In particular, very close to the magnetic separatrix in diverted tokamak plasmas, assumption 4 can be violated, in which case a two-dimensional radial and poloidal description of plasma transport is needed. Also, in the edge of L-mode tokamak plasmas uctuations can become so large that assumption 5 is violated, in which case a uid-type turbulent plasma description may be needed there. Because of assumption 7, while scalings of all the collisional radial particle uxes will be correct, some of their numerical coefcients will only be approximate. Finally, for simplicity we restrict our analysis to a two species plasmaelectrons plus hydrogenic ions. This paper is organized as follows. Section II species the plasma and magnetic eld models and perturbation procedure. Section III discusses the successive time scales and the constraints they impose. The radial particle uxes induced by the various effects indicated above and the plasma toroidal ow evolution equation that results from setting to zero the net radial currents from their nonambipolar components are developed in Sec. IV. Appendix A describes the relation of uid ow velocities to guiding center velocities, and Appendix B describes the gyroviscosity and its effects. Some interesting properties of the various effects in the evolution equation for the plasma toroidal rotation are discussed in Sec. V. Finally, Sec. VI summarizes the main results obtained in this paper.
II. PLASMA AND MAGNETIC FIELD MODELS
We use a plasma description based on uid moments of a general plasma kinetic equation for each species, which includes the Vlasov operator, the Coulomb collision operator, and sources. Some of the possible plasma particle and momentum sources and sinks are those due to collisions with neutrals, fast ions injected via energetic neutral beams,26 current sources, and radio-frequency waves interacting with the plasma. Such a description is sometimes called a two-uid description. However because we are considering the banana-plateau collisionality regime, we will use the kinetically derived closure relations for the typically rst order in the small gyroradius expansion neoclassical parallel viscous forces B0 .6 The magnetic eld description will be based on a lowest order axisymmetric magnetic eld, which is composed of nested magnetic ux surfaces. That is, any
082504-3
Toroidal ow and radial particle ux

2 0 d gf , 2 0 d g
Phys. Plasmas 16, 082504 2009

2 0 d f , /B0 2 0 d/B0
NA magnetic eld perturbations will be assumed to be small enough that they do not cause magnetic islands in the radial regions being considered. The dimensionless radial coordinate will be based on the toroidal magnetic ux, as is customary for tokamak plasma transport modeling codes. The rst order perturbations will include both the average axisymmetric poloidal variations in the equilibrium PS effects and effects due to zero-toroidal-average, gyroradiussmall perturbations. The conservative forms of the density and momentum force balance equations for each plasma species obtained from the d3v1 , mv uid moments of a full plasma kinetic equation including sources are in laboratory coordinates x
Flux-surface-averaging is an annihilation operator for the parallel derivative along a magnetic eld line, B0 f = 0. 6
In addition, the FSA of the divergence of a vector function C , is C = 1 d d C V = VC , dV V d 7
n/ tx + nV = Sn ,
nV + mnVV
x
= nqE + V B p + R + Sm .
Here, V is the species ow velocity, is the viscous force density, R Re neeJ is the Coulomb collision dynamical friction force density, and the other notation is standard.3 The sources of density Sn and momentum Sm will be assumed to be small and to mostly contribute at the transport level i.e., at second order in the gyroradius expansion. Also, the species label s has been suppressed for simplicity; it will be added as needed. The lowest order axisymmetric equilibrium magnetic eld B0 is composed of toroidal Bt and poloidal B p components. It will be represented by B0 = Bt + B p I + p = p q = t p . 3
3 in which V 0 d x = 20 d2 gd is the volume of the 0 ux surface and V dV / d = 22 gd. Thus, the FSA 0 is f = 2d gf / V. The contravariant base vectors ei ui for the i u = , , coordinate system are e , e , and e . The covariant base vectors ei x / ui are e = g , e = g , and e = g . These nonorthogonal base vectors satisfy Kronecker delta relations: ei e j = ij. Because of toroidal axisymmetry, / R and e = R2 = Re in which e / is e = e the unit vector in the direction. To lowest order in the gyroradius expansion the density n, temperature T, and pressure p nT of both plasma species will be constant5 on the poloidal magnetic ux surfaces p. In rst order we allow for zero-toroidal-average NA 3D perturbations -dependent, denoted by tilde, due to instability-induced uctuations plus those induced by NA externally imposed magnetic eld components and poloidal variations in the -average denoted by an overbar plasma parameters. Thus, we expand spatial dependences of n, T, and p as
p 1 , + p1, , + O2 , px = p0 +
Here, I p RBt is the poloidal current function, 2 p is the equilibrium poloidal magnetic ux, which will be allowed to evolve on the slow transport time scale, is the toroidal axisymmetry angle, q p B0 / B0 = dt / d p is the inverse of the rotational transform of the magnetic eld B0, and is the straight-eld-line on a magnetic ux surface poloidal angle. The Jacobian g of the transformation from laboratory x to the nonorthogonal toroidal magnetic ux coordinates , , and is
in which s / a 1 is the small gyroradius expansion parameter. Here s = vTs / cs is the most probable gyroradius for a species s with thermal speed vTs 2Ts / ms and gyrofrequency cs qsB0 / ms. The electric potential will be expanded similarly. In addition, we dene E = + EA in which EA A / t with A t p for the magnetic eld in Eq. 3. The -average inductive electric eld is O 2 . EA = / t + p t 9
1 d p/ d p g = = , B0 I/qR2
in which p d p / d aRB p. Here, t / ta is the dimensionless radial ux surface label coordinate based on the toroidal ux 2t dS B0 = d3x B0 . It ranges from zero at the magnetic axis to unity at the plasma edge or divertor magnetic separatrix of average minor radius a. Also, Rx = R , is the major radius to the laboratory position x. The ux surface average FSA of a scalar function f , is dened by
t inHere, 2 / t is the vacuum loop voltage Vloop duced by the rate of change in magnetic ux in the Ohmic transformer central solenoid. As indicated, it is spatially constant but a function of time t; it represents a gauge trans represents the inductive formation choice. Further, 2 p Lenzs law voltage induced within the plasma by poloidal magnetic ux transients e.g., from turning on a noninductive term represents toroidal magnetic ux transients CD. The t induced by radial motion and shaping of the plasma. All these slow, transport-time-scale transient effects are discussed in Sec. IV A below.
082504-4
Phys. Plasmas 16, 082504 2009
Because of the toroidal axisymmetry in the equilibrium, the zero-toroidal-average perturbations can be expanded in a Fourier series. For example, nein, p1 = p
n
n p
1 2
dein p1 .
10
The toroidal average of a function Fx is dened by Fx 1 2
dFx ,
11
p1 = 0, which is a function of , . Note that since p x = p 0 + p 1 , + O 2 . 12
The toroidal average of NA perturbations times functions that do not depend on vanishes. However, the toroidal average of products of perturbations does not in general vanish; = n p n usually does not vanish. n p 1 for example, The total equilibrium plus perturbed magnetic eld will be expanded as +B + O 2 . B = B 0 , + B 13
The parallel and perpendicular components are dened relative to the equilibrium magnetic eld B0, A B0 A B2 0 B 0, A B0 B0 A B2 0 . 14
The magnetic eld magnitude can be approximated by

2 2 2 2 1/2 B B = B2 0 + 2 B 0B + B + B
B 0 , + B , , + O 2 .
scale, and toroidal O2, determines radial particle transport uxes and the desired toroidal plasma ow equation. Summing the density and momentum equations in Eqs. 1 and 2 over species, we readily obtain to zeroth order the ideal MHD plasma equations m / t + mV = 0 and mdV / dt = J B P. In obtaining these equations we have neglected viscous force effects, which are rst order in the gyroradius expansion and source effects, which are assumed to be of rst or higher order. Neglecting electron inertia, the lowest order electron momentum equation yields the collisionless two-uid Ohms law E + V B = J B pe / nee. In these equations m is the mass density, V Vi is the plasma ion ow velocity, J niqiVi neeVe is the current density, and P pe + pi is the total plasma pressure. Adding the nonrelativistic Maxwell equations B = 0, B = 0J, and B / t = E and an isentropic equation of state = O2 0, which is derivable from energy d / dtln P / m balance equations, completes the zeroth order plasma description. The fastest time scale MHD processes are compressional Alfvn waves, which propagate perpendicular to magnetic eld lines. On time scales longer than their natural wave periods A a / cA s, together with the condition B0 P0 = 0 P0 = P0 p, they cause tokamak plasmas to come into an MHD force balance equilibrium with J0 B0 = P0 = dP0 / d p p. This relation together with the equilibrium Maxwell equations yields the GradShafranov equation for determining the equilibrium poloidal ux function p . Replacing J0 B0 by P0 = pe0 + pi0 in the -average ideal collisionless two-uid Ohms law yields p i0 , 0 = ni0qiE0 + Vi B0 16
15
are expanded in The magnetic eld perturbations B and B Fourier series analogous to that for p1 in Eq. 10. Perpendicular gradients of instability-induced uctuations will be assumed to scale as 1 / to reect the short radial scale length of drift-wave-type perturbations. Thus, for p1 1 / 0 1. In contrast, parallel gradiexample, ents of uctuations will be assumed to scale with the overall p1 tokamak plasma dimensions and hence as 0; thus, 0 1. Gradients of average plasma properties will p1 0 1, also be assumed to scale as 0. Hence, while we scale p1 1.
III. SUCCESSIVE TIME SCALES AND PROCESSES
in which the lowest order average electric eld is electrostatic: E0 0 p. This MHD equilibrium ion force balance equation can also be obtained directly from the equilibrium limit d / dt 0 of the ion force balance equation in Eq. 2by neglecting the frictional R and viscous forces and momentum sources Sm , 2, which are all higher order in the gyroradius expansion. Taking the radial e projection of Eq. 16 yields t Vi =
1 dpi0 d0 + + qVi . d p n i0q i d p
17
The plasma toroidal ow evolves on the long transport time scale seconds. Its evolution arises from effects that are formally second order or smaller in the gyroradius expansion. To obtain an equation for evolution of the plasma toroidal ow on this long time scale, we need to take account of faster processes and constraints they impose on plasma behavior. To analyze the various time scale processes, we will analyze sequentially three independent but not orthogonal components of the momentum force balance equation: radial O0, enforces radial force balance on fast compressional Alfvn time scale, parallel O, determines rst order ows on a ux surface on intermediate ion collision time
This equation provides a relation between the average Vt / R, lowest designated by overbar toroidal ow Vi 0 = d0 / d order radial electric eld E e , toroidal ion diamagnetic ow = d0 / d p p 1 / ni0qidpi0 / d p, and poloidal ion ow Vi V p / r. However, it does not specify any of these quantities. Since in obtaining Eq. 17 we divided through by qiB0, the poloidal and toroidal ion ows are rst order in the gyroradius s 1 / qsB0. Thus, these ows within tokamak ux surfaces are rst order in the small gyroradius expansion; however, for simplicity of notation we will not always indicate this by adding a subscript 1 to these ows. Transport ows in the radial direction i.e., V will be of second or higher order in the gyroradius expansion.
082504-5
Phys. Plasmas 16, 082504 2009
Coulomb collisions cause electrons to thermalize become Maxwellian on the electron collision time scale 1 / e 10 s for Te 1 keV, ne 3 1019 m3, and Zeff 3. Similarly, ions thermalize on the ion collision time scale 1 / i ms for Ti 1 keV. In doing so they cause their corresponding species temperatures to equilibrate along magnetic eld lines over distances of order the collision length vT / 102 m. From assumption 4 above this causes5 the species density and temperature to become constant on poloidal ux surfaces on the collision time scale of the species: n0 = n0 p and T0 = T0 p for t 1 / . On this same time scale the species ow velocity becomes a denable and physically meaningful quantity. Next, consider the form of the density equation 1 on time scales longer than the species collision time. Since the lowest rst order equilibrium ows lie within a ux surface, they can be written in terms of their poloidal V and toroidal V components as V 1 e V + e V . 18
no radial component and the equilibrium B0 is axisymmetric / 0, this reduces to B0
V1 = 0. B0
23
The solution6 of this partial differential equation is that a poloidal ow function U is constant on a ux surface, U p V1 V V = + . B0 B0 B0 24
Using the poloidal component of rst order E0 B0 and diamagnetic ows within a ux surface from Eq. 20 yields 1 dps0 V I d0 I = 2 + 2 , B 0 B 0 d p n s0q s d p B0
25
in which p = is constant on a ux surface. The currents owing within the ux surface can be similarly determined. Using Eqs. 19 and 20, we obtain J n s0q sV s1 J + J = J
s
This representation applies to either electrons or ions. Alternatively, the equilibrium ows within a ux surface can be represented in terms of their components parallel to and cross , perpendicular to B0 but within a ux surface the equilibrium magnetic eld B0, V1 = V B 0/ B 0 + V . 19
B0 B0 P0 + , B0 B2 0
26
which is the usual sum of the parallel and diamagnetic current densities. Summing nq times the poloidal ow components in Eqs. 24 and 25, or using the fact that for a quasineutral plasma the current density is also incompressible J = 0, yields K J p J J I dP0 = + 2 . B0 B0 B0 d p 27
The equilibrium cross ow V in each species is obtained by taking the cross product of B0 with the lowest order equilibrium / t 0 momentum equation 2, B0 p d0 1 dps0 V s = + . 2 d p n s0q s d p B0
20
The constant KJ is obtained from the FSA of B2 0 times this equation: KJ = I dP0 B 0J . 2 + B 0 B2 0 d p B 0J B 2 0 B2 0 B2 dP0 0 1 2 . d p B 0 28
The two terms in parentheses represent the usual equilibrium E0 B0 VE and diamagnetic V ows. The comparable rst order perturbed cross ow for each species is obtained by taking the cross product of the lowest order perturbed momentum equation from Eq. 2 with B0, +V . = 1 B 1 + 1 p s1 V V s 0 E n s0q s B2 0
Using this form in Eq. 27 yields B 0J = I
29
21
B and The terms in parentheses represent the perturbed E 0 diamagnetic ows. Since the rst order perturbed ows have a nonzero radial component, they lead to the anomalous ra . The relation of the uctuating dial particle ux n 1V uid ow velocity V to the guiding center ow velocity usually calculated in gyrokinetics see, for example, Ref. 12 is discussed in Appendix A. Because the equilibrium density n0 only changes on the transport time scale / t 2 and the radial transport uxes and density sources are second order, the lowest order average density equation 1 reduces to V1 = 0 + O .
2
Here, B0J is the FSA parallel current, which will be determined from a parallel Ohms law below and the last term represents the PS current whose FSA vanishes. Similar considerations and orderings of the species energy balance equation and the heat ux uid moment equation yield analogous formulas6 for the rst order equilibrium heat ows within a tokamak ux surface, Q p q1 q q = + , B0 B0 B0 30
qs 5 I dTs0 = . B 0 2 q sB 2 0 d p
31
22
The divergence of a vector C is dened by C = 1 / gi / uigC ei. Thus, using the facts that V1 has
Finally, the FSA of the parallel B0 component of the lowest order parallel ion heat ux equation6 is B0 Rqi = 0 + O2 in which Rqi iqi is the ion heat friction force. Solving Eq. 30 for qi and substituting it in this lowest order relation yield6
082504-6
Phys. Plasmas 16, 082504 2009
Q i =
1 B2 0
B2 0
q i B0
5 ni0Ti0I dTi0 . 2 q i B 2 0 d p
32
B0J = B0 EA + /ne0eB0 e + B0 JCD + B0 Jdyn . 36
The rst order average parallel force balance for each species is obtained from the parallel B0 component of the momentum equation. Its most convenient form here is obtained by using the density equation 1 to write m / tnV + mnVV + = mn V/ t + V V + + mVSn . 33
The viscous stress tensor has components along B , across magnetic eld lines but within ux surfaces , and perpendicular to ux surfaces ,3
= 0 + + 2 .
34
The FSA parallel electron viscous force can be written as6 B0 e mene0e00B2 0Ue in which the neglected poloidal electron heat ow damping effects induced by Qe only modify some numerical coefcients. Here, e00 2.24e is the viscous damping frequency6 for Zi = 1 and 1 of the poloidal electron ow Ue in which is the inverse aspect ratio, which here represents the variation in B0 = B0 , on a ux surface: Bmax Bmin / Bmax + Bmin r / R0. Using the denition J = ne0eVi Ve and Eqs. 24 and 25 for both electrons and ions, the parallel viscous force contribution to the parallel Ohms law in Eq. 36 can be written as
For the parallel force balance the most important viscous force will be the parallel component of , which we will discuss below. The gyroviscous stress indicates a diamagnetic-type B V effect. The gyroviscous force it induces is higher order for equilibrium ows. However, it is signicant for uctuating ows see Appendix B. The forces due to perpendicular viscous stresses are higher order and will not contribute to the rst order parallel momentum equation. Thus, taking the FSA of the B0 parallel component of the average of the momentum equation in Eq. 2 with all quantities expanded as in Eq. 8, employing Eq. 33, and 1 and eliminating p1 terms using Eq. 6 yield the O parallel force balance equation for each species, mn0
e00 B0 e B0J + B0 Jbs . n e0e e

Here, the bootstrap current is dened by B0 Jbs
37
e00 I dP0 ne0eUiB2 0 . e p d
38
Thus, in Eq. 36 the parallel electron viscous force in Eq. 37 produces6,29 both the trapped particle modication e00 e of the Spitzer parallel electrical conductivity and the bootstrap current. The noninductive CD in Eq. 36 is induced by electron momentum and particle sources, Sen . B0 JCD /ne0eB0 Sem meVe 39
B 0 V = n 0q B 0 E A B 0 + B 0 R t
Sn + B0 Sm mV V + B0 mn0V 1 1 B . + n 0q B 0 V 35
Most noninductive CDs e.g., due to electron cyclotron CD result from the average electron parallel momentum B0 Sem they impart to the plasma. Fluctuation-induced electron Reynolds- and Maxwelltype stresses cause a dynamo-type parallel current, B0 Jdyn =
V + B 0 m en e0V e e e n e0e
B . + B0 V e 40
V term represents the force caused by the The mn0V 1 1 Reynolds stress introduced by the uctuations. To lowest order the -average gyroviscous force produces see, for example, Ref. 30 the gyroviscous cancellation V contribution to the Reynolds stress force of the V see Appendix B. The Maxwell-stress-type term B = n qB 1 + 1 / n 0q n 0q B 0 V p1 will yield 0 dynamo-type contributions to the parallel Ohms law below. Both of these contributions are typically small for driftwave-type uctuations in the hot core of tokamak plasmas.12,31 The electron Coulomb collision frictional force density is given by3 Re ne0eJ / + J / . The perpendicular and parallel Spitzer electrical conductivities are ne0e2 / mee and 1.96 for Z = 1. Thus, the equilibrium limit / t e of the electron momentum equation 35 yields an equation for the FSA parallel current5,6 in the plasma,
In tokamak plasmas the electron Reynolds stress and dynamo-type terms are usually negligible.31,32 Taking account of the bootstrap current and trapped particle effects from the parallel electron viscous force in Eq. 37, Eq. 36 becomes a neoclassical parallel Ohms law that includes noninductive and dynamo CDs, B0 EA = nc B 0J 1 / B 0 J bs + B 0 J CD + B0 Jdyn . Here, the neoclassical parallel electrical resistivity is 41
nc
1 e00 1+ . e
42
Summing the electron and ion FSA parallel momentum equations 35 and using the quasineutrality constraint
082504-7
Phys. Plasmas 16, 082504 2009
sns0qs = 0 and the momentum conservation relation for Coulomb collisions between species of sRs = 0 yield the overall plasma parallel force balance
Ui p = Ui0 +
V + B 0 m in i0V i i i mini0i00B2 0 mini0i00B2 0 .
+ B S m V B0 J B 0 s sm s sSsn
B 0 V s B0 i m t
V + s B 0 m sn s0V s s s + B0 J B
46 The additions to the poloidal ion ow Ui0 in Eq. 45 are induced by Reynolds stresses, dynamo-type effects, and ion ow drive effects due to B0 Sim, respectively. They are usually thought to be small in tokamak plasmas see Eqs. 22 and 25 and discussion in Ref. 12. Having determined the average poloidal ion ow Vi = UiB0 , we substitute it into the t expression obtained in Eq. 17 to obtain the more specic toroidal rotation frequency relation for t 1 / i ms, t =
+ B0 s Ssm msVs Ssn .
43
The sums over species are dominated by the ion contributions since the corresponding electron momentum contributions are typically a factor of me / mi 1 / 60 1 smaller. The FSA of the equilibrium parallel ion viscous force, which is O, is given by6 B 0 i m in i0 B 2 0 i00Ui + i01
2 Q i . 5 n iT i0 44
1 dpi0 d0 + + p = i + p . d p n i0q i d p
47
Here, the poloidal ow contribution to t is p , I c pI2 dTi0 U , i R2 q iR 2 B 2 0 d p 48
Here, i00 1.1i and i01 1.17i00 for Z = 1 and 1 in the asymptotic banana collisionality regime6 i.e., for i iRq / 3/2vTi 1. They are damping frequencies for the poloidal ion ow and heat ow. Physically, the ion parallel viscous force B0 i is caused by ion collisional drag by trapped particles f t 1.46 on the untrapped circulating, f c 1 f t ions that carry the parallel poloidal ow. This viscous force damps the poloidal ion ow to an ion temperature gradient diamagnetic-type ow because hotter ions are more collisionless than bulk ions and hotter regions at smaller radii damp less than colder ones at larger radii. Thus, Eq. 43 is an evolution equation for the average parallel ion ow Vi or the average poloidal ion ow function Ui. The parallel and poloidal ion ows come into equilibrium see Ref. 33 and references cited therein on the ion collision time scale t 1 / i ms. In the absence of the usually small parallel momentum sinks and sources on the second through fourth lines of Eq. 43, the equilibrium poloidal ion ow can be obtained by setting the parallel ion viscous force in Eq. 44 to zero, Ui0 I dTi0 i01 2 Q =c . i00 5niTi0 i p qiB2 0 d p 45
in which the last approximate form neglects the usually small parallel momentum sink and source terms. Since for rigid body rotation t must be a ux function, tokamak plasmas rotate toroidally as a rigid body only when the poloidal ow is negligible so t i p. The neoclassical literature contains a number of forms for the average ion ow velocity V1 in addition to those given in Eqs. 18 and 19. In particular, using the denition of at the end of Eq. 25 in Eq. 47, one can write the rst order average ion ow within a ux surface as6 V i = ie + U iB 0 = i p R 2 + U i p B 0 . 49
The projection of this representation of Vi readily yields the toroidal rotation frequency given in Eq. 47 above. Also, the plasma parallel ow speed can be written as V B 0 V i/ B 0 = i p I p / B 0 + U i p B 0 R
R2 1 dpi0 d0 0 dTi0 +k . + q iR d p d p n i0q i d p
50
The poloidal ow coefcient k = c p i01 / i00 is 1.17 for i 1 and 1. However, c p depends on the ion collisionality regime and ;6 with impurities it also depends on gradients of impurity densities and temperatures.34 It is often evaluated using the NCLASS code.35 When sinks and sources on the second and third lines of Eq. 43 are signicant, they cause a poloidal ion ow of
In Eqs. 4850, R R , R01 + cos . For given radial ion pressure and temperature gradients, Eq. 47 provides a consistency relation between the plasma Vt / R and the lowest order ratoroidal rotation t V dial electric eld E pd0 / d p. However, it does not specify either of these quantities. To proceed further we need to determine either the radial electric eld or the toroidal owfrom their own evolution equations. We proceed by calculating the radial particle transport uxes and hence net radial current in the spirit of seeking to calculate the radial electric eld.
082504-8
Phys. Plasmas 16, 082504 2009
IV. TRANSPORT TIME SCALE
The collision-induced radial particle uxes of electrons and ions are mostly determined from the second order ux n0V2 p. To determine them we rst note that e V B0 = V e B0 = V p = pV . 51
Thus, the radial particle ux can be determined from the toroidal angular component e of the force balance Eq. 2, i.e., from the toroidal torques on the plasma species. Next, consider the e projection of the mnVV term in Eq. 2. Employing a vector, tensor identity for a general tensor T, it can be written in the form e T = e T e : T . 52 Using a cylindrical coordinate system R , , Z with R = 0 R and e being the symmetry axis, since R = e e / = e 1 / R e R, one can show that =e = e Re e e R. e = Re 53 Since this is an antisymmetric tensor, its contraction with any symmetric tensor TS vanishes: e : TS = 0. Hence, for any symmetric tensor TS, Eq. 52 simplies to e T S = e T S . 54
To do so, we need to determine our density and toroidal ow equations on poloidal ux surfaces p, upon which drift kinetics and gyrokinetics are basedbecause Grad Shafranov equilibria determine the poloidal ux surfaces, and neoclassical transport22 and microturbulence-induced anomalous transport are determined relative to them. Magnetic eld transient effects were not considered above because they are O2; hence they were negligible in obtaining the O0 radial force balance and O parallel force balance and ows within ux surfaces. Equations for the slow transport-time-scale evolution of the poloidal p and toroidal t magnetic uxes can be obtained from Faradays law. Representing the magnetic eld by B = A, Faradays law can be written as A / t x + EA = 0. Using the vector potential A = t p for the B0 eld given in Eq. 3, the temporal evolution of the toroidal and poloidal magnetic uxes at a given laboratory position x are

t t p t
= e EA =
qR2B p EA , I
57
= + e EA
x
= R2 EA . t t
58
Therefore, since mnVV is a symmetric tensor, the toroidal component of the terms on the left side of the momentum equation 2 can be written as e / txmnV + mnVV = / txmne V + mne VV . 55 Thus, the average of the toroidal e component of the momentum equation 2 for each species with all quantities expanded as in Eq. 8 can be written in laboratory coordinates x to lowest order as
/ txmn0e V1 + mne V1V1

1V = n0qe EA + V2 p + qn 1 p B e +e R+e S . + n 0q e V 1 m 56 After taking into account the requirement for ambipolarity of net radial particle uxes, the FSA of Eq. 56 will produce an equation for the toroidal angular momentum density mn0e V1 and hence t.
Here, / t Vloop t / 2 is a spatial constant gauge variable that is usually positive see Eqs. 6668 below to drive current in the plasma. It represents the toroidal loop voltage induced in vacuum by the rate of change in magnetic ux in the central solenoid Ohmic transformer of a tokamak. In tokamak modeling codes that include ux surface evolution,2325 it is usually taken to be zero; then the transformer-induced ux change is imposed as a boundary condition on the plasma domain and causes the poloidal ux to increase linearly with time for an Ohmically heated tokomak plasma in steady state. The / t 0 term is introduced here so that, in near steady-state conditions, the poloidal ux function p , t represents poloidal ux changes within the plasma and does not increase approximately linearly with time t. Multiplying these equations by 1 / R2, taking their averages and ux-surface-averaging them yields

t t p t
t q B p E A t , uG 2 IR
59
=
x
E A . R2 t
60
A. Transient magnetic eld effects
The poloidal and toroidal magnetic elds in tokamaks evolve during plasma startup and the approach to steady state on the slow magnetic skin diffusion time scale. Also, noninductive CD and other poloidal ux transient effects, particularly radially localized ones such as electron cyclotron heating ECH and CD, can cause transient poloidal eld effects on transport time scales of the plasma thermodynamic parameters n, T, and e V1. Thus, we need to allow for transient magnetic ux effects in analyzing Eq. 56.
The small radial grid speed uG of toroidal ux surfaces relative to laboratory coordinates x dened in Eq. 59 is23 uG uG = B p E A
2 pI R
61
Since our radial coordinate is based on the toroidal ux surfaces, this grid speed represents the radial speed of our coordinate system relative to xed laboratory coordinates x. It is caused by inductive poloidal electric elds produced by the poloidal magnetic eld system as it moves the toka-
082504-9
Phys. Plasmas 16, 082504 2009
mak plasma radially or changes its cross-section e.g., ellipticity and triangularity in the = constant plane. Note also that on a ux surface the time rate of change in the toroidal ux surface differential volume V dV / d is given by23,25 the compressibility of the toroidal ux surfaces,
1 V V t
= u G =
1 VuG . V
62
Using the relation B p = B0 I , Eq. 61 can be rearranged to yield for the toroidal electric eld u G EA/R2 = B0 EA/IR2 p. 63
The parallel electric eld B0 EA in the plasma is obtained from the parallel Ohms law in Eq. 41. The parallel plasma current can be written in terms of p as5,29
0B0J = IR2+ p ,
in which the second order cylindrical-type operator is + p
64 2 V p 1 p . 2 R I 65
netic ux surfaces, so the guiding center canonical toroidal angular momentum pg mvI / B q p naturally emerges as a constant of motion. Transforming the drift-kinetic equation and hence its uid moments in Eqs. 1 and 2 from laboratory to p , , coordinates requires taking account of the fact that the equation for the poloidal magnetic ux p in Eq. 66 includes possible transient and grid motion effects and poloidal ux diffusion. A transformation procedure based on the mathematical characteristics of the drift-kinetic equation on poloidal ux surfaces has been developed;36 also, a simplied model illustrating its key physics has been presented.37 In response to a Comment38 on the derivation in Ref. 36, a more systematic derivation based on an analysis of the second order radial guiding center motion was published.39 The net result36,37,39 of this coordinate transformation is that a paleoclassical diffusion-type operator D f is added to the right side of the drift-kinetic equation, which then becomes
2 R V
I

f t
f = C f + D f . + v + v d f + E E p
70
Substituting Eq. 63 into Eq. 60 and using the neoclassical parallel Ohms law in Eq. 41 and the denition of B0J in Eq. 64 yield a diffusion equation for the poloidal ux,29

p t
D
Here, in general we have f = f p , , , , E , t. For a lowest order distribution function f that is constant on a ux surface i.e., f f p , , E , t, the FSA of the paleoclassical operator can be written using Eq. 7 as39
= D + p S u G p.
x
66
f 1 2 D f V D + f u + G f . 71 p V 2
indicates motion in the toroidal-ux-based coorHere, p dinate as the radial position of a given p surface changes during poloidal ux transients within the tokamak plasma. Since remaining on a given p surface during transients requires d p = 0 = d p / + dt p / t, the motion required to remain on the same t toroidal ux surface is36,37,39 / . p p p 72
Here, the magnetic eld diffusivity D and parallel CD sources S of poloidal ux are29,36,37
nc , 0
67
S =
1/ + B0 Jbs + JCD + Jdyn . t IR2
68
As indicated in Eq. 59, the toroidal ux moves only slightly relative to laboratory coordinates due to the usually small grid speed uG. In contrast, the poloidal ux is much more mobile; it moves radially in response to nonideal MHD effects from the parallel electric eld B0 EA caused by and B0J, the bootstrap current, and noninchanges in nc ductive and dynamo CDs. Thus, tokamak modeling codes23,25 usually use a magnetic ux coordinate system based on the toroidal ux in the plasma region, as we have in this paper. Taking into account Eq. 59, the temporal change in the poloidal ux on a toroidal ux surface is given by p p t
The second term in Eq. 71 indicates grid coordinate speed effects as indicated in Eq. 62. The nal term represents paleoclassical transport effects due to the guiding centers of charged particles being carried along with the diffusing poloidal magnetic ux.36,37,39 The normalized magnetic diffusivity in Eq. 71 has units of s1,29 D , D a2 2 1 1 . R2 a2 R2
73
= D + p S .
69
The density and momentum equations in Eqs. 1 and 2 are written in terms of laboratory coordinates x. However, drift-kinetic and gyrokinetic analyses used for obtaining collision- and uctuation-induced transport uxes are performed22 in terms of poloidal magnetic ux coordinates p , , . In particular, they are performed on poloidal mag-
Here, a is an effective minor radius of the ux surface. The scaling of the operator D f in Eq. 71 is determined from the scaling of its diffusion coefcient, which is the magnetic eld diffusivity from Eq. 67: D nc / 0 e 2 e in which e c / p is the electromagnetic skin depth. Since e is of order the ion gyroradius, the terms D f introduced in the drift-kinetic equation are O2; they have been negligible up to now because they were not needed in the O0 , 1 analyses in Sec. III.
082504-10
Phys. Plasmas 16, 082504 2009
B. Density conservation equation
Before solving the -average toroidal momentum equation 56 for the collisional radial particle ux n0V2 p, we discuss the net particle ux that arises in the density conservation equation. The relevant density conservation continuity equation can be obtained from the FSA of the d3v, moment of the drift-kinetic equation in Eq. 70. First, we note that25 for a lowest order Maxwellian distribution function f M p that is constant on a ux surface, 3 d v f M / t p = n / t p = 1 / V / t pVn0. Also, the FSA of the D f contribution to the right side of the density f M = Dn0. Thus, taking the d3v equation is d3vD moment of the drift-kinetic equation in Eq. 70, averaging over toroidal angle and taking its FSA using Eq. 7, we nd that in terms of the toroidal-ux-surface-based coordinate , the O2 density conservation equation on a poloidal ux surface p is

1 V t
and neoclassical transport theory determines particle uxes across poloidal ux surfaces, it uses a toroidal ux coordinate system t that moves with grid velocity uG. All the contributions to the particle ux in Eq. 75 are O2 or smaller in the gyroradius expansion. The units of are m3 s1 because is dimensionless and 1 / a m1. Note also that the particle uxes are dened relative to here rather than to p as was used in Ref. 2; thus, the p p = p in Ref. 2 is p times the dened here. The toroidal force balance equation 56 also needs to be transformed from laboratory to poloidal ux coordinates. Analogous to the effects in the density equation see discussion between Eqs. 69 and 74 above, the transformation adds a term e Dd3vmv f = e Dmn0V1 to the right side / = R of Eq. 56. Then, since the unit vector e does not change with and hence e / = 0, we nd that the FSA of this contribution becomes e Dmn0V1 p
V n 0 + p
n0 1 + V = S n . V
74
mn0e V1
This density equation can also be obtained by taking the FSA of the average of Eq. 1 with all quantities expanded as in Eq. 8 and adding Dn0 to its right side to take account of its transformation from laboratory x to poloidal ux coordinates. The quantity dN Vn0d is the number of particles contained between the and + d ux surfaces. The quantity Vn0 is an adiabatic plasma property5,6,2325 that is conserved in the absence of particle uxes, density sources, and effects caused by poloidal ux changes induced by p 0.
p
+ mn0e V1uG + 1 2 mn e V . V D 0 1 V 2 80
For each plasma species the particle ux in Eq. 74 is
Here, the approximate equality indicates that O2 inverse aspect ratio terms have been neglected in bringing the R through the derivatives. factor in e Re Thus, transforming Eq. 56 to poloidal ux coordinates, taking its FSA, and solving it for the second order radial particle uxes for each species yield
a na n 1V p p + an n0V2 uG p +
a na a + an + pc .
75
It includes the direct second order ambipolar superscript a collision-induced ux,

a n0V2 uG ;
1 = e R + n0qe EA e q 1 e Dmn0V1 q +
76
possibly nonambipolar superscript na anomalous subscript an uxes induced by correlations of rst order density uctuations and uid ows,
na ; n 1V an

1 qV t
Vmn0e V1
77
1 V Vmn0e V 1 1 qV 81
and the ambipolar paleoclassical particle ux given by

a = pc
1 n =D n0 n V . V D 0 0 pc V
1
78
B 1 e S . e n 0V 1 m q
Here, the paleoclassical particle pinch speed s is Vpc = 1 . V D V 79
The grid velocity uG has been incorporated into the denia tion of the collisional part of in Eq. 76 via V2 5,6,22,23 collisional classical V2 uG because while
Here, Eq. 7 has been used to obtain the uctuation-induced contribution to the particle ux on the fourth line. Successive lines in Eq. 81 indicate that the radial particle ux has many O2 components: collision-induced uxes, collisioninduced paleoclassical processes, polarization ux due to time-dependent toroidal ows, microturbulence-induced JB Reynolds stresses, effects due to the Maxwell stress induced by uctuations and NA magnetic eld components, and toroidal momentum sources.
082504-11
Phys. Plasmas 16, 082504 2009
C. Ambipolar collisional particle uxes
Consider rst particle uxes induced by the collisional friction force Re neeJ / + J / = Ri neglect of poloidal heat ows changes some numerical coefcients but does not affect the qualitative results or scalings,
a p
CD
n e0I 2 B0 JCD , p B 0 n e0I 2 B0 Jdyn , p B 0
91
dyn
92
1 e R s n 0 e E A . qs
82 EA
A vector identity relating the toroidal, parallel, and poloidal directions is useful for analyzing e Rs, e R2 = IB0 B0 p . B2 B2 0 0 83
I B p E A n e0 I2R2 e E A 1 . 93 B2 B2 0 0 p
Note that all seven of these particle uxes are the same for electrons and ions.5,6,40 Thus, they are ambipolar superscript a and cause no FSA radial plasma current,
a a J a q s s + spc = 0. s
The collisional-friction-induced toroidal torque is thus 1 I B0 R e R = qs qs B2 0 =
n e0I B 0J B2 0

1 B0 p R qs B2 0 +
94
ne0 p2 dP0 . d p B2 0
84
The FSA J term can be worked out using Eqs. 29 and 36,

B 0J B2 0
= =
B 0J + B2 0

JB0 B2 0
1 B2 0
1 B2 0
B0 EA + /ne0eB0 e B2 0
+ B0 JCD + B0 Jdyn
For a Ficks law particle ux form = D n = D2dn0 / d, the collision-induced diffusion coef2 cients are of the order Dcl e 2 e , DPS q / Dcl, Dbp 2 2 2 2 3/2 e00 e Bt / B p q / Dcl, and Dpc nc / 0 D cl / e, which is the largest D for the usual situation where e 3/2 2 pe / B2 0 / 20 / q 1. In addition there are nondiffusive radial uxes due to the Ware pinch5,6 B0 EA, from bp via the parallel viscous force in Eq. 37 using Eq. 41, paleoclassical pinch ow 79, outward ows due to noninductive co-CD and dynamo effects, and EA B0 / B2 0 inward pinch.5
B2 dP0 1 0 1 I d p B2 B2 0 0
D. Possibly nonambipolar particle uxes
85
Next, consider the particle ux induced by the toroidal e component of the viscous force density ,
Next, we use this result in Eq. 84 to obtain the total collision-induced particle ux from Eq. 75, which has seven contributions: classical cl, PS, banana-plateau6 bp collisionality regime, and paleoclassical pc uxes plus the radial particle uxes due to the noninductive and dynamo parallel currents B0 JCD and B0 Jdyn, and the 5 EA B0 / B2 0 pinch effect,
a a + pc = cl + PS + bp + pc + CD + dyn + EA . 86
p 1/qse .
95
The viscous stress tensor was described in Eq. 34 above. The parallel stresses, which are in principle the largest, can be written in the ChewGoldbergerLow form
= p pBB/B2 I/3 .
96
They can be split into their axisymmetric superscript A and NA superscript NA parts, which when averaged over the uctuations can be written as
NA = A . +
Here, for each species we have dened ne0 2 dP0 cl , d B2 0 PS
B2 n e0I 2 1 0 1 2 B2 B2 0 0 p
97
87 dP0 , d
Using equilibrium average quantities in the denition of in Eq. 96 and J0 B0 = B0 B0 / 0 = P0 yields A = B 0 B 0
88
p p B2 0
p p B2 0
B2 0 + 0 P0 2
bp
I B0 e , eB2 0 p
89
1 p . p 3
98
dne0 n V , pc D e0 pc d
90
Taking the toroidal angular e component of this viscous force, the terms on the second line vanish because of axisymmetry / 0 in the equilibrium. Thus, using Eq. 6, the
082504-12

A e = 0.
Phys. Plasmas 16, 082504 2009
FSA of the toroidal viscous torque in an axisymmetric magnetic eld vanishes,5,6 e A B0 = I
102

p p B2 0
= 0.
99
Alternatively, since A is a symmetric tensor, one can use the vector, tensor identity in Eq. 54, and the FSA property in Eq. 7 to obtain the same result. Physically, this toroidal viscous force vanishes because there is no toroidal inhomogeneity in the equilibrium magnetic eld strength to impede ow in the toroidal axisymmetry direction. However, NA 3D magnetic eld components in Eq. 15 cause toroidal torques on toroidally owing plasmas that are typically second order in B and hence in the gyroradius expansion. They are usually calculated1517 in the absence of uctuation and momentum source effects from the NA NA relation e s = q s ps. The net ion torques, which are typically a factor of order mi / me1/2 60 larger than the electron ones, can be written in the generic form2,1417,41
2 Beff
e NA i = mini0it
B2 0
R2t R2 .
100
Here, it s1 is an effective ion viscous toroidal damping frequency due to transit-time magnetic pumping,14,15 banana-drift,16,41 and ripple-trapping effects see Ref. 17 and references cited therein. Moreover Beff / B02 is a weighted sum over all m , n components of B, which represents the effective magnetic eld NA through which the tokamak plasma ows toroidally. The intrinsic or offset42,43 rotation frequency R2/R2 c p + ct1/qidTi0/d p 101
However, for uctuations it does not vanish. That is, e 0 see Appendix B. The ion perpendicular stresses cause both classical3 and neoclassical4,7 radial transport of the plasma toroidal angular rotation t through their respective perpendicular viscous cl nc forces e s and e s. They cause diffusive radial transport of t with diffusion coefcients classical3 4 2 nc 2 2 cl t 0.3i i and neoclassical t 0.1q i i . It is important to note that the neoclassical toroidal angular momentum 3/2 smaller than the ion diffusivity nc t is a factor of order 0.1 neoclassical radial ion heat diffusivity5 i q2 / 3/2i 2 i . In addition, neoclassical radial transport of toroidal momentum includes a momentum pinch type effect4,7 that is comparable in magnitude to the diffusive radial transport of toroidal momentum. In most tokamak plasmas of fusion interest, the classical and neoclassical toroidal angular momentum diffusivities t are less than 0.1 m2 / s; hence the radial transport of t they induce is usually negligible. The combination of all these cross and perpendicular classical and neoclassical viscous stress effects is discussed in detail in Ref. 7. Paleoclassical processes due to the e Dmn0V1 dened in Eq. 80 also produce radial transport of toroidal angular momentum. The D-induced part of electron and ion paleoclassical effects can be characterized as perpendicular viscous torques on the toroidal angular momentum density of each species. The ion paleoclassical viscous torque is larger than that for electrons by mi / me 3672 1. It is caused by the ion D-induced term in Eq. 80, e ipc = 1 2 L V D t V 2 1 Lt + L V V D t pc V
103
is a diamagnetic-type toroidal rotation frequency proportional to the radial ion temperature gradient. It is caused by ions of different energy drifting radially at different speeds. In most NA transport processes, ions on the tail of the ion distribution drift radially more rapidly than thermal ions; this causes 0 for dTi0 / d p 0. The poloidal coefcient c p in Eq. 101 was discussed after Eq. 45; the toroidal coefcient ct ranges from 0.67 to 2.4, depending on which NA process is involved.2,41 Next, we consider the cross and perpendicular stresses and their effects. The cross stress B0 V is a symmetric diamagnetic-type stress tensor see Ref. 30 and references cited therein for which an approximate form is given in Eq. B1. The gyroviscous force it induces is given in Eq. B2. The FSA of its toroidal angular component vanishes for the equilibrium rst order neoclassical ow velocity V1 for up-down symmetric plasmas,7 which we will assume. For up-down asymmetric plasmas it is very small unless there is extreme asymmetry, perhaps near a divertor separatrix, in the PS collisionality regime.7 Also, for collisionless plasmas it vanishes.44 Thus, for equilibrium ows we will assume
in which the plasma toroidal angular momentum torque density is L t m in i0 e V i = m in i0 R 2 t . 104
/ mini0e Vi in Eq. 80 leads to an The ion term p additional toroidal torque density on the right side of Eq. Lt / . It is caused by poloidal magnetic eld 56 of p transients in the plasma that produce p 0 and is analogous n0 / term in Eq. 74. The mini0e ViuG to the p ion term in Eq. 80 contributes, as in particle uxes, to properly determining the neoclassical perpendicular viscosity discussed above when there is grid motion of the toroidal ux surfaces. Paleoclassical radial momentum transport in Eq. 103 causes radial diffusion of Lt with a diffusivity pc t D, plus a momentum pinch Vpc that is the same as that for the density, i.e., Eq. 79. The paleoclassical momentum diffusivity is the same as the magnetic eld diffusivity: D nc / 0 1400Zeff / TeeV3/2, which becomes less than about 0.1 m2 / s for Te 1 keV. While it can be signicant in
082504-13
Phys. Plasmas 16, 082504 2009

na + S . an = + + pol + Rey + Max + JB + p
Ohmic-level tokamak plasmas and toward the plasma edge, it is usually negligible in the hot plasma core unless uctuation-induced transport is suppressed. Taking into account all of the classical, neoclassical, and paleoclassical processes of radial transport of t, the total collision-induced perpendicular viscous torque is
cl nc pc e s e s + s + s pc e s .
115 These eight particle ux components are all, in principle, different for electrons and ions and hence nonambipolar superscript na.
105
E. Radial current and toroidal rotation equation
Combining this with the parallel viscous force due to 3D NA magnetic eld components, the radial particle uxes induced by and viscous forces become
NA p 1/qse s ,
106 107
Next, we consider the FSA of the plasma current continuity equation obtained by multiplying the FSA density equation 74 by the charge qs and summing over plasma species q sns0qs,
p 1/qse s .
The other particle uxes in Eq. 81 are induced by polarization ows, perturbation-induced Reynolds and Maxwell stresses, poloidal magnetic ux transients, and momentum sources,

1 V t
Vq + p
q 1 + VJ = 0. V
116
ppol
1 q sV t
Vmsns0e Vs1 ,
108
pRey
1 Vs , q sV
109 110 111 112 113
Here, VJ sqsVs is the net FSA plasma current in amperes owing radially across a magnetic ux surface. For simplicity, we have assumed that the particle sources are Ssn = 0, because inside tokaambipolar and hence that sqs mak plasmas the electron and ion density sources Ssn are usually equal and hence ambipolar. Thus, using the FSA of Gausss law 0 E0 = q and Eq. 7, this last equation becomes 1 V
pMax e ns0V1 B ,
p JB
NA e n s0V mn
NA , B mn
p mn0e V1 , p p qs
+ p
0 VE0 = 0. V
V 0 E 0 + V J
117
pS 1/qse Ssm .
The toroidal Reynolds stress s has been considered in uid models9,10 but is now usually obtained from gyrokinetic-based theories.1113 The denition of it that emerges from this uid-based approach is V e + e . 114 s msns0 V s s The gyroviscous stress in the last term is described in Appendix B and an approximate form of it is specied in Eq. B4; this term provides a partial gyroviscous cancellation see discussion before Eq. B2 of the rst term in Eq. 114. B contributions into a Also, we have split the V uctuation-induced Maxwell stress part Max and a part JB induced by externally imposed NA resonant at q = m / n magnetic perturbations, which will be discussed after Eq. 120 below. The sum of the rest of the particle uxes in Eq. 81, in addition to those included in Eq. 86, for each species is thus
The radial electric eld contribution here is E0 = 2 0 / . It is proportional to one of the key terms in the toroidal rotation t given in Eq. 47. Hence, the radial ion current due to the ion polarization particle ow ipol in Eq. 108 includes a term that is of the form / t mini0R2 0 / p in Eq. 117. This term is larger than the vacuum eld term 0E0 / t in Eq. 117 by a 2 2 2 2 factor of mini0R2 / 0 p c / cAp 1, in which cAp B p / 0mini0 is the Alfvn speed in the poloidal mag is of a netic eld. The electric eld term proportional to p similar magnitude relative to this inertial electric eld term 2 105 1, both of the from ipol. Thus, since typically c2 / cAp E0 electric eld terms in Eq. 117 are negligible compared to the neoclassical MHD polarizability45 arising from the / tmini0R2 0 / p term in Lt / t. This inertial term produces the neoclassical perpendicular dielectric 2 = c2 / cAp for this plasma when the poloidal ow is in equilibrium on the transport time scale, for which the equilibrium poloidal ow is given in Eq. 46. Thus, as in stellarator neoclassical transport theory,15,27,28 since the E0 terms are negligible in Eq. 117, this equation requires that there should be no net radial current in the plasma on the transport time scale,
082504-14
Phys. Plasmas 16, 082504 2009
na 0 = J = q s s = q s s an . s s
118
F. Rotation, radial electric eld, and net particle ux
As indicated, only nonambipolar particle uxes cause net radial currents. Setting to zero the net radial plasma current caused by contributions produced by all eight nonambipolar particle uxes in Eq. 115 for both species yields 1 V t e
NA i
The FSA toroidal plasma rotation frequency t is dened in terms of Lt by t Lt R 2 t = m in i0 R 2 R 2
V Lt
p
= +
i
inertia
non-res. NA B
cl,neo, paleo
1 d0 1 dpi0 c p dTi0 + . d n i0q i d qi d p
1 dpi0 d0 R 2 p I = + + 2 U i 2 R d p n i0q i d p R
121
1 V
+ e J
ion Reynolds stress
Maxwell stress
Thus, determination of Lt and hence R2t from a solution of Eq. 119 also determines the radial electric eld, E d0 E0 = d
+ e JNA mn
NA B mn
L t/
resonant NA Bmn
p 0 transient
= t p+ 119 t p+
+ e
Ssm .
mom. sources
1 dpi0 I 2 U i p n i0q i d R 1 dpi0 c p dTi0 . n i0q i d qi d
122
This comprehensive conservation equation for the toroidal angular momentum Lt mini0R2t of a tokamak plasma is the primary result of this paper. Here, the approximate equality indicates that we have neglected the electron , viscous, and Reynolds stresses, which are usually smaller by a factor of the order of me / mi 1 / 60. Similar to the density, VLt represents the plasma toroidal angular momentum between the and + d ux surfaces; it is also an adiabatic plasma property that vanishes in the absence of dissipative processes and momentum sources. The Maxwell stress contributions in Eq. 119 are induced by microturbulence. They can be made more explicit by using Amperes law for the uctuations and the vector identity C C = C2 / 2 C C: = e 1 B B e JB 0 = e 1 . B B 0
The factor accounts for the bunching together 1 / a of poloidal ux surfaces on the tokamaks outboard side but stretched separations 1 / a on the inboard side due to the Shafranov shift, etc. The resultant E or t causes the electron and ion nonambipolar radial particle uxes to become equal i.e., ambipolar and produce no net radial current,
na na na e anE = ZiianE Jan E = 0.
123
120
However, for radially local uctuations that do not connect to external magnetic elds, the average of this contribution over a narrow radial region of the plasma vanishes see Eq. 28 and discussion in Ref. 12. toroidal torque vanishes for ideal MHD JB The e perturbations throughout the hot core of tokamak plasmas.18 However, in the vicinity of a low order rational surface e.g., q = m / n, nonideal effects can allow an externally imposed NA magnetic eld component to induce a nite parallel cur rent Jmn and nonzero B mn within the thin nonideal bound18,19 ary layer around the low order rational surface. This pro NA cess produces the toroidal torque e JNA mn B mn in Eq. 119 in the vicinity of low order q = m / n rational surfaces in a toroidally rotating plasma.
Note that obtaining no net radial current in the plasma is the same as determining t or E from the evolution equation for Lt given in Eq. 119. Because Eq. 118 and hence Eq. 119 are determined predominantly by the sum of the nonambipolar ion particle uxes, the radial electric eld E obtained here is what is referred to as the ion root in the stellarator literature.46 Thus, from Eq. 75 the net ambipolar radial particle ux a a + pc and is the sum of the intrinsically ambipolar uxes na the nonambipolar uxes an evaluated at E,
net a na a net e = e + epc + eanE = i .
124
As indicated, since the nonambipolar ion particle uxes mostly sum to zero to yield J eZiina anE 0, it is easiest to determine the net ambipolar particle ux using the nonambipolar electron uxes. Since electron perpendicular-viscosity-driven classical and neoclassical particle uxes are me / mi 1 / 3672 1 smaller than the corresponding already small ion ones, the lowest order electron particle ux is likely to be determined by the paleoclassical particle ux pc plus the electron Reynolds and Maxwell stress particle uxes,
082504-15
Phys. Plasmas 16, 082504 2009
na na a dne0 n V net e0 pc e pc + eReyE + eMaxE = D d
1 1 1 . 125 B Ve e ne0V e e pV p
However, electron particle uxes due to parallel viscous force effects from NA magnetic elds via e and momentum source effects via eS may be important for particular tokamak plasma situations. Also, during poloidal magnetic ux transients, the FSA density equation 74 has the addi / n / . This n e0 / = tional contribution p e0 p p added ambipolar radial motion effect could lead, for example, to the density pump-out effect see, for example, Ref. 47 as ECH is applied to a tokamak plasma, locally 3/2 rapidly increases Te, reduces D nc 1 / T e , and causes 0 and hence n / t p ne / 0 there. p e0 p
V. DISCUSSION
The seven contributions on the right of the toroidal torque equation 119 have a wide variety of effects. In the following we describe briey for each contribution, in order of its appearance in Eq. 119, the parameter regime in which it is likely to be signicant, and if it were dominant what effect it would have on the plasma toroidal rotation frequency t. 1 NTV due to NA elds. When eld errors are very large or large NA control elds are deliberately applied, the neoclassical toroidal viscous torque in Eq. 100 acts as a drag throughout the plasma. It relaxes the plasma toroidal rotation toward an intrinsic or offset42,43 rotation frequency in the counter to the current direction given by Eq. 101: t c p + ct1 / qi p dTi0 / d 0. 2 Collision-induced perpendicular viscosities. In plasmas with approximately Ohmic-level heating and near the plasma edge where Te 1 keV, and when uctuationinduced transport is suppressed, paleoclassical perpendicular viscosity in Eq. 103 may be dominant and cause t 0 through a combination of radial diffusion of Lt with diffusivity D from Eq. 67 and an inward pinch Vpc from Eq. 79. 3 Fluctuation-induced ion Reynolds stress. Various properties of microturbulence can lead813 to radial diffusion, plus nondiffusive typically co-current intrinsic rotation or pinch velocity12,4853 and residual stress13,54 effects that do not depend on either t or its radial derivative. 4 Fluctuation-induced Maxwell stress. This tends to be small for low plasmas, but its electron component could be important in the radial particle ux. 5 Resonant NA magnetic elds. They produce a toroidal torque in the vicinity of low order resonant rational surfaces where q = m / n. Toroidal ow inhibits penetration of resonant eld errors into the plasma by producing a shielding effect on the rational surfaces. Above a critical eld error amplitudetermed the penetration
thresholdthe plasma rotation can no longer suppress the resonant torque, and plasma rotation at an m / n rational surface rapidly in a few milliseconds locks to the wall laboratory frame.18,21 Then, the diffusive radial momentum transport due to perpendicular collisioninduced viscosity and the ion Reynolds stress slows toroidal rotation throughout the plasma on a fraction of the global momentum connement time scale.55 The net result is then a toroidal rotation prole that is locked to the wall on low order rational surfaces with some residual toroidal rotation between rational surfaces if the plasma toroidal torque source e sSsm is large. 6 Poloidal magnetic ux transients. Poloidal ux tran2 sients that cause p 0 move Lt mini0R t radially / . The rotation frequency = at a speed d / dt = p p p t typically decreases with minor radius; thus, from Eq. 72 one usually has Lt / 0. If, in addition, 0 e.g., due to EC heating or turn-on of a non p inductive co-current source, this effect dynamically reduces the plasma toroidal angular momentum on the magnetic eld diffusion time scale: 1 / V VLt / t p p Lt / D Lt / 0. 7 Momentum sources. A net toroidal momentum source e s Ssm applies toroidal torque to the plasma. For example, unidirectional tangential energetic NBI used to heat tokamak plasmas also causes a toroidal torque on the plasma via the injected fast ions collisionally transferring26 their momentum to the background plasmain the direction of the NBI. In steady state this torque on the plasma is usually balanced by the uctuation-induced ion Reynolds stress discussed in 3 above. Similar considerations apply to the radial current induced by energetic electrons involved in lower hybrid CD LHCD experiments.56 In LHCD, the wave momentum input transferred to the energetic electrons provides a plasma momentum source via collisional transfer of the momentum in energetic electrons to the background plasma. This LHCD plasma momentum source is in the direction opposite to the wave momentum input because the negatively charged energetic electrons cause a current in the opposite direction from the wave momentum input. Thus, co-current LHCD produces a plasma torque in the counter-current direction. In the preceding discussion, particularly point 7, it was assumed that the only effects of the sources and sinks are through their particle and momentum inputs. However, for tangential NBI the mechanical momentum source from the collisional slowing down of the injected untrapped fast ions only comprises26 a fraction 1 of the total momentum input, because at birth the v of the NBI-injected fast ions is less than their toroidal velocity. The remaining fraction results from the change in the fast ion guiding center canonical toroidal angular momentum pg mvI / B qi p caused by the difference in their birth radius from the bounce average of their radial guiding center position: f 26 f this causes a = n p f . For a fast ion injection rate radial fast ion current J f n f q f f . At the plasma
082504-16
Phys. Plasmas 16, 082504 2009
edge this radial current can also be induced by direct fast ion losses to the surrounding vacuum vessel wall. Equation 118 did not include this fast-ion-induced current. However, it should be added there,57 in which case this equation becomes J + J f = 0. Multiplying Eq. 51 by nq, we see that J p = J p = e J B RJrB p. Thus, the toroidal torque the fast ion radial current induces can be accommodated in the formalism of this paper by including an extra fast ion momentum source,57 f q f f . e S fm J f p RB pn 126
This is positive for tangentially co-injected ions that on average drift inward of their birth radius and cause f 0. This extra toroidal angular momentum input due to a fastion-induced radial current causes a return current in the plasma:57 J = J f . For tangential NBI this momentum input is typically a small fraction of the direct momentum input and hence can usually be neglected; however, for near-perpendicular NBI, this fast-ion-induced radial current momentum source effect can be dominant. Similar considerations apply to the radial current induced by the energetic electrons involved in LHCD Ref. 56 due to the average radial guiding center motion of the extra species of energetic electrons. In addition, a probe inserted into a plasma58 that draws or emits a current Iprobe VJprobe produces an analogous toroidal momentum source e S pm = Iprobe / V p at the radial position of the probe.
The net radial current is determined by summing over species current contributions from all eight nonambipolar particle uxes. The toroidal rotation frequency t in Eq. 121 and hence radial electric eld in Eq. 122 are obtained from Lt. For this radial electric eld, the net electron and ion particle uxes become equal ambipolar Eq. 124 and are determined mainly by the electron particle uxes indicated in Eq. 125. This paper has also developed a general self-consistent framework for taking account of all the anomalous transport effects of microturbulence on tokamak plasmas. These effects are explicitly included in equations for the parallel Ohms law 36 and 40, poloidal ion ow 46, particle uxes 109 and 110, momentum ux 114, and plasma toroidal angular momentum 119.
ACKNOWLEDGMENTS
VI. SUMMARY
The authors are grateful to K. C. Shaing for many useful collaborations and discussions over the years and to their DIII-D colleagues and S. A. Sabbagh Columbia University/ NSTX whose experimental results stimulated this research in large part. They are particularly grateful to A. M. Garofalo, K. H. Burrell, J. S. deGrassie, M. J. Schaffer, and E. J. Strait for many stimulating discussions about toroidal ows and their recent experimental tests Refs. 42 and 43 of one of our key predictions. They also gratefully acknowledge useful discussions of a preliminary manuscript draft with N. M. Ferraro, S. C. Jardin, J. E. Rice, and R. E. Waltz and of ECH density pump-out effects with R. Prater and F. Volpe. Finally, they thank R. J. Groebner for careful reading of the draft manuscript. This research was supported by U.S. DOE under Grant Nos. DE-FG02-86ER53218 and DE-FG02-92ER54139.
Three independent components of the plasma force balance have been considered on sequential time scales: 1 plasma MHD force balance 16 on the fast Alfvn time scale, which leads to a fundamental relation 17 between toroidal and poloidal ows, and E B and ion diamagnetic ows; 2 plasma parallel force balance 43, which on the intermediate ion collision time scale leads to specication 46 of the ion poloidal ow in terms mainly of the radial ion temperature gradient; and nally, 3 toroidal force balance 56 on the slow transport time scale, which leads to radial particle uxes. The transport-time-scale density equation on a poloidal ux surface is specied in Eq. 74. The net particle ux is composed of seven ambipolar collisional uxes 86 due to classical, PS, banana-plateau neoclassical, paleoclassical, CD, dynamo current, and EA B / B2 pinch transport processes plus eight nonambipolar uxes 115 due to parallel NA and perpendicular viscosities, polarization ows, uctuation-induced Reynolds and Maxwell stresses, externally imposed resonant NA magnetic elds, poloidal ux transients, and momentum source effects. Finally, a new comprehensive evolution equation for the plasma toroidal angular momentum density Lt mini0R2t in Eq. 119 has been obtained by requiring the net radial current to vanish on the transport time scale.
APPENDIX A: FLUID AND GUIDING CENTER FLOWS, FLUXES
Since the rst order perturbed ows have a nonzero radial component, they lead to the anomalous radial particle . The species uid ow velocity V V ux n 1V E + V is related to the guiding center ow velocity usually calculated in drift kinetics and gyrokinetics via3 nqVE + V = nqVg + M. Here, Vg d3vvd f M / n is the ow velocity of particle guiding centers having drift velocities vd, and M pB / B2 is the magnetism produced by the magnetic moments of the charged particles. Since to lowest order vd E in 20 p / B2 0 1 and the gyroradius parameter , 2 2 B / B , and M B p / B , we obtain
0 0 0 0
= V g
, d3v vd f M /n0 V E
A1
= 1/n q M V . V M 0
A2
+V + O , 2. Moreover in general, since =V Thus, V g
082504-17

13
Phys. Plasmas 16, 082504 2009 P. H. Diamond, C. J. McDevitt, . D. Grcan, T. S. Hahm, W. X. Wang, E. S. Yoon, I. Holod, Z. Lin, V. Naulin, and R. Singh, Nucl. Fusion 49, 045002 2009. 14 T. H. Stix, The Theory of Plasma Waves McGraw-Hill, New York, 1962, p. 206. 15 K. C. Shaing, S. P. Hirshman, and J. D. Callen, Phys. Fluids 29, 521 1986. 16 K. C. Shaing, Phys. Plasmas 10, 1443 2003. 17 K. C. Shaing and J. D. Callen, Nucl. Fusion 22, 1061 1982. 18 R. Fitzpatrick, Nucl. Fusion 33, 1049 1993. 19 A. J. Cole and R. Fitzpatrick, Phys. Plasmas 13, 032503 2006. 20 A. J. Cole, C. C. Hegna, and J. D. Callen, Phys. Rev. Lett. 99, 065001 2007. 21 A. J. Cole, C. C. Hegna, and J. D. Callen, Phys. Plasmas 15, 056102 2008. 22 R. D. Hazeltine, F. L. Hinton, and M. N. Rosenbluth, Phys. Fluids 16, 1645 1973. 23 S. P. Hirshman and S. C. Jardin, Phys. Fluids 22, 731 1979. 24 S. C. Jardin, J. Comput. Phys. 43, 31 1981. 25 J. Blum and J. Le Foll, Comput. Phys. Rep. 1, 465 1984. 26 J. D. Callen, R. J. Colchin, R. H. Fowler, D. G. McAlees, and J. A. Rome, Plasma Physics and Controlled Nuclear Fusion Research 1974 International Atomic Energy Agency, Vienna, 1975, Vol. 1, p. 645. 27 K. C. Shaing and J. D. Callen, Phys. Fluids 26, 3315 1983. 28 K. C. Shaing, Phys. Fluids 29, 2231 1986. 29 J. D. Callen, Phys. Plasmas 12, 092512 2005. 30 J. J. Ramos, Phys. Plasmas 12, 112301 2005. 31 F. L. Hinton, R. E. Waltz, and J. Candy, Phys. Plasmas 11, 2433 2004. 32 K. C. Shaing, Phys. Fluids 31, 8 1988. 33 A. L. Garcia-Perciante, J. D. Callen, K. C. Shaing, and C. C. Hegna, Phys. Plasmas 12, 052516 2005. 34 Y. B. Kim, P. H. Diamond, and R. J. Groebner, Phys. Fluids B 3, 2050 1991. 35 W. A. Houlberg, K. C. Shaing, S. P. Hirshman, and M. C. Zarnstorff, Phys. Plasmas 4, 3230 1997. 36 J. D. Callen, Phys. Plasmas 14, 040701 2007. 37 J. D. Callen, Phys. Plasmas 14, 104702 2007. 38 J. W. Connor, R. J. Hastie, and J. B. Taylor, Phys. Plasmas 15, 014701 2008. 39 J. D. Callen, Phys. Plasmas 15, 014702 2008. 40 S. P. Hirshman, Nucl. Fusion 18, 917 1978. 41 A. J. Cole, C. C. Hegna, and J. D. Callen, Low Collisionality Neoclassical Toroidal Viscosity in Tokamaks and Quasi-Symmetric Stellarators Using an Integral-Truncation Technique, Report No. UW-CPTC 08-8, June 2009 available from http://www.cptc.wisc.edu. 42 A. M. Garofalo, K. H. Burrell, J. C. DeBoo, G. L. Jackson, M. Lanctot, H. Reimerdes, M. J. Schaffer, W. M. Solomon, and E. J. Strait, Phys. Rev. Lett. 101, 195005 2008. 43 A. M. Garofalo, W. M. Solomon, M. Lanctot, K. H. Burrell, J. C. DeBoo, J. S. deGrassie, G. L. Jackson, J.-K. Park, H. Reimerdes, M. J. Schaffer, and E. J. Strait, Phys. Plasmas 16, 056119 2009. 44 J. J. Ramos, private communication 2009. 45 J. D. Callen, W. X. Qu, K. D. Siebert, B. A. Carreras, K. C. Shaing, and D. A. Spong, Plasma Physics and Controlled Nuclear Fusion Research 1986 International Atomic Energy Agency, Vienna, 1987, Vol. 2, p. 157. 46 H. E. Mynick and W. N. G. Hitchon, Nucl. Fusion 23, 1053 1983. 47 C. Angioni, A.G. Peeters, X. Garbet, A. Manini, F. Ryter, and ASDEX Upgrade Team, Nucl. Fusion 44, 827 2004. 48 A. G. Peeters, C. Angioni, and D. Strintzi, Phys. Rev. Lett. 98, 265003 2007. 49 T. S. Hahm, P. H. Diamond, O. D. Gurcan, and G. Rewoldt, Phys. Plasmas 14, 072302 2007. 50 . D. Grcan, P. H. Diamond, and T. S. Hahm, Phys. Rev. Lett. 100, 135001 2008. 51 T. S. Hahm, P. H. Diamond, O. D. Gurcan, and G. Rewoldt, Phys. Plasmas 15, 055902 2008. 52 A. G. Peeters, C. Angioni, and D. Stinzi, Phys. Plasmas 16, 034703 2009. 53 T. S. Hahm, P. H. Diamond, O. D. Gurcan, and G. Rewoldt, Phys. Plasmas 16, 034704 2009. 54 H. L. Berk and K. Molvig, Phys. Fluids 26, 1385 1983. 55 M. Yokoyama, J. D. Callen, and C. C. Hegna, Nucl. Fusion 36, 1307 1996.
= M = 0, we have 1 / V / V n 1V nV 1 / V / V , where nV is the = g g g uctuation-induced particle ux evaluated in gyrokinetic codes see, for example, Ref. 12. Also, since E B ows are the same for electrons and ions, the net particle uxes g determined from guiding center ows are ambipolar to lowest order.
APPENDIX B: GYROVISCOSITY AND ITS EFFECTS
The gyroviscous stress is a symmetric tensor that is caused by diamagnetic-type B V effects. It is described in detail in Ref. 30. The gyroviscous stress tensor can be written to lowest order in and neglecting heat ux effects for illustrative purposes as3
T = p/4cW + W .
B1
W I + 3b b in which b B / B, I is the idenHere, W b tity tensor, W = V + VT 2 / 3I V, and the superscript T indicates the transpose of that tensor. For this form of the gyroviscous stress where the heat ux effects are neglected, to lowest order in and the gyroviscous force can be written as30 mnV V = m/q M V . B2
This gyroviscous force cancels the V V part of V V in Eq. 33 and causes it to become approximately VE V, which has been speculated59 to be in general Vg V. Since the -average uctuation-induced gyroviscous stress is a symmetric tensor, using Eq. 54 we nd in general that e = e = 1 e . B3 V
To lowest order in , neglecting heat ux effects and gradients of equilibrium quantities, the gyroviscous stress induced by uctuations is +W T . 1/4c0 pW
1
B4
R. Aymar, V. A. Chuyanov, M. Huguet, Y. Shimomura, ITER Joint Central Team, and ITER Home Teams, Nucl. Fusion 41, 1301 2001. 2 J. D. Callen, A. J. Cole, and C. C. Hegna, Nucl. Fusion 49, 085021 2009. 3 S. I. Braginskii, in Reviews of Plasma Physics, edited by M. A. Leontovich Consultants Bureau, New York, 1965, Vol. 1, p. 205. 4 M. N. Rosenbluth, P. H. Rutherford, J. B. Taylor, E. A. Frieman, and L. M. Kovrizhnikh, Plasma Physics and Controlled Nuclear Fusion Research 1971 International Atomic Energy Agency, Vienna, 1972, Vol. 1, p. 495. 5 F. L. Hinton and R. D. Hazeltine, Rev. Mod. Phys. 48, 239 1976. 6 S. P. Hirshman and D. J. Sigmar, Nucl. Fusion 21, 1079 1981. 7 P. J. Catto and A. I. Simakov, Phys. Plasmas 12, 012501 2005. 8 N. Mattor and P. H. Diamond, Phys. Fluids 31, 1180 1988. 9 A. G. Peeters, Phys. Plasmas 5, 763 1998. 10 B. Scott, Phys. Plasmas 10, 963 2003. 11 . D. Grcan, P. H. Diamond, T. S. Hahm, and R. Singh, Phys. Plasmas 14, 042306 2007. 12 R. E. Waltz, G. M. Staebler, J. Candy, and F. L. Hinton, Phys. Plasmas 14, 122507 2007; 16, 079902E 2009.
082504-18
56

58
Phys. Plasmas 16, 082504 2009 R. J. Taylor, M. L. Brown, B. D. Fried, H. Grote, J. R. Liberati, G. J. Morales, P. Pribyl, D. Darrow, and M. Ono, Phys. Rev. Lett. 63, 2365 1989. 59 N. M. Ferraro, Non-Ideal Effects on the Stability and Transport of Magnetized Plasmas, Ph.D. thesis, Princeton University, November 2008, p. 18.
J. E. Rice, A. C. Ince-Cushman, P. T. Bonoli, M. J. Greenwald, J. W. Hughes, R. R. Parker, M. L. Reinke, G. M. Wallace, C. L. Fiore, R. S. Granetz, A. E. Hubbard, J. H. Irby, E. S. Marmar, S. Shiraiwa, S. M. Wolfe, S. J. Wukitch, M. Bitter, K. Hill, and J. R. Wilson, Nucl. Fusion 49, 025004 2009. 57 F. L. Hinton and M. N. Rosenbluth, Phys. Lett. A 259, 267 1999.

PhysPlasmas 16 082504

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

PhysPlasmas 16 082504

Загружено:

Авторское право:

Доступные форматы

PHYSICS OF PLASMAS 16, 082504 2009

Toroidal ow and radial particle ux in tokamak plasmas

Electronic mail: callen@engr.wisc.edu. Website: http://www.cae.wisc.edu/ ~callen.

Callen, Cole, and Hegna