Documente Academic
Documente Profesional
Documente Cultură
• 核融合基礎学(プラズマ・核融合基礎学)
• 李継全(准教授)/岸本泰明(教授)/今寺賢志(D1)
• 2007.4.9 — 2007.7.13
2. Kinetic Theory.
Boltzmann Equation → statistical description → transport coefficients
All descriptions should be consistent. Sometimes they are only different ways to
approximately look at the same thing.
Further, some approximate models have been developed such as: fluid-kinetic hybrid
model; gyrofluid model.
2
Basic equations of plasma physics
Electric and magnetic fields (E & B) are generally determined by Maxwell’s
equations, with corresponding boundary conditions and the sources (charges and
currents). Gauss’s law
E 4
No magnetic poles B 0
1 B
Faraday law E
c t
4 1 E
Ampere’s law B J
c c t
Sometimes E & B are expressed in terms of an electric potential φ and vector
potential A:
E ( r , t ) ( r , t ) A( r , t ) B( r , t ) A( r , t )
ct
In this case, electromagnetic field equations are written in the form
1 2
1 A
2 4
2
4 ( r , t ) A
2
J (r , t )
c t 2 c t 2 c
1
With Lorentz gauge A 0
c t 3
Basic equations of plasma physics (cont.)
Equation of motion
The motion of charged particles is determined by the electromagnetic fields through
the equations of motion – Lorentz equation
d i Fi qi 1
( E i B)
dt mi mi c
dxi
i
dt
The forces include two contributions from external electromagnetic field and also
internal field, which is produced by other particles. The latter should be evaluated
self-consistently.
Hence, the electromagnetic forces in a plasma depends on the current and charge
densities which are determined by the collective particle interaction.
4
Kinetic equations
To describe a plasma with a large number of particles, one can solve the coupled
system of Maxwell’s equations and the equations of motion for each particle. This is
a terrible job!
However, there are more efficient methods to solve the plasma dynamics using
statistical approximation – kinetic equation.
Consider the single particle distribution function f(t, r, v) which gives the density of
particles in the six-dimensional space (r,v), The single particle distribution function
satisfies the Boltzmann equation
f q 1 f
f ( E B ) f
t m c t c
The charge and current densities can be evaluated as
(r , t ) qs f s (r , , t )d 3
s
J (r , t )
s
qs f s ( r , , t )d 3
To describe a plasma, it needs only to solve Maxwell equations and kinetic equation!
5
Single Particle Approach
6
Single Particle Approach
– orbits & drifts of particle in electromagnetic fields.
7
Gyro motion & Larmor radius
L
Guiding center
From equation of motion, we can easily know that particle moves along the magnetic
field with υ//0 and gyrates around the filed. Here the second magnetic field produced
by moving charged particle is ignored!
Orbit of charged particle is
x x cos( g t )
Initial velocity ofcharged
particle in 0
magnetic field 0 ( 0 , // 0 ) g
y y sin( g t )
Charged particle is only experienced 0
g
Lorentz force q B z z t
m 0 //
qB v mcv
gyrogrequency g gyroradius L
mc g qB
8
Gyration is the most basic motion of charged particle in a magnetized plasma!
Electric field drifts (E×B drift)
Equation of motion is
d //
m dt qE //
d qE q B
dt m mc
cE B
The “E × B drift” of the gyrocenter is d
B2
Homework: Problem 3
Derive the orbit of positive charged particle q with initial velocity [ 0 ( 0 , // 0 )] in a
constant uniform electric field (0,E┴,E//) and magnetic field (0,0,B), express the 9
velocity of particle gyrocenter.
Drifts due to general force F
Lorentz equation is
d F q
B
dt m mc
This general force can be gravity, force due to non-uniform magnetic field
(gradient or curvature)
10
Magnetic field gradient drifts
Rc 2 1
Gradient and curvature drifts
Fc m 2 m // 2 ( B )B
2
// are related through Maxwell’s
Rc B
equations, which depends on
In the frame of the guiding center a force the current density j. A
appears because the plasma is rotating about particular case of interest is j =
the center of curvature. 0: vacuum fields.
B
Curvature drift velocity (B )r
Rc
mc 2 1 2 cRc B
c 4
B [( B ) B]
//
B c m // m 2 2
2
qB 2 qB
12
Rc
Drifts in varying electric field
– polarization drift
If electric field E is time-varying, the particle experiences a acceleration,
d d cE B
d
dt dt B
2
In the frame of the guiding centre which is accelerating, a force is felt except for
the force due to uniform electric filed.
d EB
Fp mc
2
dt B
An additional drift is produced as
mc d E B mc dE
Polarization drift p 2 2 B 2
qB dt B qB dt
13
Remarks for single particle drifts
All these drift velocities and the particle orbit above can be derived directly by
solving the motion of particle with an initial velocity (υ┴0, υ//0) in assumed time-
varying non-uniform electric E(t, x) and magnetic fields B(t, x), i.e.,
d i qi 1 Gyromotion; electric field drift
[ E ( t , x ) i B( t , x )]
dt mi c with Larmor radius modification;
magnetic field gradient and
dxi
i with 0 ( 0 , // 0 ) curvature drifts; polarization
dt drift; magnetic mirror.
These drifts have been determined by assumed electric and magnetic fields. They
describe test particle motion. However, it should be noticed that the currents due to
the drifts alter the fields. If these changes are small compared to the background
field it is justified to apply the drift model. The derived particle drifts do not contain
any collective behavior. For this reason it is a nontrivial aspect to compare particle
and fluid plasma drifts.
Hence, single particle approach has ignored the interaction among charged particles,
14
it is only suitable for enough low density plasma.
Plasma Kinetic Description
15
Plasma Kinetic Theory
– why need kinetic description
Many particle: For a plasma, the plasma parameter is g=1/(nλ3d)<<1. Thus a
plasma consists of a very large number of particles. It is too tough work to calculate
the orbits of all particles even if for assumed electric and magnetic fields.
Long-range force: The charged particles of a plasma are both responding to the
electromagnetic fields and acting as their sources. This means charged particle
moving under the influence of both the external fields and the fields generated by
the particles themselves. Namely, the plasma behaves collectively. It is almost
impossible to calculate the motion of all particles in a plasma self-consistently.
Fields as an average: Actually, the orbits of all particles are not so important in a
plasma, the spatial and temporal development of statistical measurable quantities
as a fluid, i.e., particle density, particle flux, temperature or pressure, heat flux, and
so on, are more interesting. Because the collective behavior of the charged particles
is a fundamental property of plasmas, we do not always need to know anything
about the individual particles but, instead, we are interested in the average
properties of the gas or fluid.
Liouville equation approach: This approach starts with distribution functions and
avoids δ-functions and ensemble averaging. (we will not talk about this approach in
this lecture) 17
Klimontovich equation approach
Consider a single particle with orbit (xi(t);vi(t) ) in 6-dimensional phase space. The
“density” of this particle is, i.e., the distribution function of single particle,
N (t , x, ) [ x xi (t )] [ i ( t )]
For particles in a plasma, the microscopic distribution function is the summation
N
Klimontovich N ( t , x , ) [ x xi ( t )] [ i ( t )]
m
i 1
Here xi(t) and vi(t) are the spatial and velocity trajectories as the particles move.
∆v
All particles (i=1, N) have time-dependent
(x(t);v(t)) position xi(t) and velocity vi(t). The particle
path at subsequent times is a curve in phase
∆x space. 18
x
Basic equations for particle simulation
Here xi(t) and vi(t) are determined by the equations of motion and Maxwell
equations,
dxi E 4
m
i
dt
B 0
d i Fi qi 1
( E i B) 1 B
dt mi mi c E
c t
The microscopic sources are determined by
4 m 1 E
B J
m ( x, t ) qs N sm ( x, , t )d 3 qs ( x xi ) c c t
s s
m m
J ( x , t ) qs N s ( x, , t )d qs i ( x xi )
3
s s
N
with N ( t , x , ) [ x xi ( t )] [ i ( t )]
m
i 1
To do so, it may be convenient to have a single evolution equation for the entire
microscopic distribution. Such an equation can be obtained by calculating the total
time derivative of microscopic distribution:
dN m dx d N
[ x xi (t )] [ i (t )]
dt t dt x dt i 1
dg ( f ( x )) dg df
By using relations: g( x ) ( x y ) g( y ) ( x y )
dx df dx
dN m N
dxi d i
[ x xi (t )] [ i (t )] 0
dt i 1 t dt x dt
Inserting equation of motion, we have
Klimontovich equation N m
dx N m q m 1 m N m
E ( t , x ) B ( t , x ) 0
t dt x m c 20
Properties of the Klimontovich equation
Klimontovich equation together with the Maxwell’s equation and the definitions for
charge and current densities also provide an exact and complete description of the
plasma dynamics!
Klimontovich equation actually incorporates all particle equations of motion into one
equation since its “characteristic curves” in (t,x,v) phase space are the equations of
motion.
21
From Klimontovich equation to
Plasma kinetic equation
Define the fluctuation (deviation from the averaged level) of complete microscopic
distribution function Nm from the averaged one fs, i.e.,
N m ( t , x , ) f s ( t , x , ) N m ( t , x , )
We have N m ( t , x , ) 0 f
Distribution function
s
The average distribution function fs represents the
smoothed properties of the plasma species for ∆x
>λD; the microscopic distribution δNm represents the
“discrete particle” effects of individual charged Nm
particles for n-1/3<< ∆x<λD .
Similar separation for the fields
E m (t , x, ) E (t , x, ) E (t , x, )
∆x ∆v
B m (t , x, ) B(t , x, ) B(t , x, ) 23
E ( t , x , ) 0 B ( t , x , ) 0
Fundamental plasma kinetic equation
Substituting these forms into the Klimontovich equation and averaging it using the
procedure above, we obtain our fundamental plasma kinetic equation:
f s f s q m 1 m f s q m 1 m N m
E B E c B
t x m c m
The left side describes collective effects in the plasma, i.e., the evolution of the
smoothed, average distribution function in response to the smoothed, average
electric and magnetic fields.
The right side represents the small two-particle correlations between discrete
charged particles within about a Debye length of each other. In fact, the term on the
right represents the collisional effects, i.e., Coulomb collision effects on the average
distribution function fs.
Similarly averaging the microscopic Maxwell equations and charge and current
density sources, we obtain corresponding average equations that have no extra
correlation terms.
24
Fokker-Planck equation or Boltzmann equation
Rewriting the right side of the fundamental kinetic equation as (∂fs/∂t)c, a collision
operator on the average distribution function fs. We can have Fokker-Planck (FP) or
Boltzmann equation
f s f s q 1 f s f s
( E B)
t x m c t c
The form of the collision term on the right side depends on the nature of collisions:
– Boltzmann equation: for hard collisions and localized in space and time.
– FP equation: for collision through cumulative contribution of many small angle
Coulomb scatterings.
With corresponding averaged Maxwell equations and charge and current densities,
( x, t ) qs f s ( x, , t )d 3
1 B
E 4 ; E
c t s
J ( x, t ) qs f s ( x, , t )d 3
4 1 E
B 0; B J
c c t s
Homework: problem 4
26
Derive this conservation form of Boltzmann equation.
Reduced forms of Boltzmann equation –
Vlasov equation
For the fluctuation with short time scale in high temperature laboratory plasmas or
space plasmas, the collision is typically small, i.e., ω>>ν, we have so-called Vlasov
equation
f s f s q 1 f s
( E B) 0
t x m c
1. Due to no collision, the filamentary structures in Vlasov plasma can become more
contorted as time evolution. Hence, Vlasov code can follow the distribution
function in physics for long time only before the numerical problem occurs.
2. Due to no collision, Vlasov equation has no discrete particle correlation (Coulomb
collision) effects in it, it is completely reversible (in time) and its solutions follow
the collisionless single particle orbits in the six-dimensional phase space.
3. A Vlasov plasma is stable since the stable distribution with d fs/dε<0 minimizes
the kinetic energy.
4. Any free energy related to dfs/dε<0 may drive collective instability, profile non-
homogeneity; velocity anisotropies; flows such as beams and currents.
27
Reduced forms of Boltzmann equation (cont.) –
gyro-averaged kinetic equations
In a magnetized plasma, many plasma phenomena involve processes which are slow
compared to the gyrofrequency and which vary slowly in space compared to the
Larmor radius of individual ions or electrons. That is, the fluctuations in plasma
are characterized by longer spatial scale compared to the gyroradii (L>>ρg) and
by slow processes compared to the gyrofrequency (ω<<ωc).
Drift-kinetic equation:
This is a form of Fokker-Planck (Boltzmann or Vlasov) equation, which describes
the evolution of distribution function fs under conditions where it occurs slowly in
time compared to the gyro-period and the gradually in space compared to the gyro-
radius of particle orbits. Actually, this is an equation of fs at the guiding center
position xg. In principle, we should transform the results back from guiding center
to real space coordinates after solving it. However, this procedure is usually
neglected since the gyroradius is small and the effect is ignorable.
The conditions for applying this model are: ω<<ωc ; k┴ρg<<1
Gyro-kinetic equation:
This equation is similar to drift-kinetic equation, but it can describe the significant
change of electromagnetic field across a Larmor radius by averaging their effect
over the Larmor orbit.
The conditions for applying this model are: ω<<ωc ; k┴ρg~1
2 2B
Re-write kinetic equation by using new variables
f s dx g f s d f s d f s d f s f s
t dt x g dt dt dt t c
~ ~ ~
Define f s f s f s f s
f s with fs 0
~
where f s is small quantity with Larmor radius order
g q
Assuming ~ k g ; ~ c ; ~ c ; s E // ~ c
L t m
qs EB
E ~ ~ c ; ~ c
m t
30
Derivation of drift kinetic equation (cont.)
The change of total kinetic energy
can be subject to the gain of energy of the guiding
center in the electric field qs E dx g and the change of the perpendicular energy due
to a change of the magnetic field d dB / B
d dx m 2 B
qs E
g
dt dt 2 B t
d
0 due to magnetic moment conservation
dt
dx g B B cE B mc 2 mc //2
// D // 2
3
B B 4
[ B ( B ) B ]
dt B B B 2qs B qs B
guiding center drift velocity
Substitute all relations into rewritten kinetic equation in guiding center coordinate
and performing gyro-averaging, we can get
f s ( t , x g , , ) f s ( t , x g , , ) d f s ( t , x g , , )
sf ( t , x g , , )
( // D )
t x g dt
t c
This equation is used in linear and nonlinear studies of low frequency and long
wavelength instabilities, in neoclassical transport theory where the contribution
31
from Larmor gyration is not so important.
Gyrokinetic equation J Wesson, TOKAMAK (second edition), 1997
Drift-kinetic equation with the lowest order is sufficient for most applications.
However, like the guiding center orbits it is based on, it is incorrect at second order
in the small gyroradius expansion. More precise and complete equation is
gyrokinetic equation.
In long wavelength limit and L→0, the distribution function is reduced to the result
32
from drift kinetic equation.
References for the derivation of nonlinear gyrokinetic equation
(classical and modern gyrokinetic theories, collected by T S Hahm)
1. Hazeltine and Meiss, Plasma confinement (book)
2. Frieman and Chen, Phys. Fluids 25, 502 (1982)
3. Lee, Phys. Fluids 26, 556 (1983)
4. Dubin, Krommes, Oberman, and Lee, Phys. Fluids 26, 3524 (1983)
5. Hagan and Frieman, Phys. Fluids 28, 2641 (1985)
6. Hahm, Lee, and Brizard, Phys. Fluids 31, 1940 (1988)
7. Hahm, Phys. Fluids 31, 2670 (1988)
8. Brizard, J. Plasma Phys. 41, 541 (1989)
9. Brizard, Phys. Plasmas 2, 459 (1995)
10. Hahm, Phys. Plasmas 3, 4658 (1996)
11. Brizard, Phys. Plasmas 7, 4816 (2000)
12. Sugama, Phys. Plasmas 7, 466 (2000)
13. Brizard, Phys. Plasmas 7, 3238 (2000)
14. Wang, Phys. Rev. E. 64, 056404 (2001)
15. Qin and Tang, Phys. Plasmas 11, 1052 (2004)
16. Brizard and Hahm, Foundations of nonlinear gyrokinetic theory, Rev.
33
Mod. Phys. 79, 1-468(2007)
Fluid description of plasma
34
Fluid description of plasma
‘Fluid Description’ refers to simplified treatment of plasma which does not need the
details of velocity dependence.
Considering the integration of distribution function over whole velocity space is the
density, we integrate Boltzmann equation over velocity space (0 th moment)
f s f s q 1 f s
3 f s
t
3
d ( E B ) d
x m c t c
3 f s 3 3
ns
x
3
d d f s d f s ( ) f s d f s u
d F
f s
3
d 3
F f s
F f s
d 2
F f i
i s i
Fi
fs 0
i x , y ,z
i x , y ,z i
Fi
Here, we have used f s 0 for distributi on function; 0 for Lorentz force.
i
i 38
0th moment equation— continuity equation (cont.)
The right side becomes a source term Q n of particle number density due to collision,
such as the production or annihilation of mass through ionization or recombination.
ns
uns Qn
t
For incompressible fluid, u 0
39
1st moment equation— equation of motion
f s f s q 1 f s f s
s
3 3
d m s ( E B ) d m
t x m c t c
d 3m s f s d 3m s f s
,
d 3m s [( u) u][( u) u] f s
d 3m s ( u)( u) f s m s ( ns uu)
P m s ( ns uu )
qs 1 3 qs 1
d m s ( E B ) f s d m s ( E B ) f s
3
m s c ms c
qs 1 1
d j d k m s i
( E B ) s s u B ) ns
f q ( E
sm c c
j ,k
i x , y , z
1
q s ( E u B ) ns
c
f s
Rs
3
d m
t c
s
If the collisions are frequent enough, the pressure tensor becomes diagonal, or 41
even
isotropic, so P p
2nd moment equation— energy equation
m s 2 f s f s q 1 f s
3 m s f s
2
Letting
A d 3Af s
1 2 1
Qs ns ms K s us us Ps qs qs ns ms ( u) ( u)( u)
2 2
1 m 3
K s m 2 s w 2 u2 ps ns ms us
2
2 2 2
m s 2 f s m s 3
The first term is
d 2 t 2 t d ( u u) ( u u) f s
3
m
s d 3 ( u) ( u) u 2 f s
2 t
m
s d 3 ( u) ( u) u 2 f s
2 t
3 42
p s ns m s u 2
2
2nd moment equation— energy equation (cont.)
The second term is
3 m s f s m s 3
ms 3
2
2 2
d
2
x
2
d
x
f s
2
d
x
( u u) f s
ms 3
2
d
x
2
( w
u )
f s w
u
ms 3
2
d ( uf s ( w u) ( w u)wf s
x
ms 3 2
x 2
d uf s u P q s
2
3
p s n s m s u 2 u u P q s
x 2
1 ps 5
( p u) ( P ) u q Q with
1 t x s
x cs
3
44
Chain of moment equations
Similar way to derive high order equations
This procedure shows that low order moment equation includes higher moment,
which is an infinite chain of hierarchy!
This equation chain must be truncated at somewhere and by some way. It is often
made in the second order in many practical cases, either by neglecting the heat flux,
or by using an equation of state instead of the energy equation. Here physical
insight plays a crucial role. The treatment seems become a kind of art!
45
Closure moments
The general procedure to close a hierarchy of fluid moment equations is to derive the
needed closure moments, which are sometimes called constitutive relations, from
integrals of the kinetic distribution function for higher order moments. The
distribution function must be solved from a kinetic equation that takes account of
the evolution of the lower order fluid moments. The resultant kinetic equation and
procedure for determining the distribution function and closure moments are known
as the Chapman-Enskog approach. For situations where collisional effects are
dominant, the resultant kinetic equation can be solved asymptotically via an
ordering scheme and the closure moments. This approach has been developed in
detail for a collisional, magnetized plasma by Braginskii.
In the limits of high density and lower temperature, the collision is high, the fluid
theory is valid. But, a plasma is often described as a fluid even when it is far from
being collision dominated !!! This condition means that the effects of collisions is
negligible compared with the coherence produced by the self-consistent fields. 47
Why kinetic? Why fluid?
Plasma fluid theory is relatively simple and fluid quantities are measurable
experimentally. Plasma fluid theory can describe most of basic plasma phenomena.
For example, drift waves; cold plasma waves; MHD fluctuations; …...
The advantage of fluid theory lies in the fact that the dynamics of neutral fluid has
been extensively studied and many aspects of their behaviors are well understood.
Although the motion of plasma fluid is much more complex than that in the neutral
fluid, it is often useful to be able to draw analogies with the behavior of a plasma.
From the viewpoint of calculation (simulation), fluid codes require relatively less
CPU time compared to kinetic simulation (PIC or Vlasov codes).
For the first principal simulation, kinetic (or reduced kinetic) theory should be
48
employed.
Fluid/kinetic hybrid model – a mixed description
D Winske, Space Science Review 42, 53-65 (1985);
Computer space plasma physics (book) (1993)
Plasma phenomena are characterized by a multiple space and time scales, primarily
due to the different responses of electrons and ions to electric and magnetic fields.
Generally speaking, the fast varying and small scale physics phenomena require
kinetic descriptions, slow varying and large scale processes can be described by
more fluid models.
Some particularly interested processes occur on some of these scales but other
processes occur usually. This can be described by a mixed kinetic/fluid model.
Hybrid model describes this plasma system by using kinetic model for one species
(or part of one species) and by using fluid model for the rest.
The hybrid codes are defined as those numerical algorithms in which PIC particle or
Vlasov codes are applied for the species treated by kinetic description and fluid code
is for the species treated as a fluid.
49
Examples of fluid/kinetic hybrid model
1. The interaction of a small, cold electron beam (kinetic) with a hot background
electron population (fluid) because the unstable waves generated by the presence
of the beam strongly affect it. (O'Neil et al., 1971)
4. ……
50
Example for equations of hybrid model
As an example in the case with kinetic ions and fluid electrons, we have equations
One obtains a hierarchy of evolution equations for gyrocenter-fluid moments, i.e., for
density, parallel velocity, pressure, etc. To obtain a closed set of these gyrofluid
equations, one needs to invoke a closure approximation i.e., expressions for higher-
order fluid moments in terms of lower-order fluid moments.
To date, gyrofluid model still loses to involve properly some important physics
processes. For example, it cannot describe the zonal-flow damping accurately and
52
may overestimate the turbulence level in fusion plasmas.
Example for deriving a set of gyrofluid equations
Hammett & Perkins, PRL 1990
The calculation to get a satisfactory gyrofluid model (finite Larmor radius effects
and Landau damping) is much more complicated, sometimes up to 6- or 8-moment 54
equations.
Simplified two-fluid equations
We have derived fluid moment equations for each species. In the case without
ionization and energy exchange collisions, it can be written down a set of simplified
two-fluid equations for ions and electrons.
ni
ni ui Qni
t
1
(mi ni ui ) Pi (mi ni ui ui ) qi ni ( E ui B) Ri
t c
1 pi
( p u ) ( P ) u q Q
1 t i i
x i i x i ci
ne
ne ue Qne
t
1
(me ne ue ) Pe (me ne ue ue ) qe ne ( E ue B) Re
t c
1 pe
( p u ) ( P ) u q Q
1 t e e
x e e x e ce
With friction term Re ei me ne (ui ue ) Re Ri 0 Qci Qce (ui ue ) Re 5 3
In some cases, it is also assumed that the pressure is scalar. The heat conduction
term is neglected.
The separate two (ions and electrons) fluids interact through collisions and 55
through electromagnetic interaction.
Two-fluid equilibrium –
diamagnetic drift and diamagnetic current
At equilibrium, we have t 0 and collisionless limit ei 0
In slab geometry with straight B-field B Bẑ , assuming equilibrium fields and
profile vary only in x, the momentum equation is.
ps ms ns (us us ) qs ns ( E us B) 0
From the equation in y direction, we have usx 0 is a solution, we can have the
equilibrium equation
d qs ns 1 dp
ms ns usx usy Busx q n
s s E x usy
B 0
dx c c dx
d 1 dp E 1 dp
ms ns usx usx qs ns ( E x usy B) usy x
dx c dx B q n B dx s s
cE B cB p
Fluid drifts: uD 2
Second term is diamagnetic drift
B qs ns B 2
Diamagnetic current:
cE B cB ( pi pe ) cB ( pi pe )
j qi ni ui qe ne ue ( q n
i i q n
e e )
B2 B2 B2 56
Quasi-neutrality diamagnetic current
Single fluid model
Two fluid equations contain still considerable complexity which is not needed for
many plasma systems. It is desirable to formulate a more appropriate set of
equations which include most of the macroscopic properties of a plasma. This set of
equations are so-called magnetohydrodynamics (MHD) equations. This model
consists of a one-fluid model for the plasma and the Maxwell equations for the
electromagnetic fields.
In a charge neutral plasma, qi=-qe=e, define the following quantities for single fluid
model
Mass density: m ni mi ne me
Mass flux velocity: u ( ni m i ui ne me ue ) / m
Total current density: j qi ni ui qe ne ue qe ne ( ue ui )
Total pressure: p pi pe
t
MHD plasmas are quasineutral and have no net charge density, we can not calculate
the electric field from the Gauss' law Maxwell equation E .qHowever,
i ni qe ne
since a plasma is a highly polarizable medium, the electric field E in MHD can be
determined self-consistently from Ohm's law, Ampere's law and the charge
continuity equation.
58
Single fluid model – equation of single fluid motion
Adding the electron and ion momentum equations to derive a one-fluid momentum
equation (equation of motion) for a plasma
1
(mi ni ui me ne ue ) ( Pi Pe ) (mi ni ui ui me ne ue ue ) (qi ni qe ne ) E (qi ni ui qe ne ue ) B 0
t c
m e j m e j j j
mi ni ui ui me ne ue ue mi ni u u
me ne u
u
m i n e e
q m i n e e
q n e e
q ne e
q
me me2 me me
( mi ni me ne )uu ( j u uj ) jj ( j u uj ) jj
qe mi ne qe2 qe ne qe2
m m m jj
( mi ni me ne )uu e 2 j j ( mi ni me ne )uu i 2 e
ne qe qe m
m u m i me jj 1
( m uu ) 2 p j B
t qe m c
Using density equation and quasi-neutrality j 0
1
m u u j B p
t
59
c
Single fluid model – generalized Ohm’s law
Multiplying the equation of ion or electron motion by qi/mi or qe/me and sum them,
we can get generalized Ohm’s law.
qi qe
(qi ni ui qe ne ue ) Pi Pe (qi ni ui ui qe ne ue ue )
t mi me
qi2 ni qe2 ne qi2 ni qe2 ne qi q
E ui ue B Ri e Re
mi me mi c me c mi me
j j j qe2 ne 1 qe 1 qe me
uj j u
E u B pe j B ei 1 j
t qe ne me c me c qi m i
me mi j j j 1 mi 1 me ei
uj j u E u B
pe j B 2 j j
qe m t
2
qe ne c qe m c qe ne
61
Single fluid model – pressure equation (or state equation)
Add ion and electron pressure equation, we have
1 p
( p u p u ) p u p u (q q ) Q Q
1 t x i i e e
i i e e x i e ci ce
1 p
( pu) p u (q q ) j
2
1 t x x i e
Like the two-fluid equations, the pressure equation depends on next order fluid
moment. We can not obtain a closed set of MHD equations without some further
approximations. One approximate treatment is to roughly drop out the next order
term
1 p
( pu) p u j
2
1 t x
The chain is also often cut by employing an equation of state under an adiabatic
process.
p C m
In a 3-dimensional plasma, the specific heat ratio γ=5/3; for an isothermal process,
γ=1; In the case with constant pressure, γ=0; for an isometric process, γ→∞, i.e.
p→0 in a very low β plasma.
62
MHD equations
These MHD equations can describes many
physical processes in small gyroradius,
m ( m u) 0 magnetized plasmas — macroscopic plasma
t
equilibrium and instabilities, Alfven waves,
1 magnetic field diffusion. It is the simplest,
m u u j B p
t c lowest order model used in analyzing
magnetized plasmas.
1
E u B j
c
Used approximations used in MHD equations:
1 p 1. Exclude single particle effects such as gyro
( pu) p u j
2
1 t x
motion;
2. Exact charge neutrality for low frequency
1 B phenomena much smaller than plasma
E frequency;
c t
3. The displacement current has been
4 neglected due to the slow propagation of
B j
c electromagnetic waves;
4. Assuming isotropic pressure near thermal
B 0 equilibrium;
5. Different approximation can be taken in
generalized Ohm’s law. 63
Properties of MHD equations
1. Conservation laws of ideal MHD system mass; momentum and energy
m ( m u) 0
t
2
1
m u ( m uu) p B I BB 0 m u T 0
1
t 8 4 t
MHD stress tensor
Reynold Isotopic Expansion and tension of
stress pressure magnetic field
m u 2
m u2
t 2
p
cB 2
1 8
1
c
p u
4
EB 0
2
MHD energy density MHD energy flux w S 0
t
Kinetic Internal Magnetic Kinetic Internal electromagnetic field
energy energy energy energy energy energy flux
flux flux
From energy conservation, it can be seen that in an isolated system, total energy is
constant while exchange between kinetic; internal and magnetic energy
3 m u cB 2
2
p (Using Gauss’ theorem)
V d x 2 1 8 Wk W p WB constant
64
Properties of MHD equations (cont.) –
Details of deriving conservation laws
In deriving MHD momentum conservation, we used
c B 2
c
(a b ) a ( b ) (a )b b ( a ) (b )a jB B B BB
4 4 2
In deriving MHD energy conservation, we used
B ( u B ) B E E B; (a b ) b ( a ) a ( b )
B 2
u ( B ) B ( B ) ( u B ) [ B ( u B )]
2t
From momentum equation,
1
m u u m u (u u) u ( B) B u p
t 4
m u u
2 2
c B 2 1
( m u) m u (u u) [ B (u B)] u p
t 2 2 8 t 4
m u2 cB 2 m u2 u 1
[ B ( u B )] u p
t 2 8 2 4
Adding this equation with the energy equation,
m u2 cB 2 p 1 m u2 u 1
( pu) p u [ B ( u B )] u p
t 2 8 1 1 2 4
m u2 cB 2 p m u2 u c
( pu)
4 ( E B)
8 1
65
t 2 1 2
Properties of MHD equations (cont.)
2. Ideal MHD frozen flux theorem
In an ideal MHD, since the resistivity disappears, E u B c 0 , it can be proved
that the magnetic flux Φ through every surface moving with the plasma is constant,
i.e. d d
dt dt S
B dS B dS B dsˆ 0
S t C
In ideal MHD, plasma carries the magnetic field to move, namely, it always contain
the same amount of magnetic flux. We may simply understand “MHD frozen flux” as
that two fluid elements are always connected by a magnetic field line if they were
connected at one time by a field line, which is defined by the direction of the
magnetic field at any moment in time. In other word, fluid elements flow freely
along the line of magnetic force.
3. Magnetic diffusion
When the fluid velocity is zero, from Ohm’s law we have B t 4 2 B
The magnetic field changes by diffusing through the electrically conducting fluid,
the change rate depends on the plasma characteristics. The field decays in a
characteristic time ~ L2 4 with magnetic gradient length L.
4. Entropy conservation
The entropy is defined as s ln( p m ) , the overall entropy production rate is
d
p m
ds d dp
m u p u m u u 0 66
dt dt dt dt t t
MHD equilibrium
In a plasma equilibrium, ∂/∂t ~ 0, u=0, MHD equations consist of force balance and
Maxwell equations
1 4
j B p B j B 0
c c
Taking the scalar product of the force equilibrium equation with B and j, we have
B p j p 0
It indicates the pressure is constant on magnetic field lines and on current lines.
In space plasmas and also in magnetic fusion plasmas, MHD can be reduced further.
Including the effects of electron viscosity, the equation of motion and Ohm’s law
become
u 1 2
m m u u p j B u
t c
1 mi 1
E u B j pe j B j
2
c qe m c
Define Ψ is parallel vector potential; stream function Φ, let ∂/∂z=0, and constant
density, we can write
B zˆ Bz zˆ; u zˆ uz zˆ; Ez t
uz 1
zˆ (equation of motion ) m m [ , uz ] [ Bz , ] 2 uz
t 4
z 1
[ zˆ (equation of motion )] m [ , z ] [ jz , ] 2 z
t c 68
Reduced MHD equations (cont.) – two-field MHD
zˆ (Ohm' s law ) mi
[ , ] [ , Bz ] 2 4
t 4qe m
When the parallel magnetic field is very strong, the perpendicular gradient is small,
we get reduced two-field MHD equations
z 1
m [ , z ] [ jz , ] 2 z
t c
[ , ] 2 4
t
These equations are usually used to study the tearing mode and the magnetic
reconnection in space plasmas and magnetic fusion plasmas. 69
Classification of equations (Elliptic type—Poisson)
We will summarize
typical equations. From Maxwell equation, in electrostatic limit
we can define E , so the Gauss’ law become Poisson equation
Perpendicular to 2 2
4
2
2 4
magnetic field x 2
y
This is an Elliptic-type equation
2 2
2 0
x 2
y
In plasma physics, it is often to calculate Poisson equation to solve electric field. It
has been paid much attention to look for good Poisson solver in simulation.
70
Classification of equations (hyperbolic type—wave; convection)
Sometimes E & B are expressed in terms of an electric potential φ and vector
potential A:
E ( r , t ) ( r , t ) A( r , t ) B( r , t ) A( r , t )
ct
In this case, Maxwell equation can written as electromagnetic field equations are
written in the form
1 2 1 2A 4
2
4 ( r , t ) A
2
J (r , t )
c t 2
c t 2
c
1
With Lorentz gauge A 0
c t
It can be clearly seen that they are hyperbolic-type equation (for example 1D case)
2u 2u
2 0
t 2
x
Furthermore, the fluid equation system is typical hyperbolic and the equation is of
convection equation m u u p
u m 0 m mu
t x t x x
The solution of such convection problem is the propagation of initial profile F(0,x).
71
f ( t , x ) F ( x ut )
Classification of equations (parabolic type— diffusion)
Another common equation describes diffusion process such heat diffusion, magnetic
diffusion in MHD. For the latter, when the velocity of fluid is zero, the Ohm’s law
becomes
B 2
B
4 2
t x
This is of Parabolic type such as
T 2T
2
t x
In the energy equation in fluid equations, assuming density is constant and using
p=nT, further assuming heat flux q T, hence, we have
1 p T T 2T
( pu) p u q j
2
u 2 ......
1 t x
t x x
When fluid velocity is zero, it is a diffusion (transport) problem
T 2T
2 ......
t x
When heat is isolated, it mainly describes a convection process
T T
u ......
t x
In general case, it describes the mixed processes of diffusion and convection. 72
Final remarks T. Tajima, computational plasma physics, Chpt. 5