4.

The magnetohydrodynamic (MHD) description of plasmas

4.1 The MHD equations
• Plasmas are many-particle systems dominated by collective behaviour
• One approach to modelling such systems is kinetic theory – this models the
distribution functions fs(r,v,t) for each species (s) – beyond scope of this
course
• From kinetic theory we may take moments (integrate over velocity space)
and derive multi-fluid models (each species treated as a separate fluid) or
single fluid models
– For full derivation see text books e.g. Gurnett and Bhattarcharjee
• MagnetoHydroDynamics (MHD) treats the plasma as single electrically-
conducting fluid which interacts with magnetic fields
• Use equations of fluid dynamics and Maxwell’s equations
• Valid for:
– Collisional plasmas/large-scale phenomena (λmfp << L)
– Low frequency phenomena (ω << ωp , ωce )
– Sub-relativistic (u, L/T << c ) – usually
• But actually works well for many aspects of collisionless plasmas e.g.
Tokamak; Earth’s magnetosphere!
• Plasma state specified by mass density ρ, temperature T, velocity v and
magnetic field B at each point in space (r) and time t
 Derivation of MHD equations
Derivation 4.1 Starting from Ampere’s Law, Faraday’s Law and Ohm’s Law, derive the
Induction Equation B
    v  B    2 B,
t
where  =1  0 
• Note that conductivity σ is caused by collisions between ions and electrons and so
is temperature-dependent (see Chapter 2). We use
 = 7 X 10-4 T 3 2 mho m-1 (or Ohm-1 m-1 ) - where T in K
(but in derivation above we assume σ is constant)
• The equation of motion or momentum equation is
dv
  p  jXB   g
dt
• The RHS is total force per unit volume: pressure gradients, Lorentz force and gravity.
(Other forces such as viscous drag could also be included). Using Ampere’s law, note
j  B  1 0   B   B
• The LHS is the mass (per unit volume) multiplied by the acceleration seen by a
moving fluid element – this is given by a convective derivative
dv  v
(  v.v)
dt t
• We also have an equation of mass conservation (rate of change of mass in
a small volume is equal to mass flux into volume):

 .(  v)  0
t
• For incompressible flows this reduces to  v  0 kB
• The equation of state is taken to be the Perfect Gas Law p  T
m
where m is the mean particle mass. Note that for a fully-ionised hydrogen
plasma (protons and electrons) m  1 m p
2
or equivalently p  2nkBT
• Finally we need an energy equation. Most generally this can be written
  p j2
(  v.)(  )  .( .T )   Q(T )   H
2

 1  t  
• LHS is rate of change of internal energy (per unit volume)
• RHS is sources/sinks of energy: conduction (where conductivity κ is related
to collision-frequency and is also non-isotropic – different across
field/along field); radiation; Ohmic heating; other sources of heat (e.g.
Viscous damping).
• We will not make use of this in this course – only the simpler adiabatic
equation d p 
 0 γ is ratio of specific heats = 5/3
dt    
 The MHD equations
B 1
   ( v  B )   B ,  
2
,
t 0
v dv 1
 (  v.v)    p     B   B   g,
t dt 0

 .(  v)  0,
t
p  ( k B / m) T ,
 p
(  v.)(  )  0
t 
+ solenoidality constraint  B  0
9 highly-nonlinear, coupled equations in 9 unknowns (components of B, v; p,
ρ, T)
•Plasma flows → magnetic field (induction equation)
•Magnetic field → plasma flows (momentum equation)
 More about momentum equation and Lorentz force
• Using the vector identity
  A  B    B   A   A   B  A   B   B   A 
and setting A= B (using also B.B = B2 ) we obtain an expression for the Lorentz
force:
1  B2 
j  B    B   B       B   B
0  0
2 
= magnetic pressure force + magnetic tension force
• Magnetic pressure is B2/2μ0
• Total pressure = magnetic pressure + thermal pressure = B2/2μ0 + p

• Gravity is often negligible. The relative important of the pressure force and
magnetic force is determined by the “plasma beta” (see Chapter 2)
Thermal pressure p 2 0 p
  2 
Magnetic pressure B 20 B2
Example 4.1 Estimate β in the solar corona – hence show corona is dominated
by magnetic forces
 More about induction equation – Magnetic Reynolds number
Derivation 4.2 Magnetic Reynolds number Show that the relative importance
of plasma motions and Ohmic resistivity is determined by the magnetic
Reynolds number (a dimensionless number similar to Reynolds number in
fluid dynamics) Lv
Re   0 Lv

where L and v are typical length-scale and velocity
• Consider limits of large and small Re seperately
Derivation 4.3 Discuss behaviour in limit Re << 1 (high resistivity)
 Ideal MHD : Re >> 1 (high conductivity)
• The limit Re >> 1 – (high conductivity) is known as
Ideal MHD
• Relevant for tokamaks and astrophysics
– Note this is a singular limit in some senses (see
magnetic reconnection , later)
• The induction equation becomes
B
    v  B
t
• Consider rate of change of magnetic flux through a
   B.dS
S
fluid surface S, bounded by closed curve C
Derivation 4.4 (nonexaminable) – Part 1. Ideal MHD flux conservation and
frozen in fields
Flux Φ may change due to –
•Field B changing in time
•Surface S moving/distorting as fluid moves
d B
  .dS   B.  v  dl 
dt S t C C(t) C(t+dt)
(1) Due to explicit (2) Due to motion
time variation of B of surface S
Substitute from induction equation (1)
and re-arrange vector triple product (2)
→
d
   v  B .dS     v  B .dl
dt S C
Moving fluid sweeps out area
dS = vdt X dl in time dt.
Use Stoke’s Theorem on (2) → Hence change in flux is B.dS
d
    v  B .dS     v  B .dS  0
dt S S
 Derivation 4.4 Part 2. Ideal MHD: Motion of magnetic field lines
• Consider two points on a magnetic fieldline P1 and P2
B
separated by δr
• By definition of a fieldline  r  B  0 δr(t+dt)
P2
• Consider time evolution of this quantity
δr(t)
P1
d d r dB
 r  B   B r Using induction
v
dt dt dt equation and
vector identity    v  B   v  .B 
   r.  v   B   r   
 B.  v  B  .v  -B  .v  +  B.  v -  v.  B
Interchange δr and B
Vanishes since δr X B = 0
   B.  v    r   r    B.  v  For fuller
derivations see,
0 e.g.,Gurnett and
P1 and P2 continue to be connected by a field line Bhattarchajee
Chap 6.3.1
 Frozen-in fields
• In ideal MHD, magnetic fieldlines move with the fluid:
Magnetic field is frozen to the plasma

• The topology/connectivity of the magnetic field is invariant

B v

t=0 t >0
 Alfven’s Theorem
• In a perfectly-conducting plasma, the magnetic flux through any fluid
surface is conserved.

• Example 4.2 A star of initial radius 106 km and magnetic field strength 0.01
T collapses to form a neutron star of radius 10 km. Estimate the magnetic
field in the neutron star.
B

R
Very strong field! A pulsar.
 Example of frozen-in fields – the Solar Wind magnetic field
• The solar wind is a supersonic flow of plasma away from the Sun - typical
speed 350 km s-1 (see Chapter 1 – full derivation later in this Chapter)
• The Sun also rotates every 27 days or so, giving an angular velocity of
about 2.7 X 10-6 rad s-1
• The magnetic field is frozen to the spinning outflowing plasma which forms
spiral pattern (like garden sprinkler!) – hence
B Br  r vsw
• At 1 au this gives a spiral angle ψ (where tan ψ = Bθ/Br about 45o

Particles from flare on Sun travel to

Earth along spiral field lines

See Case study 2: Owens et al JGR Space

Physics 113, A11104 (2008) – shows how
high energy particles propagate along
spiral fieldlines, conserving magnetic
moment (but modified by scattering
which changes pitch angles)
• Since radial field (mainly) has different
polarities in N and S hemisphere, there is a
current sheet in the equatorial (ecliptic plane)
• Note that a current sheet corresponds to
reversal of sign of B due to Ampere’s Law
• The current sheet is warped (“ballerina skirt”)
giving magnetic field in ecliptic a sector
structure
• www.youtube.com/watch?v=2434rAbImf0
 Example of frozen-in fields -dynamos
• Magnetic field in Sun, stars, galaxies, planetary interiors usually generated
by a dynamo – plasma motions create magnetic field

•Differential
rotation generates
toroidal field (EW)
from poloidal field
(NS)
•Also need
mechanism to
generate poloidal
field from toroidal
e.g. Associated
with turbulence
See e.g. P Charbonneau, Living
review of Solar Physics (2010)
Geomagnetic dynamo
 Wave speeds and time scales of MHD
• Since magnetic field lines have tension (T), they can carry waves like
stretched strings – replacing “T” by the magnetic tension B2/ μ0 the wave
speed is
vA  T   B2 0 
• This is called the Alfven speed (full derivation will be given in Section 4.3) –
the speed at which information about changes in magnetic field propagates
• As in a gas, information about pressure changes propagates at the sound
speed
cs   p 
where we usually assume the ratio of specific heats   5 3.
• There are thus two natural time-scales for a plasma of typical length scale
L:
– The Alfven time tA = L/vA
– The sound travel time tS = L/cs
• In a strongly magnetised plasma, the Alfven time is the shorter of the two
(see Example sheet)
 4.2 Magnetostatic equilibrium and force-free fields
• If the plasma flows are weak and the fields are slowly-evolving, we can set
LHS of momentum equation to zero – we have an equilibrium in which all
forces balance
j  B  p   g  0

– This can be done if any flows v are weak compared with Alfven speed and
sound speed and time-scales of evolution T are slow compared with Alven
time and sound time (see Example sheet)
v cs , vA , T ts , t A
• If we can also neglect gravity (always a very good approximation for
laboratory plasmas) we have a magnetostatic (or magnetohydrostatic) field
1
j  B  p where j    B 
0
– Gravity can be neglected if length-scale L is much less than gravitational scale-
height Λ = kBT/mg
• If the plasma beta is also small  1 , then Lorentz force is dominant
and we have a force-free field
j B  0
 Magnetostatic fields in a cylinder
Derivation 4.5 Cylindrical magnetostatic fields In a cylindrical field where all
quantities depend only on r, the magnetostatic equation can be written as an
ODE d  Bz2  B2  B2
  p  0
dr  20  0 r
Total pressure gradient + magnetic tension (circular Bθ lines) = 0
• Twisted cylindrical field sometimes called a “flux rope” e.g. sunspot, coronal
loop (ignoring curvature), large-aspect ratio tokamak, linear confinement
device (e.g. Z pinch), flux rope in planetary atmosphere.....
• Consider a system with only axial field (straight field lines) and thermal
pressure (Bθ = 0) – then total pressure is constant
dptot B2
 0  ptot  p   constant
dr 2 0
• In a laboratory, this is a device in which magnetic field is generated by an
external solenoid (current in θ direction) – a θ-pinch
• Note that where p is large, B is weak – plasma diamagnetism (see Chapter 3)
Example 4.3 A sunspot has vertical magnetic field 0.25 T and temperature
4000 K; the surrounding plasma has zero magnetic field, density 6 X 1023 m-3
and temperature 6000 K. Estimate the density of the sunspot plasma.
 Z pinch (linear pinch) and Theta pinch

Current in z direction,
field in θ direction

Current in θ direction,
field in z direction
Lightning – inward
collapse due to Sunspots – roughly cylindrical bundle of
current - similar to Z magnetic field – similar to theta pinch
pinch
 Z-pinch (cylindrical pinch)
• Consider linear magnetic confinement device in which plasma current in z-
direction generates magnetic field (in θ direction)
• Plasma is confined by inwards force due to tension of circular field lines Bθ
– equivalently, current I provides confinement
Derivation 4.6 Discuss force balance in Z pinch and derive the Bennett relation
8 N line kBT
I2 
0
where Nline is line number density and I is total current
• Note that on large-scale, Z pinches are unstable and hence are ineffective
at confinement (see later)
• Can achieve stability in a screw pinch (use both Bθ and Bz)
• But very small, dense Z pinches with rapid plasma due to current pinch
could be effective fusion devices
hybrid of inertial and magnetic
confinement
 Dense Z pinches and wire arrays
See Case study 3 – Report on Z
pinch approach to fusion – J P
Chittenden, Physics World May
2000

Wire array on
MAGPIE dense
Z pinch at
Imperial
 Tokamak confinement (revisited)
• Pure toroidal field Bz ( leads to loss of
particles through drifts (Chap 3)
• Pure poloidal field Bθ is unstable to kink
instability (later in Chap 4)
• Tokamak has both components (combined
θ and Z pinches bent into torus to avoid
end losses)
• The pitch of the helical field is defined by
the safety factor q = number of toroidal
rotations of fieldline per poloidal rotation
– Tokamak: q is mainly > 1 and Bz >> Bθ (strong
toroidal field – weak plasma current)
• If the minor radius is a, the aspect ratio is
defined as R0/a (≈ 3 for JET, ITER) -
cylindrical model is good approximation for
large-aspect ratio
where R0 is
• Equilibrium must be corrected for toroidal rBz  r 
q(r )  the major
effects (e.g. Toroidal field stronger on R0 B  r  radius
inboard side)
 Force-free fields
j  B  0   B   B  0
• This means currents flow parallel to magnetic field
j // B or    B  // B
   B    r  B where  is a scalar
• Cylindrical force-free fields satisfy 1 d 2
2 dr
Bz  B 
2
B2
r
0
• Given either field component (or some relation between them) the other
component is determined
Example 4.4 The angle rotated by a field line Φ (= 2π/q) is given by
L L
B LB
  r    d   dz  
0 0
rBz rBz
In a constant-twist field, Φ is independent of r - this is produced by solid-body rotation
of the footpoints of the field. Calculate the components of a uniform-twist force-
free field and discuss force balance.
 2D force-free fields and Grad-Shafranov equation
• Consider a field which depends only on x, y - -independent of z
• Since div B = 0. we can express field in terms of a flux function ψ
Bx By
B  0   0
x y
  x, y    x, y 
 Bx  , By   or B    x, y   zˆ  Bz zˆ
y x
• It can be shown ψ is constant along field lines and also ψ is z component of
vector potential (exercise for student)
Derivation 4.7 Show that a 2D force-free field satisfies the Grad-Shafranov
equation dB
 2  Bz z , where Bz  Bz  
d
• Nonlinear partial differential equation.
• Special case:
Potential field (j = 0) – Laplace’s equation    0, Bz  0
2

• Similar equation can be written for axisymmetric fields in cylindrical

coordinates (more useful for tokamaks and coronal loops) (see textbooks)
 TRACE loops
The late “Transition Region And Coronal Explorer”
 Reading list for Chapter 4 – part 1
• Gurnett and Bhattarcharjee Chapter 6

• Friedberg Chapter 11
• Kulsrud Chapter 3
• Boyd and Sanderson Chapters 3, 4

• www.sandia.gov/z-machine/
• http://www.spaceweathercenter.org/living_with_a_star/01/wrath.html

Case studies:
• J P Chittenden, Physics World May 2000
• M Owens et al JGR Space Physics 113, A11104 (2008)