Swinburne March 12 2007 1

Magnetohydrodynamics Basics
or MHD in one lecture?! You've gotta be kidding!
Paul Cally, Centre for Stellar & Planetary
Astrophysics, Monash University

MHD Equations

Physical Interpretation

MHD Waves
Swinburne March 12 2007 2
Solar
atmosphere
dominated
by
magnetic
fields
Swinburne March 12 2007 3
MHD

MHD = fluid dynamics + Maxwell's eqns -

displacement current + Ohm's Law
[j=(E+vXB) or generalization]

Applies to (at least partially) ionized gases

(plasmas)

Nonrelativistic

Assumes highly collisional, low frequency

(c.f. cyclotron frequency)
Swinburne March 12 2007 4
MHD Approximation

v
2
<< c
2

Then Ampre's Law simplifies:

The electric field E plays no independent
role in MHD: it is entirely derivable from
v and B:
B=j j +
1
c
2
E
t
E=vB+
j
c
=vB+
B
jc
Swinburne March 12 2007 5
MHD Equations
j
t
+(jv)=0
j
Dv
Dt
= p+j g+j B+viscous terms
B
t
=(vB)+j
2
B with B=0 and j=1/jc
jT
Ds
Dt
=viscous heating + Joule heating - radiative losses - conductive losses+.
Lorentz force

Continuity

Momentum
where j =
1 -
XB = current density; =4X10
-7
(SI units)

Energy (many different forms)

Induction (=electrical conductivity; =magnetic diffusivity)

Swinburne March 12 2007 6
Magnetic Reynolds Number

Induction eqn:

Ratio of advective to diffusive terms:

the Magnetic Reynolds Number

V=typical velocity, L=typical lengthscale

R
m
>> 1 dominated by advection
R
m
<< 1 dominated by diffusion
B
t
=(vB)
_
advective
+j
2
B
_
diffusive
R
m
=
V L
j
Swinburne March 12 2007 7
Ideal MHD Equations

Continuity

Momentum

Energy (many different forms)

Induction
j
t
+(jv)=0
j
Dv
Dt
= p+j g+ j B+viscous terms
B
t
=(vB)+j
2
B with B=0
jT
D s
Dt
=viscous heating + Joule heating - radiative losses - conductive losses+.
i.e. ,
D p
Dt
=c
2
Dj
Dt
( adiabatic; c=
.
p/ j=sound speed)
Swinburne March 12 2007 8
Alfvn's Theorem: flux freezing
In Ideal MHD, the magnetic field and the
fluid are frozen together.
That is: two fluid elements originally connected by a field
line are always so connected.
So what?

If the field is very strong (<<1), plasma is

constrained to flow along it;

If the field is weak (>>1), it is advected

with the plasma.
Swinburne March 12 2007 9
Swinburne March 12 2007 10
Lorentz Force
jB =
1
j
(B) B
=
1
j
B B
B
2
2j
=
(
BB
j

B
2
2j
I
)
_
magnetic stress tensor

Stress:

tension along field lines B

2
/
isotropic pressure p
mag
= B
2
/2

j
Dv
Dt
=( p+p
mag
)+j g+
1
j
B B+viscous terms
Swinburne March 12 2007 11
Example 1

What do you
expect this to do?
or

Pressure pushing
away from high B.
B=( 0, e
x
2
, 0)
j=(0,0,2 x e
x
2
)/ j
jB=( 2x e
2x
2
, 0,0)/ j
B B=0 and B
2
/ 2j=(2xe
2x
2
, 0,0)/ j
jB=B B/ j B
2
/ 2j=( 2xe
2x
2
, 0,0)/ j
Swinburne March 12 2007 12
Example 2

And this?

or

Slingshot effect to the right!

important in flares and CMEs
B=( 2y , 1,0)
j=(0,0,2)/ j
j B=(2,4y , 0)/ j
B B=( 2,0,0)
B
2
=(0,8 y , 0)
So
1
j
BB
B
2
2j
=
(
2
j
, 0,0
)

(
0,
4y
j
, 0
)
Swinburne March 12 2007 13
Plasma beta

A simple measure of the relative

importance of the gas and magnetic
forces

If >> 1 the gas dominates (e.g. stellar

interiors)

If << 1 the magnetic field dominates (e.g.

stellar coronae)

If ~ 1 then get complicated interaction

of field & fluid (e.g. sunspots)
=
p
gas
p
mag
Swinburne March 12 2007 14
MHD Waves

All waves result from a restoring force

Sound waves are produced by compression

(gas pressure is restoring force)

Magneto-acoustic waves derive from gas

& magnetic pressure, and magnetic
tension

Alfvn waves produced by magnetic

tension only.
Swinburne March 12 2007 15
MHD Waves: Linear Theory
j
1
t
+j
0
v=0
j
0
v
t
= p
1
+
1
j
( b) B
0
p
1
t
=c
2
j
1
t
b
t
=(vB
0
)
Assume uniform medium, no gravity,
Set v=V exp[i(kr-t)] and similarly for p
1

etc, so and , to get
=i k / t =i o
o
2
V=c
2
k (kV )+
1
jj
0
( ((VB
0
)))B
0
Swinburne March 12 2007 16
Basic Waves: Revision

Circular Frequency
(radians per second)

Frequency =2
measured in Hz.

Period (seconds) =
1=2

Wave vector k specifies

the direction to
wavefronts

Wavelength =2|k|
k

Swinburne March 12 2007 17

Careful
Three velocities, all (generally) different:

Plasma velocity v

motion of the plasma itself

Phase velocity

velocity of the wave pattern

Group velocity

velocity of energy propagation

v
ph
=
o
k

k
v
gr
=
o
k
Swinburne March 12 2007 18
Sound Wave

Set B
0
= 0

Dot both sides with k to get dispersion

relation

Phase speed (pattern speed)

Group velocity (energy

propagation speed & direction)

Note: V 1 k, i.e., wave longitudinal

o
2
V=c
2
k ( kV )
o
2
=c
2
k
2
o
k
=!c
o
k
=c

k
Swinburne March 12 2007 19
Alfvn Wave

Driven purely by magnetic tension

Incompressive, i.e. v=0, i.e. kv=0, i.e.

transverse (like guitar string)

Then
where a
2
=B
2
/
0
defines the Alfvn speed a

Note: B
0
v=0, i.e. v to both B
0
and k
o
2
V=c
2
k ( kV ) +
1
jj
0
( ( (VB
0
))) B
0
i.e. o
2
V
2
=
( kB
0
)
2
jj
0
V
2
i.e. o
2
=a
2
k

2
Swinburne March 12 2007 20
Alfvn Wave

Phase speed where is the

angle between B and k
v
ph
=a for parallel propagation, 0 for
perpendicular!

Group velocity

i.e. energy travels directly along fieldlines at

speed a

Obvious! It's a tension driven wave.

o
k
=!a cos0
o
k
=!a

B
0
Swinburne March 12 2007 21
Swinburne March 12 2007 22
Magnetoacoustic Waves

Both pressures and magnetic tension

Dot with k and with B

0

Eliminate kV to get magnetoacoustic

dispersion relation
o
2
V =c
2
k( kV )+
1
jj
0
( ( (VB
0
))) B
0
=c
2
k( kV )+
1
jj
0
|
( kB
0
)
2
V( kB
0
)( kV) B
0
( kB
0
)(VB
0
) k+B
0
2
( kV ) k

| o
2
(a
2
+c
2
) k
2
kV+a
2
k
2
k

B
0
=0
c
2
k

kVo
2
V

B
0
=0
o
4
(a
2
+c
2
) k
2
o
2
+a
2
c
2
k
2
k

2
=0
Swinburne March 12 2007 23
Magnetoacoustic Waves

Solve biquadratic:

Two types of wave: fast & slow

NB: they involve both a & c

o
k
=
|
1
2
(a
2
+c
2
)!
.
(a
2
+c
2
)
2
4a
2
c
2
cos
2
0

1/ 2
fast
slow
Swinburne March 12 2007 24
Fast & Slow: Phase Speed
Phase polar diagrams

Phase speed depends on direction of k

k
k
Swinburne March 12 2007 25
Fast & Slow: Phase Speed

Phase speed depends on direction of k

Fast wave is fastest perpendicular to field

and slowest along it (why?)

in range (a
2
+c
2
)
1/2
/k max(a,c)

Slow wave is fastest along field lines

in range min(a,c) /k 0

The Alfvn wave is intermediate.

Swinburne March 12 2007 26
Fast & Slow: Group Speed
Group polar diagrams
Swinburne March 12 2007 27
Fast & Slow: Group Speed

Energy travels fastest across field lines

for fast wave

Slow wave energy transport along field

lines restricted to a cone about B

cone gets tighter as a/c departs further

from 1.
Swinburne March 12 2007 28
Q: How Many Wave Types?
A: Three Alfvn, fast, & slow (6
th
order
dispersion relation).

But what about the sound wave?

No! The sound wave is just a

magnetoacoustic wave (fast or slow
depending if c>a or a>c) travelling directly
along the field lines. Because it is
longitudinal, it does not bend or displace
them, so there is no magnetic effect.
Swinburne March 12 2007 29
Non-Ideal Effects

Magnetic reconnection

Solar corona (flares, CMEs)

massive source of energy

Earth's magnetosphere (interaction with

solar wind)

another day perhaps.

Swinburne March 12 2007 30
L
u
n
c
h
t
i
m
e
!