4

www.moffatt.tc
WHO1 - 78 - 6 7

NOTES ON THE 1978 S M E STUDY PROGRAM U MR ON DYNAMO MODELS OF GEOMAGNETISM IN GEOPHYSICAL FLUID DYNAMICS AT THE W O S HOLE OCEANOGRAPHIC INSTITUTION OD

Willem V. R . Malkus, D i r e c t o r and Mary Thayer, E d i t o r
8

WOODS HOLE OCEANOGRAPHIC INSTITUTION
Woods Hole, Massachusetts
02543

November 1978

TECHNICAL REPORT

Prepared for the Office of Naval Research under Contract NO001 4- 78-G-0072. Reproduction i n whoZe or i n part i s permitted for any purpose of the United States Government. This report should be c i t e d as: Woods Hole Oceanographic I n s t i t u t i o n TechnicaZ Report
WHOI-78-67.

Approved for pub l i e re lease; distribution un 2i m i t e d.
Approved f o r D i s t r i b u t i o n :

Robert W. Morse Dean o f Graduate S t u d i e s

(ZfLJ? b . , / i ,

The oxymoronic role of molecular diffusivity in the dynamo process

H.K. Hoffatt

Ab s tract

The delicate question concerning the behaviour of the regeneration coefficient
Q

and the turbulent diffusivity

B

in the limit of

vanishing molecular diffusivity

(n

+

0 in helical turbulence is )
&

discussed, in the light of an exact result of Bondi viz. when q
= 0

Gold (1950)

the external dipole moment of a current

distribution in a sphere is permanently bounded.

§1.

The oxymoron is a figure of speech which embodies an apparent contradiction; e.g. triviality, etc.
(= (p0o)-')

creative destruction, relaxed tension, devastating
rl

The oxymoronic role of molecular diffusivity that while non-zero diffusivity
(rl >

is this:

0 is directly )

responsible for the natural ohmic processes of dissipation and decay, it is also indirectly responsible for the means of regeneration of the magnetic field; the dynamo process may be described as a process of 'regenerative decay', or perhaps better 'reinvigorating dissipation'.

§2.

Consider the dipole moment p(t) distribution d(x,t)

-

associated with a current
<

-

= po-lV

, ,

in a conducting sphere V:r

a.

This is given by various alternative expressions:

where S

is the surface r = a ; and its rate of change is given by

With

=

-AB +
S

rlV

, and

n.u = 0
5 -

on S, this gives

The first term on the right describes the mechanism identified by Bondi Gold (1950) for increase of the dipole moment:

&

field sweeping towards the
p )

magnetic poles (defined by the instantaneous direction of the vector can increase

1p I

, but,

as emphasised by Bondi
0, since

&

Gold, this mechanism is

strictly limited when when all the flux of

rl =

Ipl

then attains a finite maximum

is concentrated at opposite ends of a diameter To see this explicitly

of the sphere (as in an elementary bar magnet).

from the above equations, .let S- denote those parts of S on which ,

n.!

>

or < 0 respectively, and let

c

- 2 -

so

that y = !+

+

F-

We then have

where

@ =

1
s+

(.3d gJ)S

=

-

I
s-

(.3d. gl)S

Now, when

rl =

0, @

is constant, since flux through every closed material

circuit is conserved, and so

Ill1

5

Iy+l

+

1F-l

i -a@ 4.rr

3

(7)

the maximum being attained only when the flux is entirely concentrated at the poles, as mentioned above.

§3.

, There can therefore be no doubt that, when 17 = 0 exponential

increase of the dipole moment is impossible, no matter what the complexity (laminar or turbulent) of the velocity field in V The situation is transformed if
r)

may be.

> 0, because then diffusive increase

in the dipole moment (represented by the second term of (3) is possible, provided the velocity field is such as to maintain a field with a suitably negative gradient near the boundary r=a.

§4.

The impossibility of sustained dynamo action (in the sense of an exponentially increasing external dipole moment) applied equally to such basic systems as the homopolar disc dynamo. If the disc conductivity is

0

infinite, then the magnetic flux across it cannot change with time, and exponential growth of the magnetic field associated with the device is impossible no matter how fast we rotate the disc or how ingeniously we twist the wire, and whatever conventional wisdom may tell us to the contrary. In terms of growth rate, if, in general, B
Q.

ept, then p

must depend on the disc Reynolds number Rm figure 1.

in the manner indicated in

It is reasonable to conjecture that fluid dynamos also must

behave in this manner.

c

- 3 -

Figure 1 .

Possible dependence of p on Rm for homopolar disc dynamo. () a: wire,resistance zero; () b: wire resistance non-zero.

In either case, p

+

0 as Rm

+

-.

§S.

Consider now the situation in mean-field electrodynamics, in which, in conventional notation, i where
b = B-B
=

<!!-pi ijBoj + =
~1

Bljk a Bo j ..

/ a t +

0 . .

(8)

Bo(x,t)
-0

=

<g(x,t)>

is the large-scale (mean) field, and

-

.

Under first-order smoothing theory (Moffatt 1978

- hereafter

referred to as M-chap.7) we have the results

where F k w , (,)

E(k,w)

are the helicity and energy spectrum functions of If E(k,o)
O(u2)

the random u-field. F(k,u) then clearly
f

= O(w2),

as w

+

0 ,

(11)

cc

a.

rl

,

B

5

~ ~ ' r l as

n

+

0,
> 0. )

(12)

where

oo'

and

i3,'

are in general non-zero constants (B0'

This

is clearly t h e situation when the y-field is a field of random waves with
no zero-frequency ingredients.

In this case, the regenerative process
rl

normally associated with the pseudo-scalar a vanishes as

* 0,

- 4 -

consistent with the remarks of of Bragniskii (M,

81.

It may be noted t h a t t h e theory

chap.8) gives an expression.for t h e regenerative

c o e f f i c i e n t very s i m i l a r t o (9), and again with t h e property a = O ( Q ) as
Q
-F

0.

86.

D i f f i c u l t i e s a r i s e however i f the ?-field has non-zero s p e c t r a l density a t
w = 0, as

i s the case f o r conventional turbulence.
*

The

zero-frequency

ingredients of the turbulence a r e p r e c i s e l y those t h a t

a r e responsible f o r the dispersion of p a r t i c l e s i n a turbulent flow, and they are of v i t a l importance a l s o i n the field-line-stretching context.
It must be noted however t h a t r e s u l t s such a s < & >

-

2

Q

2Dt

f o r the r e l a t i v e dispersion of two p a r t i c l e s separated by vector distance S(t)

-

i s ultimately limited by the physical dimensions of
and care may then be needed i n carrying over

the f l u i d domain;

asymptotic r e s u l t s from s t r i c t l y homogeneous turbulence t o turbulence i n a f i n i t e domain, p a r t i c u l a r l y when these r e s u l t s are s e n s i t i v e t o the l i m i t i n g ( t
+

=) behaviour.

§7.

When rl = 0, there i s an a l t e r n a t i v e approach t o the determination of the c o e f f i c i e n t s instant
t=O,
a
U

and and
b

8

using Lagrangian averages. f i e l d s a r e uncorrelated, then t=O).

I f a t some
a

the
t

-

-

and

B

are functions of

(which c l e a r l y vanish a t leads t o t h e expressions

The Lagrangian

procedure

(M, g7.10) t

0 0

where

v(t)

i s the velocity of the f l u i d p a r t i c l e i n i t i a l l y a t

- 5 -

position

a.

-

The d i f f i c u l t y h e r e i s t o determine how t h e s e e x p r e s s i o n s
t
+ a .

behave f o r a t y p i c a l f i e l d of homogeneous t u r b u l e n c e as

Kraichan (1976 a , b ) has argued t h a t , i n t h e case of t u r b u l e n c e w i t h non-zero h e l i c i t y , a(t)
't

a.

y

B(t)

,l4

B,

as

t + -

,

(15)

t h e a p p a r e n t p o s i t i v e d i v e r g e n c e i n t h e second term of (14) b e i n g c a n c e l l e d by an e q u a l n e g a t i v e divergence i n t h e t h i r d term, which i n v o l v e s t h e awkward t r i p l e Lagrangian c o r r e l a t i o n s . Kraichnan's

arguments r e s t i n p a r t on comparison w i t h t h e r e s u l t s of f i r s t - o r d e r smoothing t h e o r y i n s i t u a t i o n s where b o t h approaches ( f i r s t o r d e r smoothing and Lagrangian) may b e expected t o b e v a l i d , and i n p a r t on numerical e v a l u a t i o n of a(t) and

B(t)

f o r velocity f i e l d s with

prescribed Eulerian statistics.

F u r t h e r numerical e x p e r i m e n t a t i o n

i s needed however, b e f o r e t h e r e s u l t s (15) can b e regarded as
a b s o l u t e l y and d e f i n i t i v e l y e s t a b l i s h e d .

Let u s n e v e r t h e l e s s a c c e p t
a2- and

(15), and pursue t h e consequences
dynamo m d e l s .
88.

i n t h e c o n t e x t of

au-

For an a2-dynam i n a s p h e r e

r < a

(M. chap.9), t h e growth

rates have t h e form

where

"e

a " + @ ,

and

R
a

= I a [ a2 / q e

,
F(R,)
> 0.

and dynamo a c t i o n o c c u r s when

This g e n e r a l l y o c c u r s

f o r t h e s i m p l e s t mode of d i p o l e symmetry when

where

Rac i s a p o s i t i v e number of o r d e r u n i t y which depends on t h e

p r e c i s e assumption made about any l a r g e - s c a l e v a r i a t i o n of throughout t h e s p h e r e . L e t u s suppose t h a t , as q
+

a

0, t h e r e l e v a n t

c

- 6 -

behaviour of a and
a % a
0 ,

B
%

Xcf 15) is
8,

B

as

T I

+O.

Then (16) becomes B P ‘L FGa a The condition Ra and then p

*

L

Ra

=

bob2
60

is certainly satisfied if a is large enough,
T-I
-f

-

tends to a strictly positive value as

0, implying

exponential increase of the mean field, and in particular of the external dipole moment.. This appears to be in fundamental conflict with the Bondi
&

Gold result (7), which applies when

rl

= 0 whatever

the complications of the velocity field, and whether laminar or turbulent. The conflict does not arise under the alternative limiting behaviour (12). In this case,

and the dipole mement does not grow exponentially in the limit
§9.

rl =

0 .

For dynanros of ao-type, growth rates are generally given by

where rotation.

is a measure of the shear associated with differential

The condition for dynamo action is now of the form
v

X’XC

(23)

where Xc is model-dependent, but generally of order unity. under the behaviour (19), as
rl
-t

Again

0,

and we encounter the same fundamental conflict with the Bondi & Gold

result

.

Under the alternative behaviour (12),

- 7 -

To determine the behaviour of p behaviour of F X () p
-+

as

q

+

0, we need to know the
=

as X

-+

0. 0

If F X ()

oX ()

as X *

00,

then The

0 as

r)

-+

0, and conflict with Bondi & Gold is avoided.

asymptotic behaviour of F X ()

as X

+

Q)

does not appear to have been A clue is

investigated for aw-dynamos in a spherical geometry.

however provided by the results for an aw-dynamo in a Cartesian geometry (modelling the galactic disc). solved completely (M, §9.9), fX () and so
§lO.

For this case, which can be

log X
rl +

as

X

+ 00

p * 0 as

0 as required.

It is hard to escape the conclusion that the result (19) cannot be
correct, or that, if it is correct in homogeneous turbulence, it is, for some deep reason, not applicable when the turbulence is confined to a

6. finite region (see the remarks of $ )

References Bondi H. & Gold T (1950) . Kraichnan R H .. Moffatt H.K. (1976a,b) (1978) Mon.Not.R.Astr.Soc. 110, 607-11.

---

J. Fluid Mech. 75, 657-76 and 77, 753-68. -Magnetic field generation in electrically conducting fluids (C.U.P.)

--

Acknowledgement

I am grateful to John Chapman who helped me to sort out the argument of 52.