ISI

J International.

Vol.

29 (1989), No.

6, pp.

462~468

Three-dimensional Velocity
Melts Produced by
O.J.

Fields

for

Newtonian and Non Newtonian

a Rotating

Magnetic Field

ILEGBUSI and J. SZEKELY

of Technology,

Department of Materials Science and Engineering, Massachusetts Institute

(Received on June 27. 1988, accepted in the
final

Cambridge,

02139. MA

U.S.A.

18. 1988) form on November

A model is developed to calculate fluid flow in Newtonian and non-Newtonian systems subjected to rotational electroIn the former case, transport equations for K, the turbulence magnetic stirring. energy, and L, its rate of dissipation, were used to deduce the effective viscosity, while in the latter case, constitutive relations were used to relate the shear stress to the rate of strain. The body forces due to the rotating magnetic field are deducedfrom the established analyiical results obtained from the solution of Maxwell's equations.
velocity; KEY WORDS:

Newtonian; non-Newtonian; rotating;

magnetic field.

Introduction

stirring of continuous casting has gained widespread systems acceptance in recent excellent this subject is available of review An years. in Refs. l) to lO). Furthermore, in a recent paper Spitzer et al.11) have shown very good agreement between the theoretical predictions and experimental measurements conducted using a mercury model
fluid.

The electromagnetic

system to a somewhatidealized field confi~uration, but these results will, ofcourse, bc helpful in designing
a real rheocasting system.
2.

Formulation
Let us consider a cylindrically-shaped

Most of
tively

this work has been concerned with. relamodest fields, Iargely in the millitesla range,

metallic sysconsisting in of carbide which boron magnesium tem is particles stirred This suspended. melt by a are rotating magnetic field ofstrength Bo, such as sketched in Fig. 1.

with corresponding melt velocities rarely exceeding l m/s and often being significantly below this figure. Indeed, the electromagnetic stirring literature sugin the range of about 50 gests that melt velocities cmls are likely to provide good operating conditions. The purpose of this work is to explore the effect of significantly increasing the magnetic field; under these conditions, one can expect two things to happen : (1) The melt velocities are likely to increase sigslurry will systems, a melt-solid form, giving rise to the conditions usually termed " rheocasting ". Indeed, the works of Flemings and Mehrabianl2) and Danzigi3,14) suggest that these rheocasting conditions would be attained by shear s~1 rates in the range of I OOO Bo_th points (1) and (2) are essentially qualitative. Before an experimental program may be rationally designed, one needs information on the actual field strengths that are needed to provide these critical shear rates and to examine how a non-Newtonian fluid containing a fairly high fraction of suspended solids would respond to these fields. Or, in other words, what fleld strengths would be needed in order to produce acceptable velocities in melt-solid suspensions
nificantly. (2) For alloy

The question
fields

is,

for

both turbulent

then, to calculate the melt velocity and non-Newtonian condi-

tions.

This system maybe represented the equation of continuity:

az

by writing down
.(1)

a (u)

la ~7(rv) +T =O
au
ar

...

..

..

and the three componentsof the equation of motion :

Axial

momentum
- ar

P(v au

-+u-az
J

au
::=

ap I a - ~!+7 ar

rpeff

au a uoff7~ +2D~ oz

-+-

av )]

az
(2)
,

+F
~~

T E
u
Lr)

/Rotating
Magnet

Car Boron Carbide/ Magne Magnesium s ium Slurry

E
o
L~

or Pure Magn Magnesium

h-5,

7cm-+
stirred

In

this

paper we shall examine the response of the

Fig.

1
.

Schematic of the electromagnetically

system.

462

C 1989 ISI J

ISIJ International,

Vol.

29 (1989), No. 6
'_u'Vp~

Radial

momentum
av
ar

P(v~~+

u az
au
--ar
eff

ar - ~-

ap

+1[P
az

+~ az +

av )]

I +ar r -

= -VP V r+FE
force

.........(11)

2rpeft

av
ar

in

which ~E

is

the electromagnetic

per unit

volume.

r2~ Pw

2peffv
r2

+ Fr
..(3)

Weshall assumethat the semi-slurry system obeys relationship thus: power-law a

r ~==

Azimuthal
p(v

momentum
6lw
o~~w

ar

+u a~-:=:
ff

a + az pe
2. 1.

7 aw vw P +F ~r
ar

I a

peffr 3

~(wr )]
(4)
.

-m

V~(A A) n-1 }A
=

.(12)

where, Ais the rate of deformation tensor and mand n are empirical constants defined by the relations :

az

m= e(9.783f.+1'4345)

..........................................(13)

Turbulent Flow
viscosity
is

The effective

defined in terms of the

in

n=
It

O.1055+0.4lfs _O.308+1.78f.,
is

for for

0.15:~fs 0.03~f.s~0'60

........(14) ........(15)

molecular and turbulent componentsthus:

Peff

whichf,

the solid fraction.

= p+pt

"" "

properties of obtained from two its lr and turbulence, namely the energy rate of dispt
is

characteristic

sipation

e,

from the relation:

pt=CFPK2/e

........................(6)

K and e are,
p(v

in turn,

calculated

from the following

transport equations :

aK ~( ~i r aK ar ar + u az = --

aK

+aaz(~K __L aK
and

az

+GK-Pe

.(7)

p(vae

-+u -or az

as

Ta'~~r

I a (Pt ~i)+
a'
(~r

_~(Pt aa)
az
(T=

~~

+C K(GK+S)- C2p,r
Eqs. (7) and (8) tion ofK, and is dcfined as:

"""
is

"-"

"'(8)

The term GKin

GK= ;L ,

the rate ofproduc-

should be noted that the precise nature of this relationship has been deduced for a lead-tin system by Joly.16) It is not immediately obvious whether will hold for a magnesiumjboron this relationship it is likely that some incleed, modificacarbide system; tions will be needed. Work is currently in progress It is felt, however, that the general in that respect. particularly the shear-thinthese of systems, nature ning beh_avior, would be well represented by a power law type equation. Wenote here that the terms Fz' Fr and Fe in the equations (2) to (4) represent the appromomentum electromagnetic body force field components. priate quantities be These may evaluated through the maniDetails of the estipulation of Maxwell's equations. mation of these fbrce componentsfor a rotating magnetic field, assumedsinusoidal in time and in ang_ular and will not be coordinates, have been given inll) that the system of shall repeated here. We assume l) calculating the electrofor equations used in Ref. I field thc present applicable will force to be magnetic will, of course, behavior non-Newtonian case. The not affect these considerations and the fact that a finite, long cylinder is being rather than an infinitely used is thought to be a good flrst approximation for obtaining information on the general nature of the
svstem.

+2 2() ar
av 2

[(

+
The term

--+ar az +
Eq. (8)
et
is

au

av 2

au _) 2 aw + *~z + az [r~(~ )J} 2

'

"""""""

.(9)

Thus the
netic fleld

force

componentsfor a single-pole

mag-

with amplitude Bo are:

Fz Fr

fluence of streamline

Sin

Following Spitzer

used to account for the incurvature on the turbulence. al.,15) who assumeda suggestion
is

due to Bradshaw,13) this term

defined thus:
.(lO)
2. 3.
is

w S= Cspt7(~r
in
l.6.
2. 2.

l B~(w-T)2 = ~~ B~(w w ,~r F I

z~'

=O

.........(16)
...
~:2

por 3

...

""""'(17)
.(
. . . . .

(~~)

~r

8) 1

Boundary Conditions

which C*

is

an empirical constant whose value

Flow

JVlon-JVle~)tonian

At high solid loading, the flow becomesnon-Newtonian. For this situation, it is convenient to write the equation of motion in vectorial form in terms of the stress tensor r thus:

The boundary conditions needed to specify the following physical constraints are : -symmetry about the centerline ; slip at the solid surfaces introduced through wall -no functions; and -zero shear at the free surface. In these calculations we did not allow for the deformation of the free surface, which was certainly an
463

ISII

International.

Vol.

29 (1989), No.

thus, the computed results to be oversimplification; presented would correspond to a system where free surface deformation was artificially prevented, e.g., through the use of a " frictionless lid ". This assumption is not thought to introduce a serious error as far as the general nature of the system is concerned.
3.

~//'/'t;~;~///~~~~~~~~~=~~'~~~~'-~_-

I~~~~---~~'=~~=~"~'::':.....\\\\"'~~""~"~"':::::,::::::::::\~\:~

~-=__~
:/1//~ ~~~///

-]-___1:_.*~*

~~+
\\~\'

~~\~:~~

-~-~~'~::_~~~~.

r///ll'/r/"

':t:i'\jitittiiiiiiii

Methodof Solution

!i!!!!!!!!((\!\:~~i}!//////////"///////!i!{
i::::;:::::::;::"\:~j,_:i,;\,-.j:~~~'-~-~;-.-~-.--';~/t//'~~t~////:/""~

The governing equations were put in a finite difference form and the.n solved using a modification of the PHOENICS computational package.17) The principal novel features were : -the allowance for the electromagnetic force fie]d ; -modification of the program to represent non-Newtonian behavior; and -modification of the turbulence model to allow for the extra source term due to swirl. A spatially non-uniform 28 x 18 x 36 grid was used, and a typical computational run required about 3h on a MicroVAX11 digital computer for acceptab]e convergence.
4.

::::i;;.~~4~~~~~~~~

+~/://~'P

$';
..*,,~
....~,..~,+.+*,.',**~

'

-~

~~
.++

(a)

ii)i/:~~~:~~"~~~~~~=~=~~~--~~-=-~

~~:~:'~;::;i:::.~::~:"'\'~"~"'~""\"'~~~~'~',::;:::::::::;'::::::>,i'~:~':':~:~~

:"'//'7//~~;~:;1

Results

selection of the computedresults is given in the following. The principal input parameters employed in the computation are summarizedin Table l. Figs. and 3((a) to (c)) show a set of the computed azimuthal velocity flelds. In all these plots (a) refers to a position close to the

7///~"~/_-=__]'_'-~+'\'\\"~:
!!!i!!!ii/!/!/!/!;/'rv~"i'
'

~{~~tiitiiitiittlii~

'~~]\:::,.;;:::::.\:L:~\::;:::::i::::::;;::;:.-~:~i:::::::i::~~:!{::{::_-.,_-__~;_':':J..;!:~~;~,_../;~~~i../t4j:(::~/';///;;~/!/:;//';~;;/;/:~;:/'r"

\
'

\--\+-:;~~~~i:+';__,>i~i+':!::

"'

the top surface (z/H=0.8). It is seen that there is little variation with vertical position. 2 depicts a system with a field strength of 1.25 kG, while Fig. 3 shows the behavior of the system with a field strength of 0.5 kG. Both these refer to turbulent conditions, i.e., a solid fraction of zero. Figs. 4(a) and 4(b) showthe radial and axial variation of the maximum values of the azimuthal velocity, for turbulent flow (f.=0). respectively, The strong spatial dependenceof this variable is readily noted in Fig. 4(a). As expected, the azimuthal velocity shows a very marked radial dependenceand decrease quite sharply as the central part of the system is being approached. Figs. 5(a) and 5(b) show plots of the radial and axial variations of the maximum rate of turbulent Onceagain, there is ene-rgy dissipation, respectively. strong spatial depenclence with maximum values occurring in the top core region. However, the imrelatively Fig.

bottom of the container (z/H=0.2), (b) to a position about midway(z/H=0.5), and (c) to a plane close to

'\~
_--'~'+*~"'*

..

~~;~"~~~
'

(b)

._=_~~~:'~:~::::__

~~:~.;~~(

~--~

Table
Melt
Solid
Particle

1.

Proposed input parameters.

Megnesium
density
.

Melt densitv Kincmatic viscositY Amplitude of magnetic Angular velocity of field

Boron carbide 2 52 x 103 kg m~3 I 81 x 103 kg.m~8

.

'

(a)

field

9x l0-7 m2/s O-I .25 kg 3OOO rpm

6.

(b)
(c)

7 M/S z/H=0.2 forB0=1.25kG and.f*=0 z/H=0,5 for B0=1.25 kGandf* =0 ~/H=0.8 for B0=I ,25 kGandf* = O
field.

Fig. 2.

Velocity

464

ISIJ International,

Vol.

29 (1989), No. 6
15

4'-~~t-1~__

~_~~:\1'~~:~~~:~;~~~

O
f.
I~

~'~~
~~\~~\~~t\\~\\\~~~:~;~~(~~:

r//::::/:7:r/::/,:///iE:~

~'____;

12.5
~\~~t
~~~

~ii:.\\\\~::::\\~~:,\\~~\\

~
li!!1!!!!!!i!
Illllliiitt~~\\

~~~/;~

10

O
B = 125KG

~ \~1\:;t lil~h):_~_ / /!/// / fitt / / /////// t ~~\ \\\\\\\\\'-_1)~)i//::/~;///////////:1//! ~

fI

flttttff~

\"~
~

75
5.0

B. =
"l!!"'{1"I

O50KG

I
"
~~:~~\\~

;!!

~"~\1L \~~~~~~~~
.

*,\

_\~\\~\~~~

'\

_*1__*I]

' ii'

25

'

'/("/d!/r///~~/
~\~~~1
'~~~\~~'~L~

f
Fig. 4(a).

~'~~~~~
-\~:~~~\~-\\'~~i~'^\

~~ ~;

/!__r ~-

/~/~_

OO O
=

O2

O4

~;
~~~~

~:

*=.

~;i~~:~'~/;~~/~~:

/~ _

'/'

r/R

O6

0.8

~;

(a)

Radial variation of maxirnumazimuthal velocity forf* Oas a function of magnetic field.

15

O
is

1~.5
10.0

B. =

I e5KG

~ \~
,,)

75 50 25 Oo
O
B
=

050KG

0~

04

~'(~/~~4~~:~::~~

Fig. 4(b).

Axial variation

~~>*\'*:~~+~

forf*=0,
(b)
:::e:

of as a function of magnetic

z/H azimuthal maximum

field.

06

08

1
velocity

j;~~~~r/~_\_~~_1~'~::~'~~;~.
/~;////~i////
.

'

, ll ///////////:,:/tl,/;~~~:.-__

~:
_~'~..~

'h~'~'\~'\

.~

\
\\\\

\~~
\\~\h'

portant point to note here is the quite high value of a. Fig, 6 depicts a system with a solid fraction of 0.4, The strongly non-Newtonian behavior. exhibiting azimuthal velocity with magvariation of maximum netic fleld as a function of solid fraction is shown in Fig. 7. The effects of solid loading on the spatial azimuthal velocity are variations of the maximum

shownin
tt

!!ii!

!j!i!!ii

!i !!
,~~
\\\~~lL'~\'~~~

41;~f-_..~tL~r\1
.

/tll~~~!f~T~~r:~~j

~t::::~~
~~

1'
:.

~il

~ l

~~ \~~~~:~\\~~\\\\\\:::~

t~
..

~,,'iE~il-'

~
'

'r'~.:~~v~~

! //////////://
ll
//////

t I t tt

tt
/

titlllttft

Figs. 8(a) and 8(b). Examination of these plots showsthe following: (1) Comparisonof Figs. 2 and 7, and observation 6, Fig. of show an almost linear relationship between

~~ '~'~ t~ r
"'

the
'

'!""""'~

.
i~~~:i~'\

///~/~:./_r

//,~/!//'~v~1~"

;~/:

~:\\~\'L\~::~~~)~~

_~
-L\('~~~~~~

~l~/!

'l

___

.:~4
~4~~~~

(c)

~~ ~~~
(a)

(b)
(c)

z/H=0'2 fOrBo;=:1'25kG andfs=0 z/H=0'5 for Bo:=0'5kG andfs=0 z/H=0'8for Bo:=0'5kG andfs=0
field'

6'O M/S

F g'

3'

velocity

field strength; the expected behavior, which may be found from order-of-magnitude approximations as well as from the prior computedresults of Spitzer et al.11) (2) Perhaps of the greatest practical interest is the behavior of the non-Newtonian melt with the 40 o/o Solid fraction, where the velocities are seen to exhibit a maximum somedistance from the wall; this relationship. is due to the nature of the constitutional will local shear specifically, be at its the rate More at someposition near the vertical solid surmaximum in the face, which in turn will cause a local minimum The net effect of this behavior is apparent viscosity.

maxirnumvelocity and the applied

is

this

465

,l

ISIJ International,

Vol.

29 (1989), No.

2000

/////
f.
!=

~\~\
4rl J~f~~t
~'~---_~h
~~"rlr~~:~~1~~~\~~~,~L~:~(~~\~~~

O
'!

1500
co

lll!'1 //~///

-~f-~ l~F~~

'~-___ ~~~:~t-"~

\h
1~

~~'~

.\~
~'

e~

1000

500
B. =

125KG

B = O50KG

o
Fig.
5(a).

02
rate

04
of

r/R

06

08

1
energy
o_

Radial variation
dissipation
netic field.

turbulent maximum
as a function

forfs=0,

f mag-

(a)

3000
fs

~~~/////////~~~~~::~~~~~::~:~~~~~~~~

O
i___"_}
~"~//'~:/~~:~:;t~-;////~:/~/~//_//_

~~)~~~~~~~~'\'~

2500
2000
cl,

i.\\\'~::::':\\~~tj'\'

P~'1
'i/;'/'//

B. =

I ~5KC;
,i/~/!/f/~f!//////////////f~/;?

C\2

1500 1OOO
500

~~"I~'\ili))f
.

B = 050KG

'::::::::\:~::~:::j~ili::~_.~=::/;_:~:://7:///;i/://;;;,!;;/:;::'////

~}

if

ifffifffffffff

1
~\'
~,\\

'//'

4
,,

o
Fig.
5(b).

0~
ratc

04
of

Axial variation
dissipation
netic field.

z/H turbulent maximum

as a fhlnction

06

C8
of

1
energy
(b)

\+'~~~))~*~::i~~

ibrfs=0,

mag-

in the melt velocity a rather sharper local maximum than would be expected for turbulent flow conditions.
5.

~~
-///

4-~ ~~=

~:L:'~

Discussion

mathematical formulation has been developed to body of pure represent the fluid flow in a cylindrical magnesiumand a magnesium-boroncarbide slurry, when agitated by a rotating electromagnetic force fleld. The appropriate electromagnetic force field equations were taken from the prior work of Spitzer et al.11); in the statement of the fluid flow equations, allowance was madefor either turbulent behavior in the case of pure magnesium, or ibr non-Newtonian behavior, in the case of a magnesium-boroncarbide
slurr y.
all of the prior published work Virtually on the electromagnetic stirring of molten metal systems has been confined to relatively low flelds, such that the

(c

)
(a)

corresponding melt velocities tended to be below about I m/s or less. Furthermore, up to the present, no published data have been available concerning the

(b)
(c)

z/H=0.2for B0=1.25 kGandf*=0.4. z/H=0.5 forB0=1.25kG andf*=0.4. z/H=0.8 forB0=1.25kG andf*=0.4.

field.

5.0

l(/S

Fig.

6.

Velocity

466

ISIJ International,

Vol.

29 (1989), No. 6
that were drawn from this earlier paper were the associated with electromagnetic inherent difliculties rheocasting. While electromagnetic patents do exist,13,14) thcse do not provide either measurementor calculations concerning the melt velocities. The most important flnding of the present paper is that, for a rotating system, provided high enough fields are employed, it appears to be quite possible to agitate melt-solid suspensions at high enough shear and hence quite high rates to obtain good fluidity velocities. calculations presented in this melt The fields of I kG or the that of indicate order paper higher should perform adequately. While the calculations have been performed for a magnesium-boroncarbide slurry, qualitatively very similar considerations should bc expected for a broad range of metallic systems. Indeed, these calculations should be indicative of the behavior of both " rheocasting " and " compocasting " systems. The preliminary nature of this work must be stressed at this stage. Areas where further research is needed, and is indeed pursued by the present research team, include the following: (1 The calculations were presented for a specific, predetermined field. The actual coil design to bring field in the absence of excessive Joule about such a is practical issue. critical Nevergeneration heat a theless, the computed results presented here should provide adequate incentive for developing such coil designs. (2) The calculations were presented using one relating shear stress speciflc constitutive relationship It is uncertain slurry. strain in the melt-solid to is universally relationship actual the used whether nonetheless, the general shear thinning applicable; behavior of these systems was thought to be well represented, at least in a semi-quantitative sense. (3) In the calculations, the electromagnetic force field long cylinder. was calculated fbr an infinitely The behavior of finite systems will be somewhatdifferent, although the general nature of the findings is unlikely to be affected by such a refinement. (4) In the calculations we assumeda non-deformClearly, unless some mechanical able free surface. restraint is placed on the top free surface, significant

15

IZ.5

10

O
fs

\~

"

7.5
5,0 2.5

O
fs

08 04

f.

o.o

O~5

O5

O75 BO(KG)
velocity

1~5
field

Fig.

7.

azimuthal Maximum

with

solid fraction

with magnetic as a parameter.

15

O
Bo = 1.25KG

12.5

10

O
fs

= 08

\~
~

~,,

7.5
fD

5,0 2.5

= 04

0,0

O~

O4

0.6

O8

r/R

Fig.

8(a).

azimuthal velocity Radial variation of maximum with solid fraction as a parameter.

15.0
Bo =

125KG

l~.5
fs

10.0

OS

\s
~

~,,

7.5
5,0

fs

= O4

Indeed, deformation will occur due to melt rotation. this deformation, if not prevented, could cause sigpractical operating problems. However, in nificant

25
0.0

examining the computed results, there were no dramatic differences in the behavior between the top and

0~

04

Fig. 8(b).

Axial variation of with solid fraction as a parameter.

z/H azimuthal maximum

06

08

1
velocity

T~his would indicate the botto_m part of the vessel. that the results would be representative of a system with a solid cover. Notwithstanding these caveats, the results seemto with properly indicate that electromagnetic stirring, designed stirrers, could play an important role in both rheocasting and compocasting.

velocity fields in electroma~netically stirred melt-solid suspensions. In this earlier work, again, relatively low fields were considered, in conjunction with a vertical motion in the crucible, and again, quite low

Nomenclature
Bo: Amplitude of magnetic field
f*:

Solid fraction
force per unit

melt

velocities.

Indeed, the preliminary

conclusions

FE: Electromagnetic

volume
467.

ISIJ International,

Vol.

29 (1989), No. 6

F., Fa, F.

ofF. Components

in

r,

6and z directions,
l)

REFERENCES
R. Alberny and J. P. Birat:
Steel, Biarritz,

respectively

in

GK : Rate of generation H: Height of' melt

K:
m, n: ship

Continuous Casting of
Metall.
Trans. B,

of turbulence energy
2)

(1976),

I 16.
7B

Turbulent kinetic energy Empirical coefficients in power-law relationStatic pressure

E. D. Tarapore (1976), 343.

and J. W. Evans:

3) 4)
5)

(1978), 87. H. S. Marr: IronSteellnt.,51 K. H. Tacke and K. Schwerdtfeger: Stahl

7.

Eisen,

99 (1~79),

P:
r:

Radial coordinate

S.

Kolberg:

J., ASEA

53 (1980), 89.

R:
S:
u:

Radius of container

6)

Termin

source of e due to swirl Velocity componentin z-direction

u~:
v:

Maximum u
Maximum
z~'

7)

8)

w: w~:
z:

Velocity componentin r-direction Velocity componentin e-direction

9) 10)
ll) 12) 13) 14)

K. H. Spitzer, K. H. Tacke and K. Schwerdtfeger: Proc. Symp. Metallurgical Applications of Magnetohydrodynamics, Trinity College, Cambridge, (1982). H. K. Moffat: ZAMM, 58 (1978), T65-T71. " Electromagnetic Stirrer for Use in a K. Fujiwara et al.: Continuous Steel Casting Apparatus ", U.S. Patent No.
4,406,321, Sept. 27, 1983.
J. P. Birat

andJ. Chone : Ironmaking

Steelmaking,

10 (1983),
Iron

A:

e: e~
.

Axial coordinate Rate of deformation tensor Rate of dissipation of turbulence energy

269.

R. D. Matthewson, L.J. Long and D. T. Hackworth: Steel Eng., 63 (1986), No. 9, 36.

Maximum 5

6:
,i::

Azimuthal coordinate
Electrical

conductivity

M. Dubke and K. Schwerdtfeger: Metall. 17B (1986), 119. M. C. Flemings and R. Mehrabian: Trans. Am. Found.
K. H.
Spitzer,

Trans. B,

: pt : P*ff : po :
;L

Molecular Turbulent

(rK,

(T.

viscosity viscosity Effective viscosity Permittivity of free space Prandtl numbersfor and e, respectively

t:
a'

Shear stress tensor

: Angular velocity of

15)

81 (1973), 81. A. Danzig: " Process and Apparatus for Casting Metals Patent No. 8005620, (1980). and Alloys ", J. A. Danzig: " Process and Apparatus Having Improved Efficiency for Produc,ing a Semi-Solid Slurry ", European Patent No. 82105446.7, (1982). D. Bradshaw: " Effects of Streamline Curvature on TurSoc.,

J.

UK

electric

current
16)
17)

bulent Flow ", AGARDograph No. 169, (1973). P. A. Joly: Ph.D. Thesi_s to Department of Materials Science, Massachusetts Institute of Technology, (1979). D. B. Spalding: Mathcmatics and Computers in Simulation,

Acknowledgement This work was sponsored by the National Aeronautics and Space Administration under Research Grant #NAG~3-808.

XIII,

North Holland, Holland, (1981), 267.

468