Macroscopic equations for a binary gas mixture

Tatiana G. A. Graur* and Jean-Claude Lengrand^

G. Elizarova*, Irina A.

** Institute
Institute for
for Mathematical Modeling 1}, Russian Acad. of
Mathematical Modeling of Sci.,
Sei., Moscow, Russia
tLaboratoire CNRS, Orleans,
' Laboratoire d'Aerothermique du C'NRS, Orleans, France

Abstract. The derivation of moment equations

equations for
for the description of a flowing binary
binary non-reacting
non-reacting gas
gas
mixture is presented.
presented. The equations
equations obtained
obtained are based
based on a kinetic model in a relaxation
relaxation form and can
be regarded
regarded as the extension of the quasigasdynamic equations
equations (QGD)
(QGD) studied before. As an an example of
implementation the results
results for the shock wave structure in an argon- xenon mixture and for the binary
argon-xenon
diffusion process of argon
diffusion argon and helium are presented
presented in comparison with the DSMC results.
results.

I INTRODUCTION

There are two groups of models for calculating gas-mixture flows. The first one consists of kinetic kinetic models,
models, i.e.
i.e.
models based on direct numerical simulation methods (DSMC) or on the solution of the Boltzmann equation
(e.g. [1]).
[1]). The other group consists of systems of macroscopicmacroscopic equations which which are
are derived
derived on
on the basis of
Navier-Stokes equations, in general by by a phenomenological
phenomenological wayway (e.g.
(e.g. [2]
[2]- [4]).
[4]).
In this paper wewe propose a new macroscopic model to describe flows of a binary non-reacting gas mixture. The
model is a two-fluid approximation that consists of a system of equations equations for
for density,
density, momentum and energy
energy
of each component.
component. The system of macroscopic
macroscopic equations (named QGDM) QGDM) is is based on
on the kinetic
kinetic equation in
its relaxation approximation
approximation and is is a natural generalization
generalization of
of the
the recently
recently proposed quasigasdynamic (QGD) (QGD)
equations (e.g.
(e.g. [5]
[5] -- [7])
[7]) for
for aa gas
gas mixture.
mixture. Note
Note that
that previously
previously the
the QGD
QGD equations were were generalized
generalized for
for gas
flows in translational [8] [8] and rotational [9][9] non-equilibrium.
non-equilibrium.

II KINETIC MODEL
MODEL AND MACROSCOPIC
MACROSCOPIC EQUATIONS

We consider a mixture consisting of gas aa and gas b.b. Each gas is characterized
characterized by its density /?,-, temperature
Ti velocity «,,
T{ and macroscopic velocity {?,-, where
where i = a,b.
a, b. We introduce the gas constant Ri = Ar/m,-,
k/rrii, where
where k is the
Boltzmann constant, m,- is the
m, is the mass
mass of
of the
the molecule.
molecule. According
According to [10]
[10] the kinetical model for the mixture can
be expressed as

+ (£V)/a =
= vaa(F - ffaa) +
(Faa - + pvab(F - ffaa),
(Faa - (1)
(1)
at
dt

dh
2jj- + (£V)/6 =
= vbb(F
(Fbb -- /f6b)) +
+ vba(F - fA),
{Fbb - b), (2)
(2)
dt

where f faa and f /6b are the distribution functions

functions for species,
species, £ is the molecule velocity,
velocity, vaa
aa and v ^55
bb are the
frequencies of self-collisions, vz/ab a6 is the frequency of collisions
collisions_of
of aajnolecules with bb molecules,
molecules with molecules, vz/6 is the frequency
baa is
of collisions
collisions of b6 molecules
molecules^ with
with aa molecules.
molecules. F
Fa, FF&b and
and FFaa, FF&b are
are Maxwellian
Maxwellian distribution
distribution functions
functions based
based
on (u
(«aa,T
,T a), (u
(«6,J&), («aa,T
b,Tb), (ü ,T a),
), (u
(ubjfb). (overlined symbols)
b,Tb). The so-called free parameters (overlined symbols) can be be expressed
expressed in
in
terms of parameters m,u,T m, {?, T ofofeach
eachgas
gas

i) This work was sponsored by the Russian

Russian Foundation
Foundation for Basic Research, project
project no.98-01-00155
no.98-01-00155

CP585, Rarefied 22"ndd International

Rarefied Gas Dynamics: 22 International Symposium, edited by T. J. Bartel and M. A. Gallis
2001 American Institute of Physics 0-7354-0025-3/01/$18.00
© 2001
297
 REPORT DOCUMENTATION PAGE Form Approved OMB No.
0704-0188
Public reporting burder for this collection of information is estibated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing
and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burder to Department of Defense, Washington
Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of
law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (FROM - TO)
09-07-2000 Conference Proceedings 09-07-2000 to 14-07-2000
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Macroscopic equations for a binary gas mixture 5b. GRANT NUMBER
Unclassified 5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
Elizarova, Tatiana G. ; 5e. TASK NUMBER
Graur, Irina A. ; 5f. WORK UNIT NUMBER
Lengrand, Jean-Claude ;
7. PERFORMING ORGANIZATION NAME AND ADDRESS 8. PERFORMING ORGANIZATION REPORT
Institute for Mathematical Modeling NUMBER
Russian Acad. of Sci.
Moscow, Russiaxxxxx
9. SPONSORING/MONITORING AGENCY NAME AND ADDRESS 10. SPONSOR/MONITOR'S ACRONYM(S)
AOARD 11. SPONSOR/MONITOR'S REPORT
Unit 45002 NUMBER(S)
APO AP, xx96337-5002
12. DISTRIBUTION/AVAILABILITY STATEMENT
APUBLIC RELEASE
,
13. SUPPLEMENTARY NOTES
See Also ADM001341, Rarefied Gas Dynamics (RGD) 22nd International Symposium held in Sydney, Australia, 9-14 July 2000.
14. ABSTRACT
The derivation of moment equations for the description of a flowing binary non-reacting gas mixture is presented. The equations obtained are
based on a kinetic model in a relaxation form and can be regarded as the extension of the quasigasdynamic equations (QGD) studied before.
As an example of implementation the results for the shock wave structure in an argon- xenon mixture and for the binary diffusion process of
argon and helium are presented in comparison with the DSMC results.
15. SUBJECT TERMS
16. SECURITY CLASSIFICATION OF: 17. LIMITATION 18. 19. NAME OF RESPONSIBLE PERSON
OF ABSTRACT NUMBER Fenster, Lynn
Public Release OF PAGES lfenster@dtic.mil
8
a. REPORT b. ABSTRACT c. THIS PAGE 19b. TELEPHONE NUMBER
Unclassified Unclassified Unclassified International Area Code
Area Code Telephone Number
703767-9007
DSN
427-9007
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std Z39.18
 — — maua + mbub
uuaa==uubb == ■ ,
maa + + m
^6b
— 2mamb (/ mb ,^ _ ,2 \
Taa =Taa + - ■ — [T - Taa + —5 (u
I T5b - («b6 -- uaa))2 ) ,
u
(maa + mby V ok
6& /
—
_ 2m
2maqm mb5 f ma _ 2\
T =T
J6b = +
l bb + ■ — [T - Tb + — (M6 - ua) ■ (3)
(maa + mby V a ok I
We suppose that the distribution functions can be approximated by a gradient expansions as in [8]
[8]

/,-
fi = F -- T(fr)Fi
= Fi T((V)Fi (4)
(4)
where r is identified with ß/p, ///p, ß
// is
is the viscosity and p the pressure of the mixture. These expressions are
are
substituted for ffaa and
and ffbb into the convective terms of Eqs.l—
Eqs.1-22 to yield an approximation of these equations
Of
= vaa(F
-jf + (£V)Fa - (HV)T(tiV)Fa = - ffaa)) +
(Faa - + vab(F - ffaa),),
(Faa - (5)
(5)

^ + (fV)F6 - ((V)r((V)Fb = vw,(F

vbb(Fbb -- ffbb]) +
+ vba(F
(Fbb -- ffbb).). (6)
(6)

Equations 55 and 66 are integrated over the

the velocity space after
after multiplications by 1,
1, £ and £ /2,
/2, successively
successively
and result in a system (QGDM) of macroscopic equations
equations for
for the binary mixture

Pa + ViPa U\ - ViT (VjPa uiu{ + V*>a) = 0, (7)

dt

d
■T^Pa uka + ViPa u\uka + VkPa = V,- r (VjPa
a = Vf T (Vj/9
M»^ + V* Pa«Ju\++V*
a <<«a + V pa
k
VPa <)u*a+
Pa ) +V*
VTfe
T V;pPa
V," u\ + S^ ,
a < + SJf ,
(8)
(8)

f) 1
— Ea + Viui(Ea +Pa) = V,-
V,-rr{Vj{E
(Vj(Sa a++2p2p
a)<<
a)«'a«i++-V« afc«5pa) ++
^Vuafcujpa)

-^-rV,-r^Vp0+Pra-1^2-V,-rpaV^ + 5fI (9)

la -- 1
la Pa
Pa la -I
7a - Pa/^
where the volumic energy writes
2
Eaa =
= ({pauD/^+Pa/ila
Pau a)/<2+pa/(7a -- 1)
1) with
with pPaa =
=PaPaRR
aT a.a.
aT (10)
(10)
Similar equations hold for gas b.
b. The Prandtl number Pr
Pr and specific heat ratio 7 do not appear in the above-
described treatment and have been introduced artificially for generalization purpose to make these equations
coincide with Navier-Stokes (NS) equations in the case of a one-species gas and vanishing Knudsen numbers
(r-X))[7].
(r->0)[7].
The exchange terms S result from the integration of the right-hand sides of the equations 5, 6 and are
recognized as
S" = Vabpa($a
S% = vabpa(ua ~
-Uuaa),
], Sb =
S% =vVba pb{ub-u
bapb($b ~ Ubb]),, S® =Vab{E
Sf = Vab(Eaa-
- E
Eaa),
), S
Sfb =
= vVba
ba{E
(Ebb -
- E
Ebb),
), (11)
(11)

where Eaa and Eb

Eb have expressions similar to E
Eaa in Eq.10.
Suaa +S%
S" + 5fc« =
= 0,
0, SEa ++5fSf==00
5f (12)
(12)
which is consistent with momentum and energy conservation.
conservation.
Thus the binary mixture has been described by a self-contained set of two-fluid equations which allows for
different velocities and temperatures of the components.
different
The mixture parameters (no
(no subscript) relate to the species parameters:
n = na + nb, p = pa+Pb, P = Pa+Pb, b, u = (paau uaa + pbbub)/p,
)/p,
T = (n
(naaT
Taa +n
+ nbbT
Tbb)/n,
)/n, m
m = (m
(maan
na + m
mbbnnbb)/n,
)/n, pp = pRT,
pRT,
R=
R = (paRaRa ++pPbRb)/P
bRb)/p = k/m. (13)

298
 Ill
Ill DETERMINATION OF
OF COLLISION FREQUENCIES
To close
close the system of QGDM equations,
equations, it is is necessary
necessary to
to estimate
estimate the cross-collision frequencies vabb and
Vba and the relaxation time r.
i/ba r.
Relations between frequencies of collisions of aa molecules
molecules with
with each
each other (self-collisions) and with
with bb molecules
molecules
(cross-collisions), according to [10], can be calculated in the following way:

dab\ + mbbnnaa
maa+m
———— —, (14)
\da J maamb nnbb
V ^rn

where daa is
is the effective molecular diameter for gas a, a, ddab is the effective diameter which can be
ab is be determined,
determined,
for example, according to [1], p. 16, as
p.16, as ddab 0.5(<iaa + dd&).
ab = 0.5(<i In turn,
b). In turn, the
the collision frequency vz/aa can be
be related
related to
to
with the gas viscosity in the approximation of the VHS and and VSS models for particle interactions ( [1], p.90 ): ):

a aa
Pa 0/
P r\f
va =—\l{UJ \N
a,aa), pa = paref(/I — w«
)\u>n°,
Pa J-aref
,, rv
0 / ^^ 5(aq aa + l)(a
y va qa + 2)
where Q(u;a,a
Q,(uj a ,aaa) = —
—f - — —— ——-^-.
—-. (15)
15
aaa(J
(7- 2u;aa)(5
- 2w )(5 - 2uj
2ujaa))

For further calculations

calculations aa = 11 is
is used,
used, that conforms to the VHS
VHS model (( [1], p.41). In this case we
we write
Jfi(w
J ( wa,a , 1)
l ) = fi(w
n(waa).
).
The total number of collisions between molecules of gases gases aa and bb should be balanced, i.e.
i.e. relation naa vab
ab =
nt»
nb v z/6
baa must be satisfied.
To estimate the relaxation time r = p/p
To estimate p/p we
we determine
determine the binary-mixture viscosity using the Wilke formula
[3]:

(■\<n Pb Pb Ma\l , (, , , Pa M.b^ _1

P = Pa 1 + Crafc r— + ßb 1 + r
Gfca-
Pa Mb) \ Pb Ma
2
N
1 ' '
\JPa/Pb\/MblM a
/
where Gaab
G b = ^——— .. ,, (16)
y
2^2(1 + Ma/Mb) '

where M. Mb are the molar masses of gases a and b

Maa and Mb b respectively.
Note, that the collision frequencies and the binary-mixture viscosity are parameters "external" to the QGDM
model and could be determined by other estimations.

IV SHOCK WAVE STRUCTURE IN A HELIUM -- XENON MIXTURE

As a first example of using the QGDM equations we we considered the problem of a stationary shock wave
structure in a mixture of helium [He
(He is gas a) and xenon (Xe(Xe is gas b).
6). Density profiles for these gases,
measured with the use of an electron gun and a laser interferometer, can be found
found in [11]. Measurements were
performed for the following variants:

VI -- 98,
• variant VI 98, 5%
5% He and
and 1.5%
1.5% Xe
V2 -- 97%
• variant V2 97% He and 3%
3% Xe

variant V3
• variant V3 -- 94%
94% He and
and 6%
6% Xe
V4 -- 91%
• variant V4 91% He and
and 9%
9% Xe
For variant V2,
V2, there is a calculation by the DSMC method [1], the the results
results of
of which
which can
can be
be considered
considered as
as aa
reference. It coincides practically with the experimental data.
The system of equations (7)(7) - (9)
(9) is solved in non-dimensional variables taking as dimensional scales the
following characteristics of gas aa in the upstream flow: ppar areel
j is the density, aaare
areff = \/~f
^jaRaT
aRaTaref is the sound
aref is
velocity at temperature Tare j, XAare
aref, a r efj is
is the mean free path, that isis computed as as inin [1]:
[1]:

299
 1 _ _ _—

He
^^ /
0.75 yy / /
0) // / /
3 / /
g / / Xe
■a
tu
/ /
So.5 / /
<o / /
E / /
o
z
/ /
/i / /
0.25 / /
1 DSMC
i
QGDM

0 .w _^_ r" , 1 ,
-10 10 20 30 10
x/Lambda. x/Lambdaa

FIGURE 1. Density (a) and temperature(b) profiles in a He-Xe mixture

Ap 1
=
A = (17)
pVRT V27rß(w)

Then relations between the dimensional and dimensionless parameters have the following forms (all param-
eters of gas bb are scaled by parameters of gas a):

P = PParef , a = äaarej, u = üaarej, p = pparejalref, m = fhparej A^re/ ,

t = t n = n
T = 1T——
1 — ^ = 1TT
— are/,•>
1 aref x = xXare/ ■>
& — %^aref , ~& — ~& ~ ~>
i ^ — Ts
^ 3 •
A\
la Ra
Ta^a laref
"are/ aref
*aref

QGDM equations do not change their forms after the process of scaling. The relations between the parameters
of the gases (link equations) write as

-
— \
IJ±^±rr
V la
ry J - b j
Ra
la Ra
J-a —
laPa
HP - laj)a
~
Pa
; n laPb Ra
Pb Rb'
Rb
Pb

f^bref ( -L-J-aref
are/
7. — rpU,
Pa — ±a Pb Tt
f^aref \ Tbref
-*- bref

here ftbref
pbref and TbTbre
reff are the viscosity coefficient
coefficient and the corresponding temperature of gas b, 6, used in the
viscosity-law (15).
Table VI presents the parameters of the mixture before the shock shock wave,
wave, chosen
chosen in
in accordance
accordance with
with the
the
experimental data from [11] and with the calculation from from [1]. The Prandtl number for the
the gases
gases is
is constant
and equal to Pr = 2/3.
The boundary conditions on the right and left boundaries were taken from the the Rankine-Hugoniot condi-
tions for a stationary shock wave in a gas mixture.
mixture. The variables on the right of the discontinuity are computed
as follows
((7+ ) M a22
7 + l l)Ma 2 2
2
7 Ma -7+I
2jMa -7+1
P2 = Pi P2 = Pi-
2+ (7- l)Ma2' L
7 + 1
7+1 '
2
2+ (7- \)Ma2
_ l)Ma
«2 = «r (18)
(7 + l)Ma2 :

300
 1
s -~^*^^s
~~ -^—'^—
' s^ / y 0.5 -
T DSMC

0.75
a>
CD /7
ZJ
// // P 0.25
> // / /
"a '/ / / 5
NO.5 / //
~m if ■^

£ //
o
z //
//
0.25
//
// i DSMC
i 1 — - - QGDM
-y ll
;• Jl -0.25
0 _t=. f , 1
-10
-10 0 10 20 30 -10 0 10 20
20 30
x/Lambdaa
x/Lambda. x/Lambdaa
x/Lambda.

FIGURE 2. Mean temperature and density profiles (a) and diffusion

diffusion velocities (b)
(b) in a He-Xe mixture

where subscripts 11 and 22 refer to Rankine-Hugoniot conditions upstream (1) (1) and downstream
downstream (2)(2) of
of the
the shock
shock
wave. The component temperatures are found from the equations of state. state.
Assume that the temperatures and velocities of the components before and after
after the shock wave are equal, and
the mass-fraction
mass-fraction of the components after
after the transition through the shock
shock are
are unchanged
unchanged (n i/ni == n^/n^^
(naa\jn\ na2/ri2,
ribi/ni = ^62/^2)-
nbi/ni nb'ijn'i)- Thus on the basis of conditions (18), the parameters of each component of the mixture are
derived from the ratios:

Pa'i
Pa2 == m
^ ana2, Tal — Tbl — T\, Ual = «61 = «1

Pb2 = mbnb2, Ta2 = Tb2 = T'2, Ua2 = «62 = «2 (19)

are a discontinuity at point x = 0:

The initial conditions are 0:

< 00
at x < Pa =
Pa = Pal,
Pal, Pb=
Pb Pbi,
= Pbl, Taa = Tbb =
= Tl,
TI, uaa =
U = Uubb =
= Ml
wi.
at
at x
X >
> 0
0 Pa
/?a = Pa2,
pa2, Pb2,
P6 = Pb2,
Pb Taa =
T = T
Tbb = T22,, uaa =
W = U
ubb =
= «
u22 . (20)

To solve
solve the (7)(7) - (9)
(9) system, an explicit difference
difference scheme was applied where the steady-state solution was
obtained as the limit of a time-evolving
time-evolving process.
process. All
All spatial
spatial derivatives,
derivatives, including
including the convective terms,
terms, were
approximated by central differences (see, for instance, [12]).
-55
The problem was was solved using a uniform spatial grid with a convergence criterion ecPa 10~
pa = 10 .. When refining
the grid by a factor of 2, 2, the differences
differences between the computational results were extremely small, which allows
to conclude that grid convergence has been reached.
The profiles of gas-dynamic
gas-dynamic parameters
parameters (velocity,
(velocity, density,
density, temperature)
temperature) are given in a normalized form on on
the basis of
the of upstream and and downstream Rankine-Hugoniot conditions.
conditions. In this case, p —» —> (p —
— pi)/(p2
pi)/(p2 —— Pi)',
pi)I
for the
similarly for the temperature. For For the
the velocity u —> (u(u —
— 1/2)/(v>i
U2)/{u\ ~~— 1*2)
U'j).•
Let the computational results for variant V2 be considered in detail. In Figs. Figs. 1-3
1-3 the
the profiles
profiles of
of gas-dynamic
gas-dynamic
parameters at the shock-wave front front are shown in comparison with the the corresponding
corresponding results
results obtained
obtained inin [1]
[1] on
the basis of the DSMC method. The curves, corresponding to the DSMC calculations, are superimposed on
the QGDM data, so that the values of the mean density coincide at x = = 0.
In Fig.
Fig. 11 are the profiles of density and temperature of helium and xenon. In Fig.2a are the distributions
of the mean density and temperature for for the mixture. Like in the DSMC model, the mean temperature of the
mixture is very close to that of helium;
helium; and the temperature of xenon overshoots its final value by ~ 10% 10% and
the mean mixture temperature is close to the helium temperature.
In Fig.
Fig. 2b are presented the diffusion
diffusion velocities Ud
u^a and u^b,
w^d, reduced by the upstream flow velocity.
velocity.

Uda Udb = Ub (21)

301
301
 0.035 10

0.03

g
'o
ca
= 0.025
CD
E 1 5"
He
0.02

CD
X
0.015 DSMC
- - - - QGDM

0.01
-10 10 20 30 10
x/Lambdaa % of Xe

FIGURE 3. Diffusion
Diffusion velocities (a) and shock wave thickness (b) in a He-Xe mixture

where u is is the mixture velocity that is defined according to (13).

(13).
In Fig.
Fig. 3a the xenon concentration is presented. Within the shock wave, it falls to approximately halfhalf its
initial value.
The curves demonstrate that the QGDM model reflects at least qualitatively the main features of the the flow.
Variants VI,V3 and V4 V4 correspond to the conditions of density measurements by Walenta [11]. The compar-
ison of experimental results
results with those obtained by the present numerical work leads to the same conclusion.
On the basis of the calculations for variants VI - V4 the shock-wave thicknesses &He/^He
Sne/^He and
and Sxe/^Xe
Sxe/^Xe have
been plotted in Fig.3b against the concentration of Xe in the upstream mixture and compared with the results
of Xe
of [11].
[11]. In this case, the shock-wave thickness is calculated as

P2 -~ Pi
Pi
S6 = (22)
max(dp/dx)

The mean free path for each component is computed according to (17) for parameters of each gas component
before the shock wave. The experimental data are plotted as a solid line; the authors' results as a dashed line.
All the curves are represented in the form
form similar to [11]. For variant VI (with the smallest Xe concentration),
the experimental and computational results coincide practically. With increasing Xe concentration, the cal-
culated shock wave thickness is is larger than the experimental values. Nevertheless, the qualitative behavior is
reproduced by the calculations. This is also consistent with the well-known fact that the relative shock wave
thickness increases when the upstream Mach number decreases.
The shock wave thickness isis a very sensitive characteristic feature of the problem,
problem, and its calculation based
on moment equations for a single-component gas corresponds to the experimental data only in the case case of
of small
Mach numbers Ma Ma < 2.2.

V - HELIUM DIFFUSION
ARGON -

As a second example of application of the QGDM equations,

equations, the problem of helium and argon mass diffusion
diffusion
was studied for conditions that correspond to a computation by the DSMC method [1]. Let two reservoirs,
filled with the gases be located at
at a distance L = = 1m.
lm. He is gas a in the
the right
right reservoir,
reservoir, and
and Ar Ar is
is gas
gas 6b in
in
20 _3
the left
left reservoir. The number densities in the reservoirs are kept constant and equal to n = 1020
= 2.8 •• 10 m~3.The
m
gases in the reservoirs are assumed to have the same temperature T T = 273K and the same
same velocity
velocity equal
equal toto
zero.
Using these constants, the missing initial data can be obtained: helium density pa = = nm
nmaa = = 1.862 •
10_66kg/m
10~ kg/m3; sound velocity aaaa = \/jaRaTa = 971.9m/s;
971.9m/s; mean free path computed by formula (17) AAaa ==

302
 160

0.75

0.5
Q)

0.25

0.25 0.5 0.75

x(m)

FIGURE 4. Number densities and diffusion velocities in the Ar -- He diffusion problem.

problem.

2 5
10~2m;
1.479 •• 10~ m; argon density p^
pb =
= nm^
nrrib =
= 1.856 •• 10~ kg/m 3 ; sound velocity 05
10~5kg/m a& =
= -^jbRbTb —
= 307.81m/s; mean
free path computed by formula (17) is At A6 = 4.63 10~33m.
4.63 •• 10~ m.
As in the previous section, computing was performed in dimensionless variables with all quantities normalized
by the parameters of gas a -- helium in the reservoir.
A one-dimensional plane flow described by equations (7) (7) - (9)
(9) was considered.
considered. As boundary conditions the
following non-dimensional relations were used:
at the left-hand boundary (1) (1)

-10
10 -10 dua dub
pa =
Pa = 1. - HT ,
1.-10" Pb = 10" Ta = Th 0,
da da

(2)
at the right-hand boundary (2)

dua® dub
Pb l.-io-10, Pa = w-10, = Tlbb = 1.,
Taa=T i
1. -£— = -£—b = n0.
0.
da da

That is,
is, we 10 10 of the other gas exists. At initial time, the
we assumed that in each reservoir a fraction ~ 10~
density of the components between the reservoirs is assumed to change linearly ::

pa (X = L) - pa (X = 0)
Pa{X = L) Pa{X = ())
pa(x)
(X)= -———
= —— ^————— '—— T ^————— '- X + pa(X== 0),
,),

i N pb(x =
Pb(x = L)
L) -- Pb(x
pb(x =
= 0)
0)
pPb(x)
b(x) == — -———'— ——-———Lxx ++ ppb(x
b(x =
= 0).
0).
Ju

We used the same numerical algorithm as in the previous section when solving the QGDM equations. The
problem was solved using a uniform space grid consisting of 339 339 points
points with
with spatial
spatial grid
grid step
step hh = 0.2
0.2 ,, which
corresponded to 0.2A
0.2Aaa and 0.64At.
0.64A6.
The number densities and diffusion
diffusion velocities of both gases
gases are
are plotted in in Fig.
Fig. 44 against the position between
the reservoirs. In both figures the comparison with DSMC results [1] is given.
[1] is given. Again
Again the
the present
present results
results agree
agree
at least qualitatively with the reference results. The point of equal concentration point is is shifted from the
middle of the domain to the left, closer to the
the reservoir
reservoir containing
containing the
the heavier
heavier gas.
gas. The
The diffusion
diffusion velocity
velocity ofof
helium is larger than that of argon.
argon. The diffusion velocity of helium
helium exhibits
exhibits aa minimum
minimum in in the middle of the
computational domain.

303
 TABLE 1. Dimensional parameters of mixture
mixture components
components for variants
variants VI —VA
— V4

VI V2 V3
V3 V4
He Xe He Xe He Xe He Xe
p(kg/mdd)) -• 1055 5.15 2.57
2.57 5.16 2.22
2.22 4.91
4.91 10.3
10.3 4.57 14.8
14.8
p(Pa)
p(Pa) 33.14 0.51
0.51 33.21
33.21 1.02
1.02 31.62 2.02
2.02 29.42 2.91
2.91
T(K)
T(K) 310.
w(m/s)
«(m/s) 3076.76 2882.6 2672.8 2530.3
2530.3
Ma
Ma 2.97 17.01
17.01 2.78 15.93
15.93 2.58 14.78
14.78 2.44 13.99
13.99

VI CONCLUSION
The macroscopic system of the QGDM equations is is constructed on
on the
the basis
basis of the
the kinetic
kinetic equation system
in the relaxation approximation to describe the non-reacting-gas mixture flow. Contrary to some widely known
models, the QGDM system consists of the equations for density, momentum and energy of each component,
i.e. it is
is a two-fluid approximation,
approximation, that makes it possible to describe in detail the behavior of each component.
component.
The momentum and energy equations
equations include exchange terms,
terms, that allow for
for the appropriate exchanges
between the gas components. To calculate these terms terms the cross-collision frequencies must be estimated.
estimated. The
viscosity of the mixture must also be estimated. The QGDM model includes diffusion diffusion processes but does not
require coefficients of thermo-,
thermo-, baro-
baro- and self- diffusion, which are
are included
included in
in the
the Navier-Stokes
Navier-Stokes models,
models, and
the determination of which is a separate task.
The numerical algorithms developed on the basis of the QGDM model appear to be more stable than similar
algorithms based on the conventional conservation equations.
Thus, the QGDM equations are worth being studied further for future other applications.

REFERENCES
1. Bird G.A.
1. G.A. Molecular gasgas dynamics
dynamics and the direct simulation of gas flows, Clarendon Press, Press, Oxford, 1994.
1994.
M.H. Internal gas mixture flows. Moscow, Nauka, 1989.
2. Lapin Yu.V., Streleth M.H. 1989. (in Russian)
3. Golovachev Yu.P.
3. Yu.P.Numerical
Numericalsimulation
simulationofofthe
theviscous
viscousgas
gasflow
flowininaashock
shocklayer.
layer.Moscow,
Moscow,Nauka,
Nauka,"Fizmatlit",
"Fizmatlit", 1996.
1996.
(in Russian)
4. Zhdanov V.M.
V.M. Transport
Transportandandrelaxation
relaxationprocesses
processesininmulticomponent
multicomponent plasma.
plasma.Mir,
Mir,Energoizdat,
Energoizdat, 1982.
1982.(in
(inRussian)
Russian)
5. Elizarova T.G.,
5. T.G., Chetverushkin B.N. B.N. Kinetic algorithms for calculating gasdynamic flows, J.Comput. Mathem. and
Mathem. Phys. 25, No5, 164 164 (1985).
(1985).
6. Elizarova T.G.,
6. T.G., Graur
Graur I.A.,
I.A., Lengrand J.C.,
J.C., Chpoun A. A. Rarefied gas flow simulation based on on quasigasdynamic
quasigasdynamic
equations. AIAA Journal. 1995.1995. V.33.
V.33. NN 12.
12. P.2316
P.2316 - 2324.
2324.
7. Sheretov Yu.V., Quasi hydrodynamic equations as a model for viscous compressible heat conductive flows, in book:
7. book:
Implementation of Functional Analysis in the Theory Theory of
of Approaches, Tver University, 127 (1997) (in Russian).
Russian).
8. Elizarova T.G.,
8. T.G., Lengrand J.C.,
J.C., Graur
Graur I.A. functions and
LA. Gradient expansions for distribution functions and derivation of moment
equations, 21th
21th Intern. Symp. on Rarefied Gas Dynamics, Marseille, France, July 26-30, 1998, 1998, Ed.R.Brun et al..
al..
Cepadues, Toulouse, France, 1999,1999, Vol.1, pp.119-126.
9. Chirokov I.A.,
9. LA.,Elizarova
ElizarovaT.G.,
T.G.,Lengrand
LengrandJ.C.
J.C.Numerical
Numericalstudy
studyofofshock
shockwave
wavestructure
structurebased
basedononquasigasdynamic
quasigasdynamic
equations with rotational nonequilibrium, 21th 21th Intern. Symp. on Rarefied Gas Dynamics, Marseille, France, July
26-30, 1998,
1998, Ed.R.Brun et al..
al. Cepadues, Toulouse, France, 1999, 1999, Vol.1, pp.175-182.
10. Wu
10. Wu Y.,
Y., Lee
Lee C.H.
C.H. Kinetic theory of shock tube problems for for binary mixtures,
mixtures, J. Phys.
Phys. of Fluids. 14, NN 2,
2, 313
313 (1971).
(1971).
11. Gmurczyk
11. Gmurczyk A.S.,
A.S., Tarczynski
Tarczynski M.M. and
and Walenta
Walenta Z.A.Z.A.Shock
Shock wave
wave structure
structure in
in the
the binary
binary mixtures
mixtures ofof gases
gases with
with
disparate molecular masses, Rarefied gas dynamics, Ed. by R. Campargue, CEA, CEA, Paris, v.l,
v.l, 1979,
1979, pp. 333-
333- 341.
341.
12. Graur
12. Graur LA, T.G, Lengrand J.C.
LA, Elizarova T.G, J.C. (1997) Quasigasdynamic equations with multiple translational
translational tempera-
tures, Laboratoire d'Aerothermique du CNRS, Meudon (Fr), (Fr), R 97-1.
97-1.

304

View publication stats