marangoni convections in a horizontal porous layer ... - IJAMM
marangoni convections in a horizontal porous layer ... - IJAMM
marangoni convections in a horizontal porous layer ... - IJAMM
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
MARANGONI CONVECTIONS IN A HORIZONTAL POROUS LAYER<br />
SUPERPOSED BY A FLUID LAYER IN THE PRESENCE OF<br />
UNIFORM VERTICAL MAGNETIC FIELD AND SOLUTE<br />
R. T. Matoog 1 and Abdl-Fattah K. Bukhari 2<br />
1 Department of Mathematics. Faculty of teachers preparation. Makkah. Saudi Arabia<br />
Email: t ezat@yahoo.com<br />
1 Department of Mathematics. Fac. of Applied Sciences. Umm Al-Qura Univ. Makkah.<br />
Saudi Arabia<br />
Email: afb@uqu.edu.sa<br />
ABSTRACT<br />
Received 21 April 2008; accepted 18 May 2008<br />
A l<strong>in</strong>ear stability analysis applied to a system consist<strong>in</strong>g of a <strong>horizontal</strong> fluid <strong>layer</strong> overly<strong>in</strong>g a<br />
<strong>porous</strong> <strong>layer</strong> <strong>in</strong> the presence of uniform vertical magnetic field and solute on both <strong>layer</strong>s. Flow<br />
<strong>in</strong> <strong>porous</strong> medium is assumed to be governed by Darcy’s law, where as the flow <strong>in</strong> the fluid <strong>layer</strong><br />
is governed by Navier-Stokes equation. The Beavers-Joseph condition is applied at the <strong>in</strong>terface<br />
between the two <strong>layer</strong>s. Numerical solutions are obta<strong>in</strong>ed for stationary convection case us<strong>in</strong>g<br />
the method of expansion of Chebyshev polynomials. It is found that the spectral method has a<br />
strong ability to solve the multi-<strong>layer</strong>ed problem, the depth ratio has a strong effect and that the<br />
magnetic field has effect <strong>in</strong> this model.<br />
Keywords: Navier-Stokes equation, Darcy’s law, Chebyshev polynomials, Magnetic equations,<br />
Maragoni Convections, Magnetic field and solute, Non-dimensionalization.<br />
1 INTRODUCTION<br />
Thermal <strong>in</strong>stability theory has attracted considerable <strong>in</strong>terest and has been recognized as a problem<br />
of fundamental importance <strong>in</strong> many fields of fluid dynamics. The earliest experiments to<br />
demonstrate the onset of thermal <strong>in</strong>stability <strong>in</strong> fluids, are those of Benard (Benard 1900; Benard<br />
1901). Benard worked with very th<strong>in</strong> <strong>layer</strong>s of an <strong>in</strong>compressible viscous fluid, stand<strong>in</strong>g on leveled<br />
metallic plate ma<strong>in</strong>ta<strong>in</strong>ed at a constant temperature. The upper surface was usually free and<br />
be<strong>in</strong>g <strong>in</strong> contact with the air was at a lower temperature. In his experiments Benard deduced that<br />
a certa<strong>in</strong> critical adverse temperature gradient must be exceeded before <strong>in</strong>stability can set <strong>in</strong>.<br />
Rayleigh (Rayleigh 1916) provided a theoretical basis for Benard’s experimental results. He<br />
showed that the numerical value of the non-dimensional parameter. Jeffrey’s (Jeffreys 1930)<br />
showed that the onset of thermal <strong>in</strong>stability criterion derived for <strong>in</strong>compressible fluids could be<br />
used for compressible fluids under some certa<strong>in</strong> conditions on the adverse temperature gradient.<br />
Benard convection <strong>in</strong> the context of magneto hydrodynamic fluid has been exam<strong>in</strong>ed by<br />
Thompson (Thompson 1951) and Chandrasekhar (Chandrasekhar 1952) with a l<strong>in</strong>ear constitutive<br />
relationship between the magnetic field and the magnetic <strong>in</strong>duction. Nield (Nield 1968)<br />
considered the onset of salt-f<strong>in</strong>ger convection <strong>in</strong> a <strong>porous</strong> <strong>layer</strong>. Taunton et al., (Taunton and<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 35<br />
Lightfoot 1972) considered the thermohal<strong>in</strong>e <strong>in</strong>stability and salt-f<strong>in</strong>gers <strong>in</strong> a <strong>porous</strong> medium and<br />
solved the boundary value problem. Sun (Sun 1973) was the first to consider such a problem,<br />
and he used a shoot<strong>in</strong>g method to solve the l<strong>in</strong>ear stability equations. The results show that the<br />
critical Raleigh number <strong>in</strong> the <strong>porous</strong> <strong>layer</strong> decreases cont<strong>in</strong>uously as the thickness of the fluid<br />
<strong>layer</strong> is <strong>in</strong>creased. Nield (Nield 1977) formulated the problem with surface-tension effects at<br />
a deformable upper surface <strong>in</strong>clude and obta<strong>in</strong>ed asymptotic solution for small wave numbers<br />
for a constant heat-flux boundary condition. Sun (Sun 1973) and Nield (Nield 1977) used<br />
Darcy’s law <strong>in</strong> formulat<strong>in</strong>g the equations for <strong>porous</strong> <strong>layer</strong>. In Darcy’s low of motion <strong>in</strong> <strong>porous</strong><br />
mediums, the Darcy resistance term took the place of the Navier-Stokes viscose term while<br />
<strong>in</strong> modified Darcy’s law (Br<strong>in</strong>kman model) that suggested by Br<strong>in</strong>kman (Br<strong>in</strong>kman 1947a;<br />
Br<strong>in</strong>kman 1947b) the Navier-Stokes viscous term still exists. Chandrasekhar (Chandrasekhar<br />
1981) considered typical problems <strong>in</strong> hydrodynamic and hydro magnetic stability <strong>in</strong> his treatise.<br />
Somerton and Catton (Somerton and Catton 1982) have different formulation of the problem<br />
from Sun (Sun 1973) and Nield (Nield 1977) by <strong>in</strong>clusion Br<strong>in</strong>kman term <strong>in</strong> Darcy equation<br />
for the <strong>porous</strong> <strong>layer</strong> and solved the problem us<strong>in</strong>g Galerk<strong>in</strong> method. Hilles (Hills et al. 1983)<br />
and Maples & Poirier (Maples and Poirier 1984) discussed the stability of the boundary value<br />
problem of a thermo dynamical consistent model, the directional solidification of molten alloys<br />
as a <strong>layer</strong> of <strong>porous</strong> material of variable permeability which is separated from its melt by a<br />
mush zone of dendrites. Glicksman (Glicksman et al. 1986) described the <strong>in</strong>teraction between<br />
the solidify<strong>in</strong>g alloy and its melt by a doubly diffusive model. Chen & Chen (Chen and Chen<br />
1988) considered the multi-<strong>layer</strong> problem when the above <strong>layer</strong> is heated and salted from above,<br />
and solution of problem is obta<strong>in</strong>ed us<strong>in</strong>g a shoot<strong>in</strong>g method. Abdullah (Abdullah 1991) discussed<br />
the Benard convection <strong>in</strong> a non-l<strong>in</strong>ear magnetic field under the <strong>in</strong>fluence of vertical and<br />
non-vertical magnetic field for different boundary conditions. L<strong>in</strong>dsay & Ogden (L<strong>in</strong>dsay and<br />
Ogden 1992) worked <strong>in</strong> the implementation of spectral methods resistant to the generation of<br />
spurious Eigenvalues. Lamb (Lamb 1994) used expansion of Chebyshev polynomials <strong>in</strong>vestigate<br />
an Eigenvalue problem aris<strong>in</strong>g from a model discuss<strong>in</strong>g the <strong>in</strong>stability of the earth’s core.<br />
Bukhari (Bukhari 1996) studied the effects of surface-tension <strong>in</strong> a <strong>layer</strong> of conduct<strong>in</strong>g fluid with<br />
imposed magnetic field and obta<strong>in</strong>ed results when deformable and non-deformable free surface<br />
and he solved by us<strong>in</strong>g Chebyshev spectral method, and he obta<strong>in</strong>ed some different results from<br />
that of Chen & Chen (Chen and Chen 1988). Straughan (Straughan 2001) studied the thermal<br />
convection <strong>in</strong> fluid <strong>layer</strong> overly<strong>in</strong>g a <strong>porous</strong> <strong>layer</strong> and he considered when the problem of lower<br />
<strong>layer</strong> heated from below and surface tension driven on the free top boundary of upper <strong>layer</strong>.<br />
Also, <strong>in</strong> (Straughan 2002) he dealt with the same problem, consider<strong>in</strong>g the ratio depth of the<br />
relative <strong>layer</strong> also, <strong>in</strong>vestigated the effect of the variation of relevant fluid and <strong>porous</strong> material<br />
properties. Bukhari (Bukhari 2003a) applied the l<strong>in</strong>ear stability analysis <strong>in</strong> the system consist<strong>in</strong>g<br />
of a <strong>horizontal</strong> fluid <strong>layer</strong> overly<strong>in</strong>g a <strong>layer</strong> of a <strong>porous</strong> medium affected a vertical magnetic<br />
field on both <strong>layer</strong>s. The same author (Bukhari 2003b) applied the spectral Chebyshev polynomial<br />
method to obta<strong>in</strong> numerically the solution of a multi-<strong>layer</strong> system consist<strong>in</strong>g of the f<strong>in</strong>ger<br />
convection onset <strong>in</strong> a fluid <strong>layer</strong> overly<strong>in</strong>g a <strong>porous</strong> <strong>layer</strong>, and he studied the boundary value<br />
problem when the thermal Rayleigh number and the critical salt Rayleigh number for <strong>porous</strong><br />
<strong>layer</strong> are <strong>in</strong>creas<strong>in</strong>g due to heat<strong>in</strong>g and salt<strong>in</strong>g the lower of <strong>porous</strong> <strong>layer</strong>.<br />
In this paper we studies convective <strong>in</strong>stabilities <strong>in</strong> a <strong>horizontal</strong> multi-<strong>layer</strong> problem <strong>in</strong> the presence<br />
of uniform vertical magnetic field and solute. i.e., we shall consider the onset of thermal<br />
convection <strong>in</strong> a <strong>horizontal</strong> <strong>porous</strong> <strong>layer</strong> superposed by a fluid <strong>layer</strong> affected by a vertical magnetic<br />
field and solute. The flow <strong>in</strong> the <strong>porous</strong> <strong>layer</strong> is assumed to be governed by Dray’s law.<br />
The l<strong>in</strong>ear stability equations will be solved us<strong>in</strong>g expansion of Chebyshev polynomials. This<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
36 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
method is better suited to the solution of hydrodynamic stability problems than expansions <strong>in</strong><br />
other sets of orthogonal polynomials. Lanczos (Lanczos 1938) is the first presented of this<br />
method. This method have used and developed to solve ord<strong>in</strong>ary differential equations by Fox<br />
(Fox 1962), Fox and Parker (Fox and Parker 1968), Orszag (Orszag 1971a; Orszag 1971b).<br />
Chaves & Ortiz (Chaves and Ortiz 1968) applied it to a second order Eigenvalue problem with<br />
polynomial coefficients. Hassanien & El.Hawary (Hassanien and El-Hawary 1990) studied<br />
Chebyshev solution of lam<strong>in</strong>ar boundary <strong>layer</strong> flow. Bukhari (Bukhari 1996) has used this<br />
method to solve multi<strong>layer</strong>s region. Hassanien et. al., (Hassanien et al. 1996) studied axisymmetric<br />
stagnation flow on a cyl<strong>in</strong>der us<strong>in</strong>g series expansions of Chebyshev polynomials.<br />
2 MATHEMATICAL FORMULATION<br />
Consider the problem of two <strong>horizontal</strong> <strong>layer</strong>s L1 and L2 such that the bottom of the <strong>layer</strong> L1<br />
touches the top of the <strong>layer</strong> L2. A right handed system of Cartesian Coord<strong>in</strong>ates (xi, i = 1, 2, 3)<br />
is chosen so that the <strong>in</strong>terface is the plan x3 = 0, the top boundary of L1 is x3 = df + F (t, xα),<br />
and the lower boundary of L2 is x3 = −dm (see Fig. (1)). Suppose that the upper <strong>layer</strong> L1 is<br />
filled with an <strong>in</strong>compressible thermally and electrically conduct<strong>in</strong>g viscous fluid consist<strong>in</strong>g of<br />
melted solute which flow <strong>in</strong> it and is governed by Navier-Stokes equations, where as the lower<br />
<strong>layer</strong> is occupied by a <strong>porous</strong> medium permeated by the fluid which flow <strong>in</strong> it and is governed by<br />
Darcy’s law and is subjected to a uniform vertical magnetic field. Gravity g ¯ acts <strong>in</strong> the negative<br />
direction of x3, the heated and solutal from below.<br />
Figure 1: Schematic diagram of the problem.<br />
Convection is driven by temperature dependence of the fluid density and solut<strong>in</strong>g, and damped<br />
by viscosity. The Oberbeck-Bouss<strong>in</strong>eq approximation is used as density of the fluid is constant<br />
every where except <strong>in</strong> the body force term where the density is l<strong>in</strong>early proportional to<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 37<br />
temperature and solute concentration.<br />
ρf = ρ0[1 − α(T − T0) + α ′ (S − S0)], (1)<br />
where T denotes the Kelv<strong>in</strong> temperature of the fluid, S is the solute concentration, ρ0 is density<br />
of the fluid at T0 and S0, α (constant) is the thermal coefficient of volume expansion of the fluid<br />
and α ′ (constant) is the solut<strong>in</strong>g coefficient of volume expansion of the fluid.<br />
Furthermore, <strong>in</strong>compressibility of the fluid and the non-existence of magnetic monopoles require<br />
that V and B are both solenoidal vectors. Hence<br />
div V = Vi,i = 0, div B = Bi,i = 0. (2)<br />
Suppose also that the magnetization <strong>in</strong> the fluid is directly proportional to the applied field that<br />
the fluid behaves like an Ohmic conductor so that the magnetic field H, magnetic <strong>in</strong>duction B,<br />
current density J and electric field E are connected by the constitutive relations<br />
B = µm H, (3)<br />
J = σ (E + V × B), (4)<br />
and the Maxwell equations<br />
curl E = −∂B<br />
, (5)<br />
∂t<br />
J = 1<br />
curl H, (6)<br />
4π<br />
where µm (constant) is the magnetic permeability, σ is the electrical conductivity and the displacement<br />
current has been neglected <strong>in</strong> the second of these Maxwell equations as is customary<br />
<strong>in</strong> situations when free charge is <strong>in</strong>stantaneously dispersed. On tak<strong>in</strong>g the curl of equation (4)<br />
and replac<strong>in</strong>g the electric field by the Maxwell relation (5), the magnetic field H is now readily<br />
seen to satisfy the partial differential equation<br />
η curlcurl H = − ∂H<br />
∂t<br />
+ curl(V × H), (7)<br />
1<br />
where η = 4π σ µm is the electrical resistivity. In addition, the constant nature of µm makes<br />
the magnetic field H a solenoidal vector. Equation (7) is now reworked us<strong>in</strong>g standard vector<br />
identities to yield <strong>in</strong> sequence<br />
∂H<br />
∂t + (V · ∇)H = (H · ∇)V + η ∇2 H, (8)<br />
equation (8) describes the temporal evolution of magnetic field. Moreover, the relations (3),<br />
(4), (5) and (6) can be used to recast the Lorentz force J × B <strong>in</strong>to<br />
J × B = µm � 1<br />
(H · ∇)H −<br />
4π<br />
2 ∇H2� ,<br />
lead<strong>in</strong>g to the notion that the Lorentz force is derived from a magnetic stress tensor σ (m)<br />
ji . In<br />
view of the fact that<br />
(J × B)i = σ (m)<br />
ji,j<br />
µm �<br />
= HjHi −<br />
4π<br />
1<br />
2 H2 �<br />
δij ,j<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.<br />
(9)
38 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
so the govern<strong>in</strong>g equations of the fluid <strong>layer</strong> are<br />
�∂V f<br />
ρ0<br />
∂t + V �<br />
f · ∇V f = −∇Pf + µ∇ 2 V f + ρf g +<br />
¯ µmf<br />
4π (Hf · ∇Hf), (10)<br />
�∂Tf<br />
(ρ cp)f<br />
∂t + V �<br />
f · ∇Tf = kf∇ 2 Tf, (11)<br />
∂Sf<br />
∂t + V f · ∇Sf = Mf∇ 2 Sf, (12)<br />
∂H f<br />
∂t = (H f · ∇)V f − (V f · ∇)H f + ηf∇ 2 H f, (13)<br />
and the govern<strong>in</strong>g equations of the <strong>porous</strong> medium <strong>layer</strong> are<br />
ρ0 ∂V m<br />
ϕ ∂t = −∇Pm − µ<br />
K V m + ρf g +<br />
¯ µmm<br />
4π (Hm · ∇Hm), (14)<br />
∂Tm<br />
(ρ c)m<br />
∂t + (ρ cp)f V m · ∇Tm = km∇ 2 Tm, (15)<br />
ϕ ∂Sm<br />
∂t + V m · ∇Sm = Mm∇ 2 Sm, (16)<br />
∂H m<br />
∂t = (H m · ∇)V m − (V m · ∇)H m + ηm∇ 2 H m, (17)<br />
where PfPf + µmf<br />
8π H2 f , PmPm + µmm<br />
medium <strong>layer</strong>s respectively; ηf =<br />
8π H2 m are the modified pressure of the fluid and the <strong>porous</strong><br />
1<br />
4π σ µmf , ηm =<br />
1<br />
4π σ µmm<br />
are the electrical resistivity of the<br />
fluid and the <strong>porous</strong> <strong>layer</strong> respectively; V f, V m are the solenoidal and seepage velocity respectively;<br />
Tf, Tm are the Kelv<strong>in</strong> temperature of the fluid and <strong>porous</strong> medium <strong>layer</strong> respectively;<br />
Sf, Sm are solute concentration of the fluid and <strong>porous</strong> medium <strong>layer</strong> respectively; Mf, Mm are<br />
the mass diffusivity of the fluid and <strong>porous</strong> medium <strong>layer</strong> respectively; µmf, µmm are magnetic<br />
permeability of fluid and <strong>porous</strong> <strong>layer</strong> respectively; kf, km are the thermal conductivity of fluid<br />
and <strong>porous</strong> <strong>layer</strong> respectively; µ is the dynamic viscosity; K is the permeability; ϕ is the porosity<br />
and (ρ cp)f, (ρ c)m are the heat and overall heat capacity per unit volume of the fluid and<br />
<strong>porous</strong> medium <strong>layer</strong>s at constant pressure. In fact<br />
(ρ c)m = ϕ (ρ cp)f + (1 − ϕ) (ρ cp)m,<br />
where (ρ cp)m is the heat capacity per unit volume of the <strong>porous</strong> substrate.<br />
3 BOUNDARY CONDITIONS<br />
For the upper boundary we shall suppose x3 = d + F (t, xα) where t is the time with unit<br />
normal n = niei directed from the viscous fluid <strong>in</strong>to the passive <strong>in</strong>viscid fluid. So, the boundary<br />
conditions come from follow<strong>in</strong>g sources.<br />
General radiation conditions the heat and mass transfer:<br />
T,<strong>in</strong>i + L(T − T∞) = 0, (18)<br />
S,<strong>in</strong>i + C(S − S∞) = 0, (19)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 39<br />
where L, C constants. Material surface particles of the fluid rema<strong>in</strong> on the surface at x3 =<br />
d + F (t, xα) and so<br />
dx3<br />
dt<br />
= ∂F<br />
∂t<br />
+ ∂F<br />
∂xα<br />
∂xα<br />
∂t ,<br />
this leads to the condition<br />
V3 − ∂F<br />
Vα =<br />
∂xα<br />
∂F<br />
. (20)<br />
∂t<br />
Magnetic condition s<strong>in</strong>ce the region x3 > df is electrically <strong>in</strong>sulat<strong>in</strong>g, then Bi = µmfHi is<br />
cont<strong>in</strong>uous across x3 = df and the magnetic field <strong>in</strong> x3 > df is irrotational, that is, derived<br />
from a potential function.<br />
Stress conditions which can be decomposed further <strong>in</strong>to the tangential and normal components<br />
respectively,<br />
τ(T, S) b α α = −P∞ − (P + µmf<br />
8π H2 f ) + 2 ρ0 νVi,j n<strong>in</strong>j + µmf<br />
8π (Hjnj) 2 , (21)<br />
τ,α = ρ0 ν(Vi,j + Vj,i) nj xi,α + µmf<br />
4π (Hjnj)(Hi xi,α), (22)<br />
where ν is k<strong>in</strong>ematics viscosity. At the <strong>in</strong>terface x3 = 0 we have<br />
Tf = Tm, Sf = Sm, kf<br />
−Pf + 2µ ∂ωf<br />
∂x3<br />
∂uf<br />
∂x3<br />
= αBJ<br />
√K (uf − um),<br />
∂Tf<br />
∂x3<br />
= km<br />
∂Tm<br />
∂x3<br />
, Mf<br />
= −Pm, Hf = Hm, kf<br />
∂νf<br />
∂x3<br />
∂Sf<br />
∂x3<br />
= αBJ<br />
√K (νf − νm),<br />
∂Hf<br />
∂x3<br />
= Mm<br />
= km<br />
∂Sm<br />
, ωf = ωm,<br />
∂x3<br />
∂Hm<br />
, (23)<br />
∂x3<br />
where ωf and ωm are the normal axial velocity components of the fluid <strong>in</strong> fluid <strong>layer</strong> and <strong>porous</strong><br />
medium <strong>layer</strong> respectively; uf, νf are the limit<strong>in</strong>g tangential component of the fluid velocity<br />
as the <strong>in</strong>terface is approached from the fluid <strong>layer</strong> L1, whereas um, νm are the same limit<strong>in</strong>g<br />
components of tangential fluid velocity as the <strong>in</strong>terface is approached from the <strong>porous</strong> <strong>layer</strong> L2;<br />
the last two boundary conditions <strong>in</strong> (23), discont<strong>in</strong>uities <strong>in</strong> shear velocity across the <strong>in</strong>terface<br />
are <strong>in</strong>herent, also they calls Beavers & Joseph (Beaver and Joseph 1967). αBJ is Beavers and<br />
Joseph constant.<br />
At the lower boundary x3 = −dm we have<br />
ωm(−dm) = 0, Tm(−dm) = Tl, Sm(−dm) = Sl, Bi = µmmHi be cont<strong>in</strong>uous. (24)<br />
At equilibrium case<br />
V f = 0, V m = 0, −∇Pf + ρf g ¯ = 0, −∇Pm + ρf g ¯ = 0,<br />
∇ 2 Tf = ∇ 2 Tm = 0, ∇ 2 Sf = ∇ 2 Sm = 0, F (t, xα), (25)<br />
∇ 2 H f = ∇ 2 H m = 0, H = (0, 0, H) : Hconstant.<br />
The boundary conditions <strong>in</strong> the exterior are<br />
Tf(df) = Tu, Tm(−dm) = Tl, Sf(df) = Su, Sm(−dm) = Sl, (26)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
40 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
and the <strong>in</strong>terface conditions<br />
Tf(0) = Tm(0) , Sf(0) = Sm(0),<br />
∂Tf(0)<br />
kf<br />
∂x3<br />
∂Tm(0)<br />
= km ,<br />
∂x3<br />
∂Sf(0)<br />
Mf<br />
∂x3<br />
=<br />
∂Sm(0)<br />
Mm , Pf(0) = Pm(0), Hf(0) = Hm(0),<br />
∂x3<br />
(27)<br />
∂Hf(0)<br />
kf<br />
∂x3<br />
=<br />
∂Hm(0)<br />
km ,<br />
∂x3<br />
are the equilibrium of temperature field, solute concentration, hydrostatic pressure and magnetic<br />
field <strong>in</strong> the fluid <strong>layer</strong> and <strong>porous</strong> medium <strong>layer</strong> respectively, where<br />
Tf = T0 − (To − Tu) x3<br />
, Sf = S0 − (So − Su) x3<br />
, Pf = Pf(x3),<br />
df<br />
. H f = (0, 0, H), 0 ≤ x3 ≤ df;<br />
Tm = T0 − (To − Tu) x3<br />
, Sm = S0 − (So − Su) x3<br />
, Pm = Pm(x3),<br />
dm<br />
. H m = (0, 0, H), −dm ≤ x3 ≤ 0; (28)<br />
where the <strong>in</strong>terfacial temperature and solute concentration are determ<strong>in</strong>ed by the cont<strong>in</strong>uity of<br />
heat flux and solute flux respectively and take the values<br />
T0 = kf dm Tu + km df Tl<br />
kf dm + km df<br />
4 PERTURBED EQUATIONS<br />
df<br />
dm<br />
, S0 = Mf dm Su + Mm df Sl<br />
.<br />
Mf dm + Mm df<br />
Suppose that the equilibrium solution be perturbed by follow<strong>in</strong>g l<strong>in</strong>ear perturbation quantities<br />
V f = 0 + ν f, Pf = Pf(x3) + pf, H f = (0, 0, H) + h f,<br />
Tf = T0 − (T0 − Tu) x3<br />
+ θf, Sf = S0 − (S0 − Su) x3<br />
+ sf,<br />
df<br />
V m = 0 + ν m, Pm = Pm(x3) + pm, H m = (0, 0, H) + h m, (29)<br />
Tm = T0 − (Tl − T0) x3<br />
dm<br />
df<br />
+ θm, Sm = S0 − (Sl − S0) x3<br />
dm<br />
+ sm.<br />
Then we may verify that the L<strong>in</strong>earized version of equations (10), (11), (12) and (13) are<br />
�∂ν f<br />
ρ0<br />
∂t + ν �<br />
f · ∇νf = −∇pf + µ∇ 2 νf + ρ0 α θf g<br />
¯<br />
(ρ cp)f<br />
∂sf<br />
∂t + ν f ·<br />
− ρ0 α′ sf g ¯ + µmf<br />
4π<br />
df<br />
e3<br />
� Hf<br />
∂h f<br />
∂x3<br />
�<br />
+ hf · ∇hf , (30)<br />
�<br />
∂θf<br />
∂t + νf · � ∇θf − (T0 − Tu)<br />
df<br />
�<br />
∇sf − (S0 − Su)<br />
�<br />
= Mf∇ 2 sf, (32)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.<br />
�<br />
e3<br />
�<br />
= kf∇ 2 θf, (31)
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 41<br />
∂hf ∂ν f<br />
= Hf<br />
∂t ∂x3<br />
+ (hf · ∇)νf − (νf · ∇)hf + ηf∇ 2 hf, (33)<br />
and equations (14), (15), (16) and (17) are<br />
ρ0<br />
ϕ<br />
∂ν m<br />
∂t = −∇pm − µ<br />
K νm + ρ0 α θm g<br />
¯<br />
∂θm<br />
(ρ c)m<br />
ϕ ∂sm<br />
∂t + ν m ·<br />
− ρ0 α′ sm g ¯ + µmm<br />
4π<br />
∂t + (ρ cp)f ν m ·<br />
�<br />
∇sm − (Sl − S0)<br />
� Hm<br />
�<br />
∇θm − (Tl − T0)<br />
dm<br />
e3<br />
∂h � m<br />
+ hm · ∇hm , (34)<br />
∂x3<br />
dm<br />
e3<br />
�<br />
= km∇ 2 θm, (35)<br />
�<br />
= Mm∇ 2 sm, (36)<br />
∂hm = Hm<br />
∂t<br />
∂ν m<br />
∂x3<br />
+ (hm · ∇)νm − (νm · ∇)hm + ηm∇ 2 hm, (37)<br />
where e3 is the unit vector <strong>in</strong> the x3−direction. νf, νm are solenoidal.<br />
The boundary conditions, on the upper boundary x3 = df + F (t, xα), the conditions (18), (19)<br />
and (20) become<br />
(n3 − 1)(Tu − T0) + df ni θ,i + Nu θf + Nu(Tu − T0) F<br />
(n3 − 1)(Su − S0) + df ni s,i + Nus sf + Nus(Su − S0) F<br />
ν 3 − ∂F<br />
∂xα<br />
df<br />
df<br />
= 0, (38)<br />
= 0, (39)<br />
να = ∂F<br />
, (40)<br />
∂t<br />
where Nu and Nus are Nusselt numbers<br />
Nu = Ldf, Nus = Cdf.<br />
The surface stress conditions (21), (22), the normal component and the tangential surface stress<br />
are<br />
τ(T, S)b α α = −pf + µmf<br />
8π<br />
− µmf<br />
− ρ0 g<br />
2df<br />
�<br />
(hjnj) 2 �<br />
+ 2Hfn3hjnj + ρ0 g F<br />
8π H2 f (1 − n 2 3) + 2ρ0 ν νi,j n<strong>in</strong>j<br />
�<br />
α(Tu − T0) − α′(Su − S0)<br />
�<br />
F 2 , (41)<br />
∂τ<br />
∂T T,α + ∂τ<br />
∂S S,α = ρ0ν(νi,j + νj,i)nj xi,α + µmf<br />
4π (Hfn3 + hjnj)(hi xi,α). (42)<br />
In (41) and (42), it is assumed that the derivative of the surface tension with respect to temperature<br />
and concentration are evaluated at<br />
T = Tu + (Tu − T0)d −1<br />
f F + θf, S = Su + (Su − S0)d −1<br />
f F + sf,<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
42 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
whereas the boundary conditions on the <strong>in</strong>terface x3 = 0 are<br />
θf = θm, sf = sm, kf<br />
−pf + 2µ ∂ωf<br />
∂x3<br />
∂uf<br />
∂x3<br />
= αBJ<br />
√K (uf − um),<br />
∂θf<br />
∂x3<br />
= km<br />
∂θm<br />
∂x3<br />
= −pm, hf = hm, kf<br />
∂νf<br />
∂x3<br />
, Mf<br />
∂hf<br />
∂x3<br />
∂sf<br />
∂x3<br />
= km<br />
= αBJ<br />
√K (νf − νm),<br />
= Mm<br />
and the boundary conditions on the lower boundary x3 = −dm<br />
∂sm<br />
, ωf = ωm,<br />
∂x3<br />
∂hm<br />
, (43)<br />
∂x3<br />
ωm(−dm) = 0, θm(−dm) = 0, sm(−dm) = 0, µmmHi is cont<strong>in</strong>uous. (44)<br />
5 NON-DIMENSIONALZATION<br />
We now non-dimensionalize the equations (30 − 33) and (34 − 37) by us<strong>in</strong>g the relations<br />
x = df ˆxf, t = d2 f<br />
sf = |So − Su| ν<br />
Mf<br />
λf<br />
ˆtf, ν f = ν<br />
ˆsf, pf = ρ0 ν 2<br />
d 2 f<br />
df<br />
For the fluid <strong>layer</strong>, and us<strong>in</strong>g the relations<br />
x = dm ˆxm, t = d2 m<br />
sm = |Sl − S0| ν<br />
Mm<br />
λm<br />
ˆtm, ν m = ν<br />
ˆsm, pm = ρ0 ν 2<br />
d 2 m<br />
ˆν f, θf = |To − Tu| ν<br />
ˆpf, h f = Hf λf<br />
dm<br />
ηf<br />
λf<br />
ˆθf,<br />
ˆν m, θm = |Tl − T0| ν<br />
ˆpm, h m = Hm λm<br />
For the <strong>porous</strong> medium <strong>layer</strong>. Where λf = kf<br />
(ρ cp)f and λm = km<br />
(ρ cp)f<br />
of the fluid <strong>layer</strong> and <strong>porous</strong> medium <strong>layer</strong> respectively.<br />
The equations (30 − 33) can be written <strong>in</strong> the form<br />
P −1 ∂ˆν f<br />
rf<br />
∂ˆtf<br />
∂ ˆ θf<br />
∂ˆtf<br />
1<br />
Lef<br />
∂ ˆ h f<br />
∂ˆtf<br />
ηm<br />
ˆh f. (45)<br />
λm<br />
+ ˆν f · ∇ˆν f = −∇ˆpf + ∇ 2 ˆν f + Raf ˆ θf e3 − Rasf ˆsf e3<br />
+ P −1<br />
rf Qf<br />
∂ ˆ h f<br />
∂ˆx3<br />
+ QfP −1<br />
rf<br />
ˆθm,<br />
ˆh m. (46)<br />
are the thermal diffusivity<br />
P −1<br />
mf (ˆ h f · ∇ ˆ h f), (47)<br />
+ Prf ˆν f · ∇ ˆ θf = sign(T0 − Tu) ωf + ∇ 2ˆ θf, (48)<br />
∂ˆsf<br />
∂ˆtf<br />
= Pmf<br />
+ Scf ˆν f · ∇ˆsf = sign(S0 − Su) ωf + ∇ 2 ˆsf, (49)<br />
∂ˆν f<br />
∂ˆx3<br />
+ ( ˆ h f · ∇)ˆν f − (ˆν f · ∇) ˆ h f + Pmf∇ 2ˆ hf, (50)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 43<br />
where Prf, Pmf, Qf, Scf, Lef, Raf and Rasf are non-dimensional numbers denote the viscous<br />
Prandtl number, magnetic Prandtl number, Chandrasekhar number, Schmidet number, Lewis<br />
number, thermal Rayleigh number and solute Rayleigh number of the fluid <strong>layer</strong> where<br />
Prf = ν<br />
λf<br />
, Pmf = ηf<br />
λf<br />
Raf = α g d3 f |T0 − Tu|<br />
λf ν<br />
, Qf = µmf H2 f d2f , Scf =<br />
4π ρ0 ν ηf<br />
ν<br />
, Lef =<br />
Mf<br />
Mf<br />
,<br />
λf<br />
, Rasf = α′ g d3f |S0 − Su|<br />
,<br />
Mf ν<br />
the equations (34 − 37) can be written <strong>in</strong> the form<br />
P −1<br />
rm<br />
Da ∂ˆν m<br />
ϕ ∂ˆtm<br />
= −∇ˆpm − ˆν m + Ram ˆ θm e3 − Rasm ˆsm e3<br />
+ Qm P −1<br />
rmDa ∂ˆ hm + Qm Da P<br />
∂ˆx3<br />
−1<br />
rm P −1<br />
mm( ˆ hm · ∇ˆ hm), (51)<br />
Gm<br />
∂ ˆ θm<br />
∂ˆtm<br />
+ Prm ˆν m · ∇ˆ θm = sign(Tl − T0) ωm + ∇ 2ˆ θm, (52)<br />
ϕ<br />
Lem<br />
∂ˆsm<br />
∂ˆtm<br />
+ Scm ˆν m · ∇ˆsm = sign(Sl − S0) ωm + ∇ 2 ˆsm, (53)<br />
∂ˆ hm ∂ˆν m<br />
= Pmm + (<br />
∂ˆtm ∂ˆx3<br />
ˆ hm · ∇)ˆν m − (ˆν m · ∇) ˆ hm + Pmm∇ 2ˆ hm, (54)<br />
(ρ c)m<br />
where Gm = (ρ cp)f , Prm, Pmm, Qm, Scm, Lem, Da, Ram and Rasm are non-dimensional<br />
numbers denote the viscous Prandtl number, magnetic Prandtl number, Chandrasekhar number,<br />
Schmidet number, Lewis number, Darcy number, thermal Rayleigh number and solute Rayleigh<br />
number of the <strong>porous</strong> medium <strong>layer</strong> where<br />
Prm = ν<br />
λm<br />
Da = K<br />
d 2 m<br />
, Pmm = ηm<br />
λm<br />
, Qm = µmm H2 m d2 m<br />
, Scm =<br />
4π ρ0 ν ηm<br />
ν<br />
Mm<br />
, Ram = α g d3 m|Tl − T0| K<br />
λm ν<br />
, Lem = Mm<br />
,<br />
λm<br />
, Rasm = α′ g d3m|Sl − S0| K<br />
.<br />
Mm ν<br />
Us<strong>in</strong>g (45) and (46) <strong>in</strong> the upper boundary at x3 = 1 + f(t + xα) we obta<strong>in</strong><br />
sign(T0 − Tu)(1 − n3) + Prf ni ˆ θ,i + Nu � Prf ˆ θf − sign(T0 − Tu) f � = 0, (55)<br />
�<br />
sign(S0 − Su)(1 − n3) + Scf ni ˆs,i + Nus Scf ˆsf − sign(S0 − Su) f � = 0, (56)<br />
ˆν 3 − ∂f<br />
∂ˆxα<br />
ˆν α = P −1 ∂f<br />
rf ,<br />
∂t<br />
(57)<br />
lim<br />
x3→1− ˆhi = µ0<br />
µmf<br />
Cr −1 P −1<br />
rf<br />
lim<br />
x3→1 +<br />
∂φu<br />
∂ˆxi<br />
b α α = − ˆpf + 1<br />
2<br />
+ B0 Cr −1 P −1<br />
rf<br />
− 1<br />
2<br />
� −1 −1<br />
QfPrf Prf ( ˆ hjnj) 2 + 2n3 ˆ �<br />
hjnj<br />
1<br />
f −<br />
2 Qf Pmf(1 − n 2 3) + 2νi,j ˆ n<strong>in</strong>j<br />
�<br />
sign(T0 − Tu)P −1<br />
rf Raf − sign(S0 − Su)S −1<br />
cf Rasf Lef<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.<br />
�<br />
f 2 ,<br />
(58)<br />
(59)
44 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
−Ma<br />
�<br />
Prf ˆ<br />
�<br />
θ,α − sign(T0 − Tu) f,α<br />
Prf<br />
− Mas<br />
Scf<br />
�<br />
�<br />
Scfs,α ˆ − sign(S0 − Su) f,α<br />
= ( νi,j ˆ + νj,i) ˆ nj ˆxi,α + Qf ˆ hi ˆxi,α(n3 + P −1<br />
mf ˆ hj nj), (60)<br />
where φu is the non-dimensional magnetic potential function <strong>in</strong> the region x3 > 1 + f and Cr,<br />
B0, Ma & Mas are non-dimensional numbers denote the Crispation number, Bond number,<br />
thermal Marangoni number and solute Marangoni number where<br />
Cr = ρ0 ν λf<br />
df τ(Tu, Su) , B0 = ρ0 g d 2 f<br />
τ(Tu, Su) ,<br />
Ma = − ∂τ<br />
∂T<br />
df|T0 − Tu|<br />
ρ0 ν λf<br />
The middle boundary at x3 = 0<br />
, Mas = − ∂τ<br />
∂S<br />
df|S0 − Su|<br />
.<br />
ρ0 ν Mf<br />
ΥT ˆ θf(0) = εT ˆ θm(0), ΥS ˆsf(0) = εS sm(0), ˆ<br />
∂ ˆ θf(0)<br />
= εT<br />
∂ ˆx3<br />
∂ ˆ θm(0)<br />
,<br />
∂ ˆx3<br />
∂ ˆsf(0)<br />
∂ ˆx3<br />
=<br />
∂sm(0) ˆ<br />
εS , ˆωf(0) =<br />
∂ ˆx3<br />
ˆ ∂ ˆωf(0)<br />
d ωm(0), ˆ ˆpf(0) − 2<br />
∂ ˆx3<br />
= ˆ d2 Da ˆ pm(0),<br />
ˆhf(0) =<br />
1<br />
hm(0), ˆ<br />
ˆn εT<br />
∂ ˆ hf(0)<br />
=<br />
∂ ˆx3<br />
ˆ d<br />
ˆn ε2 T<br />
∂ ˆ<br />
∂ûf(0)<br />
∂ ˆx3<br />
=<br />
hm(0)<br />
,<br />
∂ ˆx3<br />
(61)<br />
ˆ d2 �<br />
αBJ 1<br />
√<br />
Da ˆd ûf(0)<br />
∂ ˆνf(0)<br />
∂ ˆx3<br />
=<br />
�<br />
− um(0) ˆ ,<br />
ˆ d2 �<br />
αBJ 1<br />
√<br />
Da ˆd ˆνf(0)<br />
�<br />
− νm(0) ˆ ,<br />
where<br />
ˆd = df<br />
, εT = λf<br />
dm<br />
λm<br />
ΥT = |T0 − Tu|<br />
|Tl − T0| = ˆ d<br />
Raf = ˆ d 4<br />
εT<br />
= kf<br />
, εS =<br />
km<br />
Mf<br />
, ˆn =<br />
Mm<br />
Hf ηm<br />
,<br />
Hm ηf<br />
, ΥS = |S0 − Su|<br />
|Sl − S0| = ˆ d<br />
ε 2 T Da Ram, Rasf = ˆ d 4<br />
εS<br />
, Prf = 1<br />
ε 2 S Da Rasm, Lef = εS<br />
The lower boundary at x3 = −1 the condition becomes<br />
ωm ˆ = 0, θm<br />
ˆ = 0, sm ˆ = 0, lim<br />
x3→1 +<br />
ˆh m = µ0<br />
µmm<br />
εT<br />
lim<br />
x3→1 −<br />
where φl is the non-dimensional magnetic potential function.<br />
6 LINEARIZED PROBLEM<br />
εT<br />
Lem.<br />
Prm,<br />
∂φl<br />
, (62)<br />
∂ˆxi<br />
The l<strong>in</strong>earized of equations is obta<strong>in</strong>ed by ignor<strong>in</strong>g all products, the powers more than first and<br />
by dropp<strong>in</strong>g hat the superscript ˆ ( ) for fluid <strong>layer</strong> the result<strong>in</strong>g<br />
P −1 ∂ν f<br />
rf<br />
∂tf<br />
= −∇pf + ∇ 2 νf + Raf θfe3 − Rasf sf e3 + Qf P −1 ∂hf rf , (63)<br />
∂x3<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
∂θf<br />
∂tf<br />
1<br />
Lef<br />
Pm −1<br />
f<br />
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 45<br />
= ∇ 2 θf + ξT ωf, (64)<br />
∂sf<br />
∂tf<br />
∂h f<br />
∂tf<br />
= ∇ 2 sf + ξS ωf, (65)<br />
= ∂ν f<br />
∂x3<br />
+ ∇ 2 h f, (66)<br />
and for the <strong>porous</strong> medium <strong>layer</strong> the result<strong>in</strong>g<br />
Da<br />
ϕ Prm<br />
Gm<br />
ϕ<br />
Lem<br />
Pm −1<br />
m<br />
where<br />
∂θm<br />
∂tm<br />
∂ν m<br />
∂tm<br />
∂sm<br />
∂tm<br />
∂h m<br />
∂tm<br />
= −∇pm − ν m + Ram θme 3 − Rasm sm e 3 + Qm P −1<br />
rmDa ∂hm , (67)<br />
∂x3<br />
= ∇ 2 θm + ξT ωm, (68)<br />
= ∇ 2 sm + ξS ωm, (69)<br />
= ∂ν m<br />
∂x3<br />
ξT = sign(Tl − T0) = sign(T0 − Tu) =<br />
ξs = sign(Sl − S0) = sign(S0 − Su) =<br />
+ ∇ 2 h m, (70)<br />
� +1 when heat<strong>in</strong>g from below,<br />
−1 when heat<strong>in</strong>g from above.<br />
� +1 when solut<strong>in</strong>g from below,<br />
−1 when solut<strong>in</strong>g from above.<br />
The boundary conditions on the lower boundary x3 = −1 are unchanged except to clear superscript,<br />
but the l<strong>in</strong>earized upper boundary conditions<br />
Prf<br />
Scf<br />
∂θf<br />
+ Nu(Prf θf − ξT f) = 0,<br />
∂x3<br />
∂sf<br />
+ Nus(Scf sf − ξS f) = 0,<br />
∂x3<br />
ν 3 = P −1<br />
rf<br />
∂f<br />
∂t ,<br />
lim<br />
x3→1− hi = µ0<br />
µmf<br />
lim<br />
x3→1 +<br />
∂φu<br />
,<br />
∂xi<br />
Cr −1 P −1<br />
rf bαα = −pf + Qf P −1<br />
rf h3 + B0Cr −1 P −1<br />
rf<br />
∂ν3<br />
f + 2 ,<br />
∂x3<br />
−Ma P −1<br />
rf (Prf θ,α − ξT f,α) − Mas S −1<br />
cf (Scf s,α − ξSf,α) = ν3,α + ∂να<br />
∂x3<br />
+ Qfhα.<br />
To simplize the normal stress boundary condition on the <strong>in</strong>terface plane we will elim<strong>in</strong>ate hydrostatic<br />
pressure term, then by tak<strong>in</strong>g two-dimensional Laplacian of (61)6 we obta<strong>in</strong><br />
∆2pf(0) − 2 ∂<br />
∆2ωf(0) =<br />
∂x3<br />
ˆ d4 Da ∆2pm(0), (71)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
46 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
s<strong>in</strong>ce<br />
∇νf = 0 −→ ∂uf<br />
∂x1<br />
∇ν m = 0 −→ ∂um<br />
∂x1<br />
+ ∂νf<br />
∂x2<br />
+ ∂νm<br />
∂x2<br />
= − ∂ωf<br />
,<br />
∂x3<br />
= − ∂ωm<br />
. (72)<br />
∂x3<br />
Then we take the divergence of equations (63) and (67) we get respectively<br />
∆2pf = 1<br />
Prf<br />
∆2pm = Da<br />
ϕ Prm<br />
∂ ∂ωf<br />
∂t ∂x3<br />
∂ ∂ωm<br />
∂t ∂x3<br />
2 ∂ωf<br />
− ∇ , (73)<br />
∂x3<br />
substitute from (74) and (73) <strong>in</strong>to (71) to get<br />
∂<br />
∂x3<br />
�<br />
∇ 2 ωf(0) − 1<br />
Prf<br />
+ ∂ωm<br />
, (74)<br />
∂x3<br />
∂ωf(0)<br />
∂t<br />
�<br />
+ 2∆2ωf(0) = − ˆ d4 �<br />
Da<br />
Da ϕ Prm<br />
∂<br />
+ 1<br />
∂t<br />
�<br />
∂ωm(0)<br />
. (75)<br />
∂x3<br />
Comb<strong>in</strong>e the Beavers and Joseph boundary conditions on the <strong>in</strong>terface plane (61)9 and (61)10<br />
by differentiate equations (61)9 and (61)10 with respect to x1 and x2 respectively, we get<br />
∂<br />
∂x3<br />
∂uf(0)<br />
∂x1<br />
= ˆ d αBJ<br />
√<br />
Da<br />
�<br />
∂uf(0)<br />
−<br />
∂x1<br />
ˆ d<br />
2 ∂um(0)<br />
∂x1<br />
∂ ∂νf(0)<br />
=<br />
∂x3 ∂x2<br />
ˆ d<br />
�<br />
αBJ ∂νf(0)<br />
√ −<br />
Da ∂x2<br />
ˆ d<br />
∂x2<br />
by add<strong>in</strong>g (76) to (77) and us<strong>in</strong>g (72) we get<br />
∂<br />
∂x3<br />
� αBJ<br />
√ Da<br />
ˆd ωf(0) − ∂ωf(0)<br />
�<br />
∂x3<br />
2 ∂νm(0)<br />
= ˆ d3 αBJ<br />
√<br />
Da<br />
�<br />
, (76)<br />
�<br />
, (77)<br />
∂ωm<br />
. (78)<br />
∂x3<br />
Recall that the magnetic field on <strong>in</strong>sulat<strong>in</strong>g material is irrotational and is derived from a potential<br />
function φ(t, xi) which is the solution of Laplace’s equation. Let φ = ψ(t, x3) e i(px1+qx2) then<br />
∂ 2 ψ<br />
∂x 2 3<br />
− a 2 ψ = 0, a 2 = p 2 + q 2 ,<br />
∂ψ<br />
∂x3<br />
→ 0 as |x3| → ∞,<br />
where a = � p 2 + q 2 the dimensionless wave number. Trivially φl and φu have functional form<br />
φl = Cl(t) e ax3 e i(px1+qx2) , φu = Cu(t) e −ax3 e i(px1+qx2) .<br />
When x3 < −1 then h = Cl(t) (ip, iq, a) and cont<strong>in</strong>uity of the magnetic <strong>in</strong>duction across<br />
x3 = −1 requires that<br />
h3,3 = 1<br />
µm<br />
b3,3 = −1<br />
µm<br />
bα,α = µ0<br />
µm<br />
a 2 Cl(t) = a<br />
µm<br />
b3 = ah3.<br />
With a similar argument on x3 = 1 + f(t, xα). Hence the magnetic boundary conditions are<br />
∂h3<br />
− am h3 = 0, x3 = −1,<br />
∂x3<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 47<br />
∂h3<br />
+ af h3 = 0, x3 = 1.<br />
∂x3<br />
We put ω = ν3, h = h3 and take the double curl of equation (63) and (67) to elim<strong>in</strong>ate<br />
hydrostatic pressure pf and pm respectively, and we take the third component for all equations<br />
of the system. Therefore, for the fluid <strong>layer</strong>, we get<br />
P −1<br />
rf<br />
∂θf<br />
∂tf<br />
1<br />
Lef<br />
∂<br />
∂tf<br />
∇ 2 ωf − QfP −1<br />
rf ∇2 Dhf = ∇ 4 ωf + Raf △2θf − Rasf △2sf,<br />
= ξT ωf + ∇ 2 θf,<br />
∂sf<br />
∂tf<br />
= ξSωf + ∇ 2 sf,<br />
P −1<br />
mf<br />
∂hf<br />
∂tf<br />
= ∇ 2 hf + Dωf.<br />
Also for the <strong>porous</strong> medium <strong>layer</strong> , we obta<strong>in</strong><br />
Da<br />
ϕ Prm<br />
Gm<br />
ϕ<br />
Lem<br />
P −1<br />
mm<br />
∂θm<br />
∂tm<br />
∂<br />
∂tm<br />
∂sm<br />
∂tm<br />
∂hm<br />
∂tm<br />
∇ 2 ωm − Qm P −1<br />
rmDa ∇ 2 Dhm = −∇ 2 ωm + Ram △2θm − Rasm △2sm,<br />
= ξT ωm + ∇ 2 θm,<br />
= ξSωm + ∇ 2 sm,<br />
= ∇ 2 hm + Dωm,<br />
where ∇ 2 is the 3-D Laplacian and ∆2 is the 2-D Laplacian.<br />
Now we look for solution of the form<br />
φ(t, xi) = φ(x3) e σt e i(px1+qx2) , φ = {ω, h, p, θ, s}.<br />
f(t, xα) = f0 e σt e i(px1+qx2) , f0 is constant.<br />
Thus the equation of fluid <strong>layer</strong> become<br />
σf P −1<br />
rf (D2 f − a 2 f) ωf − σf P −1<br />
mf Qf P −1<br />
rf Df hf = (D 2 f − a 2 f) 2 ωf<br />
− Raf a 2 f θf + Rasf a 2 f sf − Qf P −1<br />
rf D2 f ωf, (79)<br />
σf θf = ξT ωf + (D 2 f − a 2 f) θf, (80)<br />
σf<br />
Lef<br />
sf = ξS ωf + (D 2 f − a 2 f) sf, (81)<br />
σf P −1<br />
mf hf = (D 2 f − a 2 f) hf + Df ωf, (82)<br />
while the <strong>porous</strong> medium equations take the form<br />
−Da σm<br />
ϕ Prm<br />
(D 2 m − a 2 m) ωm − σm Da P −1<br />
mm Qm P −1<br />
rmDm hm = (D 2 m − a 2 m) ωm<br />
+ Ram a 2 m θm − Rasm a 2 m sm + Qm P −1<br />
rmDa D 2 m ωm, (83)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
48 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Gm σm θm = ξT ωm + (D 2 m − a 2 m) θm, (84)<br />
ϕ σm<br />
Lem<br />
sm = ξS ωm + (D 2 m − a 2 m) sm, (85)<br />
σm P −1<br />
mm hm = (D 2 m − a 2 m) hm + Dm ωm, (86)<br />
from these equations<br />
af = ˆ d am, σf = ˆ d 2<br />
εT<br />
σm,<br />
Dm = ∂<br />
; x3 ∈ [−1, 0], Df =<br />
∂x3<br />
∂<br />
; x3 ∈ [0, 1],<br />
∂x3<br />
where af and am are non-dimensional wave number <strong>in</strong> the fluid <strong>layer</strong> and <strong>porous</strong> medium <strong>layer</strong><br />
respectively, σf and σm are growth rates.<br />
As a preamble to the formulation of the f<strong>in</strong>al boundary conditions on x3 = 1, the pressure<br />
everywhere is given by the equation<br />
p = 1 � 3 2 −1<br />
D ω − a Dω + Q Pr D 2 h − σ P −1<br />
r Dω �<br />
a 2<br />
hence the boundary conditions on x3 = 1 are<br />
Prf Dfθf + Nu(Prfθf − ξT f0) = 0, (87)<br />
Scf Df sf + Nus(Scf sf − ξS f0) = 0, (88)<br />
ωf − σf P −1<br />
rf f0 = 0, (89)<br />
Dfhf + af hf = 0, (90)<br />
a 2 f(B0 + a 2 f)f0 − Cr Prf(D 3 fωf − 3a 2 f Dfωf − Qf P −1<br />
rf Dfωf) =<br />
= σf Cr(Qf P −1<br />
mf hf − Dfωf), (91)<br />
(D 2 f + a 2 f)ωf + Ma P −1<br />
rf (Prf θf − ξT f0)a 2 f<br />
+ Mas S −1<br />
cf (Scf sf − ξS f0)a 2 f = 0. (92)<br />
The middle boundary on x3 = 0, become<br />
ΥT θf = εT θm, ΥSsf = εS sm, Dfθf = εT Dmθm, Dfsf = εS Dmsm,<br />
ωf = ˆ d ωm,<br />
D 3 fωf − 3a 2 f Dfωf − σf<br />
hf =<br />
Dfhf =<br />
1<br />
ˆn εT<br />
ˆd<br />
ˆn ε 2 T<br />
hm,<br />
Prf<br />
Dfωf = − ˆ d4 �<br />
Da<br />
�<br />
σm + 1 Dmωm, (93)<br />
Da ϕ Prm<br />
Dmhm, αBJ<br />
√ ˆdDfωf − D<br />
Da<br />
2 fωf = ˆ d3 αBJ<br />
√ Dmωm.<br />
Da<br />
The lower boundary on x3 = −1 become<br />
ωm = 0, θm = 0, sm = 0, Dmhm − amhm = 0. (94)<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 49<br />
7 METHOD OF SOLUTION<br />
The first order Chebyshev spectral method is applied to solve the equations (79 − 86) with<br />
the relevant boundary conditions (87 − 94), and we map x3 ∈ [0, 1] and x3 ∈ [−1, 0] <strong>in</strong> to<br />
z ∈ [−1, 1] by the transformations z = 2x3 − 1 and z = 2x3 + 1 respectively, to obta<strong>in</strong><br />
∂<br />
∂x3<br />
= 2 ∂<br />
∂z , thus Df = Dm = 2 ∂<br />
∂z<br />
Let the variables y1, y2, . . . , y18 be def<strong>in</strong>ed by<br />
= D, z ∈ [−1, 1].<br />
y1 = ωf, y2 = Dfωf, y3 = D 2 fωf, y4 = D 3 fωf, y5 = θf,<br />
y6 = Dfθf, y7 = sf, y8 = Dfsf, y9 = hf, y10 = Dfhf,<br />
y11 = ωm, y12 = Dmωm, y13 = θm, y14 = Dmθm, y15 = sm,<br />
y16 = Dmsm, y17 = hm, y18 = Dmhm.<br />
Then the equations (79 − 86) can be rewritten <strong>in</strong> a system of eighteen ord<strong>in</strong>ary differential<br />
equations of first order, s<strong>in</strong>ce Df = Dm = D and if we put σm = σ then σf = ˆ d2 σ so that<br />
εT<br />
Eigenvalue problem can be reformulated <strong>in</strong> the form<br />
dY<br />
dz<br />
= AY + σBY, z ∈ [−1, 1]<br />
where A and B are real 18×18 matrices. The f<strong>in</strong>al Eigenvalue problem reduces to EX = σF X<br />
where matrices E and F have the block forms. The boundary conditions replace the Mth,<br />
2 Mth,. . ., 18 Mth. rows of E and F . First to make the boundary condition on x3 = 1 four<br />
condition. Its convenient to neglect<strong>in</strong>g f0 and neglect<strong>in</strong>g normal stress upper boundary. All<br />
boundary condition take the form<br />
y6 + Nu y5 = 0,<br />
y8 + Nus y7 = 0,<br />
y1 = 0,<br />
y3 + a 2 f y1 + a 2 f Ma y5 + a 2 f Mas y7 = 0,<br />
y10 + af y9 = 0,<br />
y1 − ˆ d y11 = 0, ΥT y5 − εT y13 = 0, ΥSy7 − εS y15 = 0, y6 − εT y14,<br />
y8 − εS y16 = 0, y4 − 3a 2 f y2 + ˆ d4 Da y12<br />
� ˆ2 d<br />
= σ<br />
y9 − 1<br />
ˆn εT<br />
y17 = 0, y10 − ˆ d<br />
ˆn ε 2 T<br />
y17 = 0, αBJ<br />
√<br />
Da<br />
εT Prf<br />
y11 = 0, y13 = 0, y15 = 0, y18 − amy17 = 0.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.<br />
y2 − ˆ d 4<br />
ϕ Prm<br />
ˆdy2 − y3 − ˆ d3 αBJ<br />
√<br />
Da<br />
y12<br />
�<br />
,<br />
ˆdy12 = 0,
50 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Figure 2: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of depth ratio when, Da = 4 × 10 −6 , Ma = 100, Mas = 10,<br />
Qm = 10, Rams = 50.<br />
8 RESULTS AND DISCUSSION<br />
The Eigenvalue problem (79 − 86) with boundary conditions (87 − 94) by us<strong>in</strong>g first order of<br />
Chebyshev spectral method is transformed to a system of ten ord<strong>in</strong>ary differential equations of<br />
first order <strong>in</strong> the fluid <strong>layer</strong> and a system of eight ord<strong>in</strong>ary differential equations of first order <strong>in</strong><br />
the <strong>porous</strong> <strong>layer</strong> with eighteen boundary conditions. In this paper we will discuss the numerical<br />
results when the Eigenvalues are real then the stationary <strong>in</strong>stability happens as shown <strong>in</strong> the<br />
figures (2) − (11). These figures display the relation between Ram and am by consider<strong>in</strong>g<br />
chosen values of the non-dimensions constants<br />
Prf = 6, Gm = 10, ϕ = 0.3, αBJ = 0.1, Lef = 1, Pmm = 6, εT = 0.7,<br />
εS = 3.75, Pmf = 6, ˆn = 1, Nu = 10, Nus = 5, ξT = 1, ξS = 1.<br />
For deferent values of ˆ d, Qm, Da, Ma, Ras, Rasm we found the follow<strong>in</strong>g results will be held:<br />
1. The <strong>in</strong>creas<strong>in</strong>g of depth ratio has unstabiliz<strong>in</strong>g effect for the system as shown <strong>in</strong> figure<br />
(2).<br />
2. L<strong>in</strong>ear magnetic field has a stabiliz<strong>in</strong>g effect for the system as shown <strong>in</strong> figure (3).<br />
3. The permeability of <strong>porous</strong> medium has a stabiliz<strong>in</strong>g effect for the system on the other<br />
hand the fluid becomes more stable when the permeability <strong>in</strong>creases as shown <strong>in</strong> figures<br />
(4) and (5).<br />
4. The <strong>in</strong>creas<strong>in</strong>g of the thermal Marangoni number has unstabiliz<strong>in</strong>g effect for the system<br />
as shown <strong>in</strong> figures (6), (7) and (8).<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 51<br />
Figure 3: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of Chandrasekhar number when, Da = 4 × 10 −6 , Ma = 100,<br />
Mas = 10, ˆ d = 0.07, Rams = 50.<br />
5. The <strong>in</strong>creas<strong>in</strong>g of the solute Marangoni number has unstabiliz<strong>in</strong>g effect for the system as<br />
shown <strong>in</strong> figures (9) and (10).<br />
6. The <strong>in</strong>creas<strong>in</strong>g of the solute Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> has a stabiliz<strong>in</strong>g effect for<br />
the system, i. e. there is a direct relation between thermal and solute Rayleigh number <strong>in</strong><br />
<strong>porous</strong> <strong>layer</strong> as shown <strong>in</strong> figure (11).<br />
9 CONCLUSIONS<br />
In this paper we have implemented a D variant of the Chebyshev numercal method and have<br />
derived accurate results for the Marangoni Convections <strong>in</strong> a <strong>horizontal</strong> <strong>porous</strong> <strong>layer</strong> superposed<br />
by a fluid <strong>layer</strong> <strong>in</strong> the presence of uniform vertical magnetic field and solute. The Chebyshev<br />
method is very important for this general class of problem s<strong>in</strong>ce it is highly accurate, the eigenfunctions<br />
are easy to generate, and it is easily generalized to the multi<strong>layer</strong> situation where there<br />
are many <strong>porous</strong>-fluid <strong>layer</strong>s superposed. The Chebyshev method used here also yields as many<br />
eigenvalues <strong>in</strong> the spectrum as one desires. This is particularly important <strong>in</strong> that for convection<br />
problems of current <strong>in</strong>terest, it is often the case that the eigenvalues change position, one eigenvalues<br />
which is dom<strong>in</strong>ant <strong>in</strong> a certa<strong>in</strong> region of parameter space be<strong>in</strong>g replaced by another <strong>in</strong><br />
another parameter region. Firstly we have <strong>in</strong>vestigated <strong>in</strong> detail the effect of depth ratio on the<br />
onset of <strong>in</strong>stability. Straughan,s (2001) study the effect of surface tension is <strong>in</strong>vestigated <strong>in</strong> detail<br />
and it is found that for the parameter ˆ d very small, the surface tension has strong effect on<br />
convection dom<strong>in</strong>ated by the <strong>porous</strong> medium,whereas for ˆ d larger the surface tension effect is<br />
observed only with the fluid mode as shown <strong>in</strong> figure (12). Also Chen and Chen (1988)produced<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
52 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Figure 4: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of Darcy number when, Ma = 100, Mas = 10, ˆ d = 0.05,<br />
Rams = 50, Qm = 10.<br />
a classical paper <strong>in</strong> which they studied thermal convection <strong>in</strong> a two-<strong>layer</strong> system composed of a<br />
<strong>porous</strong> <strong>layer</strong> saturated with fluid over which was a <strong>layer</strong> of the same fluid. The <strong>layer</strong> was heated<br />
from below and Chen and Chen (1988) considered the bottom of the <strong>porous</strong> <strong>layer</strong>,as well as the<br />
upper surface of the fluid,to be fixed. They showed that the l<strong>in</strong>ear <strong>in</strong>stability curves for the onset<br />
of convection motion,i.e; the Rayleigh number aga<strong>in</strong>st wavenumber curves,may be bimodal <strong>in</strong><br />
that the curves possess two local m<strong>in</strong>ima. A crucial parameter <strong>in</strong> their analysis is the number ˆ d.<br />
They <strong>in</strong>terpreted their f<strong>in</strong>d<strong>in</strong>g by show<strong>in</strong>g that for ˆ d small � ≤ 0.13 � the <strong>in</strong>stability was <strong>in</strong>itiated<br />
<strong>in</strong> the <strong>porous</strong> medium,whereas for ˆ d larger than this the mechanism changed and <strong>in</strong>stability was<br />
controlled by the fluid <strong>layer</strong> as shown <strong>in</strong> figure (13). Al Enizi (Al Enizi 2006) study where<br />
the solute concentration effect was <strong>in</strong>cluded, the effect depthratio become more stronger, due to<br />
the addition of solute concentration with the effect of surface tension as shown <strong>in</strong> figure (14).<br />
In our work it is clear from the figure that the Rayleigh number <strong>in</strong> the <strong>porous</strong> <strong>layer</strong> decreases<br />
cont<strong>in</strong>uously as the thickness of the <strong>layer</strong> <strong>in</strong>creases. And it is clear the magnetic field has a<br />
stabiliz<strong>in</strong>g effect on the system. F<strong>in</strong>ale we can take water,oil or glycer<strong>in</strong> as example of fluid and<br />
sponge as example of <strong>porous</strong> <strong>layer</strong> then we can obta<strong>in</strong> the results.<br />
REFERENCES<br />
Abdullah A (1991). Benard convection <strong>in</strong> a non-l<strong>in</strong>ear magnetic field. Ph. D. thesis, University<br />
of Glasgow.<br />
Al Enizi MA (2006). Convection <strong>in</strong> <strong>horizontal</strong> multi-<strong>layer</strong> problem with surface-tension’s effect<br />
<strong>in</strong> the presence of solute concentration. M.sc. Thesis Umm Al-qura University.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 53<br />
Figure 5: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of Darcy number when, Ma = 100, Mas = 10, ˆ d = 0.1,<br />
Rams = 50, Qm = 10.<br />
Beaver GS and Joseph DD (1967). Boundary condition at a naturally permeable wall. J. Fluid<br />
Mech. 30, pp. 197–207.<br />
Benard H (1900). Les tourbillons céllulaires dans une nappe liquide. Rev. Gen. Sci. Pure<br />
Appl. 11, pp. 1216–1271 and 1309–1328.<br />
Benard H (1901). Les tourbillons céllulaires dans une nappe liquide transportant de la chaleur<br />
par convection en regime permanent. Ann. Chim. Phys. 23, pp. 62–114.<br />
Br<strong>in</strong>kman HC (1947a). A calculation of the viscous force exerted by a flow<strong>in</strong>g fluid on a dense<br />
swarm of particles. Appl. Sci. Res. Al, pp. 27–34.<br />
Br<strong>in</strong>kman HC (1947b). On the permeability of media consist<strong>in</strong>g of closely packed <strong>porous</strong><br />
particles. Appl. Sci. Res. Al, pp. 81–86.<br />
Bukhari AF (1996). A spectral Eigenvalue method for multi-<strong>layer</strong>ed cont<strong>in</strong>ua. Ph. D. thesis,<br />
University of Glasgow, Scot.<br />
Bukhari AF (2003a). Convection <strong>in</strong> a <strong>horizontal</strong> <strong>porous</strong> <strong>layer</strong> underly<strong>in</strong>g a fluid <strong>layer</strong> <strong>in</strong> the<br />
presence of a l<strong>in</strong>ear magnetic field on both <strong>layer</strong>s. Umm Al-Qura Univ. J. Sci. Med. Eng. 15(1),<br />
pp. 95–113.<br />
Bukhari AF (2003b). Double diffusive convection <strong>in</strong> a <strong>horizontal</strong> <strong>porous</strong> <strong>layer</strong> superposed by a<br />
fluid <strong>layer</strong>. Umm al-Qura Univ., J. Sci. Med. Eng. 15(2), pp. 95–113.<br />
Chandrasekhar S (1952). On the <strong>in</strong>hibition of convection by a magnetic field. Phil. Mag.<br />
Ser. 7(340), pp. 501–532.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
54 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Figure 6: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of thermal Marangoni number when, Da = 4 × 10 −6 , Qm = 10,<br />
Mas = 10, ˆ d = 0.05, Rams = 50.<br />
Chandrasekhar S (1981). Hydrodynamic and Hydromagnetic Stability. Dover.<br />
Chaves T and Ortiz EL (1968). On the numerical solution of two-po<strong>in</strong>t boundary value problems<br />
for l<strong>in</strong>ear differential equations. Zeitschrift fur Angewandte Mathematik und Mechanik 48, pp.<br />
415–418.<br />
Chen F and Chen CF (1988). Onset of f<strong>in</strong>ger convection <strong>in</strong> a <strong>horizontal</strong> <strong>porous</strong> <strong>layer</strong> underly<strong>in</strong>g<br />
a fluid <strong>layer</strong>. J. of Heat Tran. 110, pp. 403–409.<br />
Fox L (1962). Chebyshev method for ord<strong>in</strong>ary differential equations. Computer. J. 4, pp. 318–<br />
331.<br />
Fox L and Parker IB (1968). Chebyshev polynomials <strong>in</strong> numerical analyses. Oxford University<br />
Press, London.<br />
Glicksman ME, Curiel SR, and McFadden GB (1986). Interaction of flows with crystal-melt<br />
<strong>in</strong>terface. Ann. Rev. Fluid Mech. 18, pp. 307–335.<br />
Hassanien IA and El-Hawary HM (1990). Chebyshev solution of lam<strong>in</strong>ar boundary <strong>layer</strong> flow.<br />
Intern. J. Computer Math. 33, pp. 127–132.<br />
Hassanien IA, El-Hawary HM, and Salama AA (1996). Chebyshev solution of axisymmetric<br />
stagnation flow on a cyl<strong>in</strong>der. Energy Convers. Mgmt 37(1), pp. 67–76.<br />
Hills RN, Loper DE, and Roberts PH (1983). A thermodynamically consistent model of the<br />
mushy zone. Q. J. Mech. Appl. Math. 36, pp. 505–539.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 55<br />
Figure 7: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of thermal Marangoni number when, Da = 4 × 10 −6 , Qm = 10,<br />
Mas = 10, ˆ d = 0.07, Rams = 50.<br />
Jeffreys H (1930). The <strong>in</strong>stability of a compressible fluid heated below. Proc. Camb. Phil.<br />
Soc. 26, pp. 170–172.<br />
Lamb CJ (1994). Hydro magnetic <strong>in</strong>stability <strong>in</strong> the earth core. Ph. D. thesis, Department of<br />
Mathematics, University of Glasgow.<br />
Lanczos C (1938). Trigonometric <strong>in</strong>terpolation of empirical and analytical functions. J. Math<br />
Phys. 17, pp. 123–199.<br />
L<strong>in</strong>dsay K and Ogden R (1992). A practical implementation of spectral methods resistant to<br />
the generation of spurious eigenvalue. Paper No. 92/23 University of Glasgow, Department of<br />
Mathematics, prepr<strong>in</strong>t series.<br />
Maples AL and Poirier DR (1984). Convection <strong>in</strong> the two-phase zone of solidify<strong>in</strong>g alloys.<br />
Metall. Trans. B. 15, pp. 163–172.<br />
Nield DA (1968). Onset of themohal<strong>in</strong>e convection <strong>in</strong> a <strong>porous</strong> medium. Water Resources<br />
Res. 4, pp. 553–560.<br />
Nield DA (1977). Onset of convection <strong>in</strong> a fluid <strong>layer</strong> overly<strong>in</strong>g a <strong>porous</strong> medium. J. Fluid<br />
Mech. 81, pp. 513–522.<br />
Orszag S (1971a). Accurate solution of the orr-sommerfeld stability equation. J. Fluid Mech. 50,<br />
pp. 689–703.<br />
Orszag S (1971b). Approximations to flows with<strong>in</strong> slabs: spheres and cyl<strong>in</strong>ders. Phys. Rev.<br />
Letts. 26, pp. 1100–1103.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
56 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Figure 8: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of thermal Marangoni number when, Da = 4 × 10 −6 , Qm = 10,<br />
Mas = 10, ˆ d = 0.1, Rams = 50.<br />
Rayleigh L (1916). On convection currents <strong>in</strong> a <strong>horizontal</strong> <strong>layer</strong> of fluid when the temperature<br />
is on the side. Phil. Mag. 32(192), pp. 529–547.<br />
Somerton CW and Catton I (1982). On the thermal <strong>in</strong>stability of superposed <strong>porous</strong> and fluid<br />
<strong>layer</strong>s. ASME J. Heat Trans. A. 104, pp. 160–165.<br />
Straughan B (2001). Surface-tension-driven convection <strong>in</strong> a fluid overly<strong>in</strong>g a <strong>porous</strong> <strong>layer</strong>. J.<br />
Comput. Phys. 170, pp. 320–337.<br />
Straughan B (2002). Effect of property variation and model<strong>in</strong>g on convection <strong>in</strong> a fluid overly<strong>in</strong>g<br />
a <strong>porous</strong> <strong>layer</strong>. Int. J. Number. Anal. Meth. Geomech. 26, pp. 75–97.<br />
Sun WJ (1973). Convection Instability <strong>in</strong> Superposed Porous and Free Layers. Ph. D. thesis,<br />
University of M<strong>in</strong>nesota, M<strong>in</strong>neapolis.<br />
Taunton J and Lightfoot E (1972). Thermohal<strong>in</strong>e <strong>in</strong>stability and salt f<strong>in</strong>gers <strong>in</strong> a <strong>porous</strong> medium.<br />
Phys. Of fluid 15(5), pp. 748–753.<br />
Thompson W (1951). Thermal convection <strong>in</strong> a magnetic field. Phi. Mag. Ser. 7, pp. 1417–1432.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 57<br />
Figure 9: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of solute Marangoni number when, Da = 4 × 10 −6 , Qm = 10,<br />
Ma = 100, ˆ d = 0.05, Rams = 50.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
58 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Figure 10: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of solute Marangoni number when, Da = 4×10 −6 , Qm = 10,<br />
Ma = 100, ˆ d = 0.1, Rams = 50.<br />
Figure 11: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> Ram and wave number<br />
<strong>in</strong> <strong>porous</strong> <strong>layer</strong> am for various of solute Rayleigh number when, Ma = 100, Mas = 10,<br />
ˆd = 0.05, Da = 4 × 10 −6 , Qm = 10.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 59<br />
Figure 12: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> and wave number <strong>in</strong><br />
<strong>porous</strong> <strong>layer</strong> for various of depth ratio when, Ma = −100 φ = 0.3, Pr = 6.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
60 R. T. Matoog and Abdl-Fattah K. Bukhari<br />
Figure 13: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> and wave number <strong>in</strong><br />
<strong>porous</strong> <strong>layer</strong> for various of depth ratio.<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.
Marangoni Convections <strong>in</strong> a Horizontal Porous Layer Superposed by a Fluid Layer 61<br />
Figure 14: The relation between thermal Rayleigh number <strong>in</strong> <strong>porous</strong> <strong>layer</strong> and wave number <strong>in</strong><br />
<strong>porous</strong> <strong>layer</strong> for various of dept ratio when, Ma = 100, Mas = 10, Da = 4 × 10 −6 .<br />
Int. J. of Appl. Math. and Mech. 4(6): 34-61, 2008.