Professional Documents
Culture Documents
Textbook Mixed Convection in Fluid Superposed Porous Layers 1St Edition John M Dixon Ebook All Chapter PDF
Textbook Mixed Convection in Fluid Superposed Porous Layers 1St Edition John M Dixon Ebook All Chapter PDF
https://textbookfull.com/product/convection-in-porous-media-5th-
edition-donald-a-nield/
https://textbookfull.com/product/multiphase-fluid-flow-in-porous-
and-fractured-reservoirs-1st-edition-wu/
https://textbookfull.com/product/fluid-dynamics-part-3-boundary-
layers-anatoly-i-ruban/
https://textbookfull.com/product/applications-of-heat-mass-and-
fluid-boundary-layers-1st-edition-r-o-fagbenle/
Physicochemical Fluid Dynamics in Porous Media
Applications in Petroleum Geosciences and Petroleum
Engineering Mikhail Panfilov
https://textbookfull.com/product/physicochemical-fluid-dynamics-
in-porous-media-applications-in-petroleum-geosciences-and-
petroleum-engineering-mikhail-panfilov/
https://textbookfull.com/product/diffusion-nmr-of-confined-
systems-fluid-transport-in-porous-solids-and-heterogeneous-
materials-rustem-valiullin/
https://textbookfull.com/product/how-to-publish-in-
biomedicine-500-tips-for-success-third-edition-john-dixon/
https://textbookfull.com/product/are-some-languages-better-than-
others-robert-m-w-dixon/
https://textbookfull.com/product/australia-s-original-languages-
an-introduction-r-m-w-dixon/
SPRINGER BRIEFS IN
APPLIED SCIENCES AND TECHNOLOGY
John M. Dixon
Francis A. Kulacki
Mixed Convection
in Fluid
Superposed
Porous Layers
123
SpringerBriefs in Applied Sciences
and Technology
Extensive research has been done in the field of natural, mixed, and forced
convection in a porous layer. Several studies have investigated natural and forced
convection in a system that includes a porous and a superposed fluid layer, but
mixed convection has not been addressed. The present monograph is motivated to
fill the gap in the literature regarding mixed convection.
We investigate mixed convective heat transfer in a long channel that is partially
filled with a porous layer and has a fluid layer above it. The channel is heated on the
bottom over a finite length with cross flow along the length of the channel. The two
sublayers are treated as a single domain numerically, and the porosity is used as a
switching parameter, causing the governing conservation equations to transition
from an extended form of the Darcy-Brinkman-Forchheimer equation in the porous
sublayer to the Navier-Stokes equations in the fluid sublayer. This methodology
avoids the need for boundary conditions at the interface between the two domains.
Dimensionless groups are varied and include the Péclet number, Rayleigh number,
the porous sublayer height, Darcy number, Prandtl number, and the conductivity
ratio between the solid and fluid phases. The impact of the various additional terms
in the extended form of Darcy’s law is also investigated.
The conductivity ratio, Darcy number, porous sublayer height, Rayleigh num-
ber, and Péclet number all have a strong effect on the overall Nusselt number, while
the Prandtl number, Brinkman effect, Forchheimer effect, and convective terms
have negligible effects on Nusselt numbers. A Péclet number is observed at which
the Nusselt number is a minimum and is shown to be proportional to the Rayleigh-
Darcy number, the product of the Rayleigh and Darcy numbers, and inversely
proportional to the porous sublayer height. This Péclet number is termed the
“critical Péclet number.” A critical porous sublayer height ratio is also observed
at which the Nusselt number is a minimum and is proportional to the Rayleigh-
Darcy number and inversely proportional to the Péclet number. Streamlines capture
the transition from the natural convection regime to the forced convection regime.
In the transition region, flow patterns have characteristics of both. Isotherms
v
vi Preface
capture the thermal plume above the heated wall and show the influence of cross
flow on the shape and character of the plume.
An experimental apparatus is designed in order to collect data over a similar
range of parameters explored numerically. The numerical results show good agree-
ment with the experimental data within the bounds of uncertainty. The experiments
confirm the presence of the critical Péclet number. However, they do not show the
same trends at intermediate porous layer heights. The effect of the dimensionless
porous sublayer height on the Nusselt number is shown to be small from 0.5 to 1.
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 General Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Effective Conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Interfacial Boundary Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 One-Domain Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.5 Mixed Convection in Saturated Porous Media . . . . . . . . . . . . . . . . . 8
1.6 Natural Convection in Porous Media . . . . . . . . . . . . . . . . . . . . . . . . 12
1.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 Mathematical Formulation and Numerical Methods . . . . . . . . . . . . . . 15
2.1 Solution Domain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 Governing Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 One-Domain Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Numerical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1 Verification of Solution Method . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.1 Rayleigh-Bénard Convection . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.2 The Horton-Rogers Lapwood Problem . . . . . . . . . . . . . . . . . 21
3.1.3 Natural Convection in Fluid-Superposed Porous Layers . . . . . 24
3.1.4 Mixed Convection in a Porous Layer . . . . . . . . . . . . . . . . . . 26
3.2 Mixed Convection in a Fluid-Superposed Porous Layer . . . . . . . . . . 27
3.2.1 Variation of Thermal Conductivity . . . . . . . . . . . . . . . . . . . . 27
3.2.2 Inclusion of Brinkman and Forchheimer Terms . . . . . . . . . . . 30
3.2.3 Effect of Prandtl and Darcy Numbers . . . . . . . . . . . . . . . . . . 32
3.2.4 Sublayer Critical Rayleigh Number . . . . . . . . . . . . . . . . . . . 34
3.2.5 Critical Péclet Number . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.6 Transition from Natural to Forced Convection . . . . . . . . . . . 37
3.2.7 Critical Sublayer Height . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
vii
viii Contents
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Nomenclature
a Area (m2)
A Heated wall area (m2)
CF Forchheimer coefficient, Eqn. (2.8)
cp Specific heat at constant pressure (J/kg K)
d Particle diameter (m)
Da Darcy number, K/H2
DH Channel hydraulic diameter
g Gravitational acceleration, (0, g) (m/s2)
h Heat transfer coefficient (W/m2K)
H Height (m)
k Thermal conductivity (W/mK)
K Permeability (m2)
Lh Length of heated wall (m)
Nu Nusselt number, hH/km
p Pressure (Pa)
p Dimensionless pressure, pH2/ρ0α2f
Pe Péclet number, U0H/αf
Pr Prandtl number, νf/αf
q Heat transfer (W)
q000 Energy generation (W)
Ra Rayleigh number, gβH3(Tw T0)/αfνf
t Time (s)
T Temperature (K)
U0 Volumetric flow/area (m/s)
v Fluid velocity, (u,v) (m/s)
v Dimensionless velocity, (u/U0,v/U0)
w Weight factor, Eqn. (1.22)
x Horizontal coordinate (m)
y Vertical coordinate (m)
ix
x Nomenclature
Greek Symbols
Subscripts
c Critical value
e Effective
f Fluid
fs Fluid-solid interface
h Heater
H Total height
m Average value
p Porous
s Solid
w Wall
0 Reference
Chapter 1
Introduction
and geometrical parameters defining the system. The problem space is limited to
heating along a finite length of the lower boundary of the porous sublayer and a cold
sink temperature on the upper surface of the fluid sublayer. Figure 1.1 shows the
general features of the problem domain.
Heat transfer and fluid flow in porous media has been studied for over 150 years.
Textbooks are available that provide clear derivations of the governing equations
and give a condensed review of the relevant literature. Nield and Bejan [2] provide
a critical review of literature through 2006, and Whitaker [3] derives various forms
of Darcy’s law through an exposition of volume-averaging methods. Approxi-
mately 200 papers in recent years have been published annually on the topic
[2]. In the following paragraphs, we establish the general background to of the
problem and point to several relevant prior investigations of the heat transfer
problem.
Darcy’s law was first established in 1856 and is given as:
K
v ¼ ∇p: ð1:1Þ
μf
This equation directly relates the bulk, or drift, velocity to the pressure gradient,
∇p; the fluid viscosity, μf; and the bulk permeability, K. This equation is simple and
easy to use, but it can clearly be seen that it does not allow for any variation in the
velocity profile. As a consequence, for a fixed pressure gradient, the velocity profile
is flat, thus ignoring any edge effects due to solid or free boundaries.
The Forchheimer term was later added to account for inertial effects in the flow
[4]. The proper definition of the constant on the Forchheimer term, CF, has been
debated [2, 5], but the general form is:
μf CF
∇p ¼ v 1=2 ρf jvjv: ð1:2Þ
K K
1.1 General Considerations 3
The Forchheimer term is also known as the quadratic drag term and becomes
significant at higher flow velocities. The pressure drop for flow through a restric-
tion, e.g., orifice, pipe fitting, or nozzle, is proportional to the square of the average
fluid velocity and including the Forchheimer term provides a more physically
realistic governing equation. However, it can safely be neglected at the typical
low flow velocities found in many porous media applications.
Brinkman [6] derived a new equation treating the problem as flow over a sphere,
resulting in what is commonly known as the Brinkman equation:
μf
∇p ¼ vþμ
e∇2 v, ð1:3Þ
K
K
v¼ ð∇p þ ρf gÞ: ð1:5Þ
μf
The Boussinesq approximation for the equation of state allows the density in the
buoyant term to vary linearly with temperature and is widely used when the
temperature variation is not too large. Carr and Straughan [8] consider water near
the freezing point and suggest using a quadratic function that allows for the
minimum density of water to occur at 4 C.
4 1 Introduction
The l.h.s. represents the transient and convective terms. The first term on the
r.h.s. is the pressure gradient, which is the driving force of any bulk flow. The
second term on the r.h.s. represents the linear drag forces, and the fourth term
represents the quadratic drag forces. The third term accounts for the edge effects in
the flow, and without it, the basic velocity profile would be uniform. The fifth term
accounts for the buoyancy effects.
The energy equation for a porous medium is much less debated than the
momentum equation. However, there are several aspects to consider owing to the
two-phase nature of a porous medium. The basic energy equations are:
∂T s
ð1 ϕÞρs cp, s ¼ ð1 ϕÞ∇ ðks ∇T s Þ þ ð1 ϕÞqs 000 ð1:7Þ
∂t
∂T f
ϕρf cp, f þ ρf cp, f v ∇T f ¼ ϕ∇ ðkf ∇T f Þ þ ð1 ϕÞqf 000 , ð1:8Þ
∂t
where Eqn. (1.7) represents the temperature of the solid phase and Eqn. (1.8) the
temperature of the fluid phase.
Heat transfer between the solid and fluid phases occurs at the interface and has
been handled in different ways [2, 3], but the results are most simply explained by
defining an effective heat transfer coefficient between them and altering Eqns. (1.7)
and (1.8) to account for the interphase transfer:
∂T s
ð1 ϕÞρs cp, s ¼ ð1 ϕÞ∇ ðks ∇T s Þ þ ð1 ϕÞqs 000 þ hfs ðT f T s Þ ð1:9Þ
∂t
∂T f
ϕρf cp, f þ ρf cp, f v ∇T f ¼ ϕ∇ ðkf ∇T f Þ þ ð1 ϕÞqf 000
∂t
þ hfs ðT s T f Þ, ð1:10Þ
where volumetric energy sources in the solid and fluid phases are formally
included. Determining the values of the interphase heat transfer coefficient, hfs,
remains a challenge, but they were experimentally determined by Polyaev [9] and
shown to be in general agreement with the theoretical work of Dixon and Cresswell
[10]. For a bed of particles, the following formulas define the heat transfer
1.2 Effective Conductivity 5
coefficient [2], where afs is the surface area per unit volume, d is the particle
diameter, and B ¼ 10 for spherical particles:
where the particle Reynolds number, Red, is defined in terms of the particle
diameter, d, drift velocity in the porous medium, |v|, and fluid viscosity, μf. Other
models [11] have been shown to produce good agreement with measurement but are
omitted from the present review.
Thorough analysis of interface heat transfer has been conducted, and volume
averaging techniques have been employed, resulting in criteria for local thermal
equilibrium [3, 7]. In many applications it is often appropriate to make the assump-
tion of local thermal equilibrium. The criteria are always met in steady state and
reduce the equations to a single governing equation:
∂T
ρm c p , m þ ρf cp, f v ∇T ¼ ∇ ðkm ∇T Þ þ qm 000 ð1:15Þ
∂t
Characterizing the heat transfer and fluid flow characteristics in a porous medium
requires large-scale averaging of the flow. In the averaging process, pore level
physics is lost and is added back in a variety of ways. For example, consider a
porous medium of glass spheres filled with stagnant water and heated at the bottom.
Heat is transferred into the water and into the glass spheres, and the overall
conductivity is an average of the conductivities of the water and the glass spheres.
6 1 Introduction
The resulting value depends on the amount of each of the phases (porosity), on the
arrangement of the solid phase, and on the conductivities of each of the two phases.
When there is fluid motion, Darcy’s law provides the bulk flow velocity, which may
appear very laminar and orderly, but at the pore level, the flow behaves much
differently. Every particle that is encountered splits the flow effectively dispersing
it. The flow path is therefore a tortuous path. In laminar flow in fluids, heat transfer
perpendicular to the flow occurs strictly through conduction in the fluid, but flow in
porous media much more effectively transfers heat in the direction perpendicular to
the flow due to tortuosity and dispersion [12]. Dispersion and tortuosity not only
affect heat transfer and effective thermal conductivity but also momentum transfer
or effective viscosity.
For the most general case of a non-isotropic porous medium with varying flow
direction, a full dispersion tensor is used [3], defining unique values for each
direction. For simpler cases, a longitudinal dispersion and a transverse dispersion
are separately defined [2, 13], with longitudinal dispersion in the direction of
streamlines and transverse dispersion perpendicular to streamlines. However, for
easy implementation into standard numerical and analytical schemes, a single,
effective value is preferred. Many authors have approached this problem and
have developed correlations and theoretical predictions for a large range of values.
Various modeling approaches and measurement techniques have been devel-
oped for estimating effective thermal conductivity [5, 14], and further review will
not be considered at this point. Generally, the effective thermal conductivity is a
function of the solid-fluid conductivity ratio; the porosity; the structure of the solid
phase, e.g., foam metal matrix, monodisperse spherical particles, polydisperse
miscellaneous shapes, etc.; and the geometry of the solid-fluid interface. For
κ ~ 1, most of the methods produce similar results, and there is a negligible
difference between predictions within the uncertainty of most measurement
methods.
It is important to properly account for the effects of dispersion due to bulk
motion of the fluid. Combining the existing experimental data, correlations have
been developed [15] that provide the longitudinal and transverse dispersion coef-
ficients, thus increasing the effective diffusivity in both directions. Depending on
the geometry of the porous medium and the local Péclet number, transverse
dispersion can either approximately equal longitudinal dispersion or be up to
100 times greater [16]. At high flow rates, most of the longitudinal heat transfer
is due to advection, and transverse dispersion can be neglected.
Investigation of the boundary condition between a porous layer and a fluid layer
was conducted by Beavers and Joseph [17]. At the time of their research, the most
basic form of Darcy’s law was being used, providing a flat velocity profile in the
porous region, and a no-slip velocity profile was used at the interface between the
1.3 Interfacial Boundary Conditions 7
porous region and the fluid, fixing the edge velocity in the fluid to be equal to that of
the porous region. However, experimental data for Poiseuille flow over a porous
block revealed higher than expected flow rates in the porous region due to a
penetration of fluid flow. The penetrating flow creates a slip velocity at the
interface. The analytical expressions is:
∂u γ
¼ 1=2 ðu um Þ, ð1:19Þ
∂y K
where um is the mean fluid velocity above the porous layer and γ is an empirical
constant. This expression provides a simple and elegant way of calculating the
velocity profile but relies on the empirical constant to assure evaluation of the slip
velocity. However, the Brinkman term in Darcy’s law can be used instead of the
Beavers-Joseph boundary condition to achieve the same result [18].
Jones [19] built on the work of Beavers and Joseph and considers flow over a thin
spherical shell. He assumes the slip velocity was based on the shear stress, resulting
in an equation that considers both horizontal and vertical velocity components:
∂u ∂v γ
þ ¼ 1=2 ðu um Þ: ð1:20Þ
∂y ∂x K
Mixed convection in porous media has been investigated by many authors over the
years. This research has provided a great deal of insight into the present work, even
though the specific problem of mixed convection in a horizontal porous layer with a
superposed fluid layer has not been addressed in the literature. In addition, studies
of natural convection in porous layers, particularly in porous layers with an
overlying fluid layer, provide valuable insight.
Wooding [29] first studied mixed convection in porous media, analyzing the
effect of large Reynolds number on the convection characteristics of the system.
Prats [30] continued the theoretical study, considering the effect of cross flow on the
convection currents. Sutton [31] revisited the classical onset of convection problem,
but included a net cross flow, and was the first to discover that the critical Rayleigh
number increases with the increasing cross flow. As the flow rate approaches zero,
the critical Rayleigh number approaches the theoretical limit, 4π 2. This result was
confirmed by Homsy and Sherwood [32].
Rubin [13] investigated the effect of heat dispersion due to bulk fluid motion,
allowing for different thermal diffusivity in the transverse and longitudinal
1.5 Mixed Convection in Saturated Porous Media 9
directions. His theoretical results further confirm the finding that the critical Ray-
leigh number increases with an increasing Péclet number.
Combarnous and Bia [33] experimentally studied mixed convection in horizon-
tal layers at low Péclet number. Their data do not show any change in the onset of
convection due to cross flow. Reda [34] conducted mixed convection experiments
in a vertical chamber around a cylindrical heat source with the flow opposing the
buoyancy induced upward motion. The parameter Ra/Pe characterizes the heat
transfer of the system, and the transition from mixed to forced convection occurs
at Ra/Pe ~ 0.5, independently of the size of the heater or the amount of total power.
Lai and Kulacki [35] investigated mixed convection in porous media analyti-
cally, numerically, and experimentally in horizontal porous layers with localized
heating from below over a range of Péclet and Rayleigh numbers. Nusselt numbers
show steady behavior in the buoyancy-dominated regime and steady behavior in the
forced flow regime but oscillatory behavior in the transition region between natural
and forced convection (Fig. 1.2). This oscillatory behavior is initially shown to
stabilize, but later findings disagreed, showing ongoing oscillations with period:
τp ¼ 2:41Pe1:096 : ð1:21Þ
Vertical flows in porous media with a finite length wall heat source have also
been considered [36–38]. Numerical solutions were obtained using the basic form
of Darcy’s law with the buoyancy term included. In the opposed flow case, the
results show a Péclet number (denoted as the critical Péclet number) that results in a
minimum heat transfer rate, but no minimum was shown to occur in aided flow. The
same problem was later considered but includes the Forchheimer term and the
15 15
Pe Pe= 8
Nu 10 Nu 10
1
0.1 5
5
5 0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
τ τ
15 15
Pe = 2 Pe = 10
Nu 10 Nu 10
5 5
0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
τ τ
15 15
Pe = 4 Pe = 12
Nu 10 Nu 10
5 5
0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
τ τ
15 15
Pe = 6 Pe = 18
Nu 10 Nu 10
5 5
0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
τ τ
Fig. 1.2 Nusselt number oscillations in mixed convection in a porous layer [35]
10 1 Introduction
I Entrance Section
II Test Section
III Exit Section
IV Cooling Chamber
Fig. 1.3 Apparatus for mixed convection in a saturated porous heated from below [39]
for the numerical results. In Eqns. (1.22)–(1.25), the length scale is the height of the
porous layer. A critical Péclet number was numerically confirmed, and the mea-
surements showed it becomes more prominent as the length of the heated section
exceeds the total layer height. Similar experiments were then reported [40] with a
sudden expansion at the beginning of the flow channel just prior to the heated wall
section. A critical Péclet number is also observed, but inclusion of non-Darcy terms
had a negligible effect even beyond Pe ~ 100.
A series of papers by various authors investigate mixed convection about a line
source in an infinite saturated porous domain. Pop et al. [41] develop a similarity
solution using the Darcy’s law with the buoyancy terms included. Jang and Shiang
[42] use an implicit finite difference method to numerically solve the DBF equation.
Nakayama [43] divides the flow regime into five distinct regions and presents
similarity solutions for each and shows where the addition of the Forchheimer
and buoyancy terms is necessary. Zhou and Lai [44] numerically solve the vertical
flow problem for Darcy flow over a heated cylinder for Reynolds number from 10 to
100 and Grashof number from 0 to 400. Oscillatory behavior is again observed for
opposed flow, and streamlines and isotherms showing the transient behavior were
presented.
Kumar et al. [45] considered mixed convection with both a fluid layer and a
fluid-saturated porous layer in vertical flow. They investigated isothermal and
isoflux boundary conditions and used the DBF equation with the buoyancy term,
algebraically solving for the steady-state velocity and temperature profiles in a
semi-infinite domain. Their solutions cover a range of viscosity ratio, conductivity
ratio, Darcy number, Grashof number, and width ratios, employing continuity and
no-slip boundary conditions. Although they were the first to consider the
two-domain system in mixed convection, the results are of limited value in the
present context owing to exceedingly high Darcy numbers, 0.125–0.5.
Bhargavi [46] studied forced convection in a porous layer with a superposed
fluid layer. He considers two-dimensional flow and solves the governing equations
for fully developed conditions in an infinite channel. Entrance effects are also
considered. Various boundary conditions are considered for the heated wall and
top surfaces, and Ochoa-Tapia-Whitaker’s stress-jump condition is applied at the
interface between the porous and fluid regions. Results from the Darcy equation
are compared to those from the Darcy-Brinkman equation and to those from
the Forchheimer equation, showing significant differences at some conditions.
12 1 Introduction
In addition, the effects of viscous dissipation were considered, including five forms
of the dissipation function. Within a certain range of Darcy number (0.1–0.001), a
peak Nusselt number occurs at a dimensionless porous sublayer height of 0.6–0.8.
Elder [47] first studied natural convection from a localized heat source, numerically
solving the problem with a finite difference method. The domain considered is
similar to that for mixed convection studies, being much longer than tall, with an
aspect ratio of 10. Streamlines and isotherms are presented for a range of heater
sizes and Rayleigh numbers. With a dimensionless heater length of δ ¼ 1, the fluid
plume rises in a single narrow column. With δ ¼ 2, two recirculating regions form,
rotating in opposite directions. With δ ¼ 3, a pair of closed cells are formed and at
δ ¼ 6, three pairs of cells. This same problem of localized heating has been
investigated by Prasad and Kulacki [48, 49]. They solve the problem numerically
over a range of aspect ratios and dimensionless heater lengths. At a fixed heater
size, it is shown that changing the aspect ratio has minimal effect on the overall heat
transfer coefficient. At small dimensionless heater lengths (δ < 2), recirculation due
to natural convection is always unicellular, but for δ > 2 recirculation becomes
multicellular.
Bagchi [1] extensively reviews the literature on natural convection in a porous
medium with an overlying fluid layer as a background for his study with a heated
segment on the bottom (Fig. 1.4). He used porosity as a switching parameter in the
DBF equation to produce the Navier-Stokes equation in the fluid sublayer. However
derivatives of the porosity are not considered to account for the rapid change at the
interface. Various heater sizes, dimensionless porous sublayer height, Rayleigh
number, Prandtl number, Darcy number, conductivity ratio, and overall aspect
ratio are investigated. Numerical results showed excellent agreement with the
classical solutions for the Rayleigh-Bénard and Horton-Rogers-Lapwood problems.
As the porous sublayer height is decreased (η < 1), the strength of the thermal
plume greatly increases in the fluid region, and circulating flow was primarily
contained in the fluid region as well. It is shown that increasing the Darcy number
increases the heat transfer coefficient due to a decreased restriction of bulk fluid
motion, that the Prandtl number had little effect on the overall heat transfer, and that
the conductivity ratio has a large impact due to an increase in convective motion.
Disagreement between numerical and experimental results pointed to either the
need for refinement of the numerical solution or a source of error in the
experiments.
1.7 Conclusion
The development of the governing equations has been discussed, historical work
considering the interfacial boundary conditions has been reviewed, various
approaches to the effective thermal conductivity have been explored, and applica-
ble research on natural and mixed convection in porous media have been reviewed.
The current state of research clearly favors the one-domain approach using porosity
as a switching parameter to avoid explicit definition of the boundary condition at
the interface. This method leaves open the question of porosity variation near the
interface. However, there is no established method to properly transition from the
porous layer to the fluid layer, and appropriate methods need to be determined.
Determining the effective thermal conductivity of a porous medium has been a
challenge, and there is an abundance of research on the subject. The effective
thermal conductivity becomes somewhat of a catchall to include various other
effects as well, such as dispersion and tortuosity. The research on stagnant thermal
conductivity provides many correlation equations that have a great deal of variation
at high conductivity ratio but general convergence at low conductivity ratios. The
effects of dispersion, tortuosity, etc., can easily be included by using readily
available correlation equations that correct for these effects.
Natural convection in porous media, particularly in a porous layer with an
overlying fluid layer, has been thoroughly addressed in the literature. Mixed
convection, on the other hand, has not. There are many investigations of mixed
convection that consider a full porous domain, but the literature is lacking on mixed
convection in a porous layer with a superposed fluid layer.
Chapter 2
Mathematical Formulation
and Numerical Methods
The general features of the solution domain and inlet velocity profile are shown in
Fig. 1.1. The top of the flow channel is held at a constant temperature, T0, and has a
no-slip velocity boundary condition. The bottom also has a no-slip boundary and
has a constant temperature, Tw, on a finite length, Lh. The bottom is otherwise
insulated. The inlet temperature is also at T0. Far upstream and downstream, the
inlet and outlet are also taken as adiabatic (∂T/∂x ! 0) and present two options for
the velocity distribution. For the natural convection limit, no-slip velocity boundary
conditions are applied at the left and right boundaries, and for mixed convection, a
fixed velocity profile based on steady-state conditions is applied at the inlet and
outlet.
Equations governing the flow and the temperature for each sublayer reduce to a
single set of equations as porosity transitions from ϕ ¼ ϕ0 to ϕ ¼ 1 when the
solution shifts from the porous to the overlying fluid sublayer. For incompressible
flow, the governing equations for the porous sublayer are the continuity equation,
the DBF equations with the addition of the buoyancy term and transient and
convective terms, and the volume-averaged energy equation.
∂u ∂v
þ ¼ 0: ð2:1Þ
∂x ∂y
∂
v
1 2 ∂
v ∂
v
ϕ Pe þ ϕ Pe u þ v
2
∂τ ∂
x ∂
y
!
2 2
∂
p PrPe ∂ v ∂ v
¼ v þ μ
bPrPe þ 2 ð2:6Þ
∂
y Da ∂
x2 ∂y
CF Pe2 gH 3
pffiffiffiffiffiffi v þ 2 þ PrRaθ:
Da αf
The r.h.s. of Eqns. (2.5) and (2.6) contain four to six terms, the first being the
pressure gradient which is the driving force for forced flow. The second term is the
Darcy term and is generally the dominant term in most natural and mixed convec-
tion applications due to the low values of the Darcy number. The third term is the
Brinkman term which captures the viscous effects due to boundaries within the
porous domain. The fourth term is the Forchheimer term which can become the
dominant term at large flow velocities. The transition from the Darcy regime to the
Forchheimer regime will vary depending on the Darcy number and the flow
velocity. The sixth term in Eqn. (2.6) captures the effects of buoyancy, causing
flow to accelerate upward when the temperature is elevated above that of the free
stream.
The energy equation is multiplied by H2/αf, and the resulting dimensionless is:
∂θ ∂θ ∂θ ∂ ∂θ ∂ ∂θ
σ þ Pe
u þ Pe
v ¼ λ þ λ : ð2:7Þ
∂τ ∂
x ∂
y ∂ x ∂
x ∂
y ∂
y
This equation neglects viscous dissipation and does not include volumetric
energy generation.
The momentum equations are next reduced to the stream function—vorticity
form to eliminate the pressure gradient. The partial derivatives of Eqn. (2.5) with
respect to y and Eqn. (2.6) with respect to x are taken to derive the variable porosity
form of the equations. The Darcy number is a function of the porosity and therefore
varies in the y-direction.
Although the Forchheimer constant is written as a function of porosity by some
authors, the more widely accepted form is not a function of porosity but rather the
bead diameter and the hydraulic diameter, DH, of the channel [2]:
d
CF ¼ 0:55 1 5:5 : ð2:8Þ
DH
18 2 Mathematical Formulation and Numerical Methods
With the over bars dropped, the dimensionless momentum equations are written as:
∂uy
ϕ1 Pe þ ϕ2 Pe2 uuxy þ uy ux þ vuyy þ vy uy
∂τ
Dauy uDay
¼ Pxy PrPe 2
þμbPrPe uxxy þ uyyy ð2:9Þ
Da
pffiffiffiffiffiffi 1 1
Da jvjuy þ jvjy u jvju Da2 Day
CF Pe2 2
Da
∂vx
ϕ1 Pe þ ϕ2 Pe2 uvxx þ ux vx þ vvxy þ vx vy
∂τ
PrPe
¼ Pxy vx þ μ
bPrPe vxxx þ vxyy ð2:10Þ
Da
CF Pe2
pffiffiffiffiffiffi jvjvx þ jvjx v þ PrRaθx :
Da
Subtracting Eqn. (2.9) from Eqn. (2.10), the governing equation for the porous
sublayer is:
∂ω h i
ϕ1 Pe þ ϕ2 Pe2 ðuωÞx þ ðvωÞy
∂τ
¼μbPrPe ωxx þ ωyy
PrPe CF Pe2
þ pffiffiffiffiffiffi jvj ω
Da Da ð2:11Þ
jvjy u jvjx v
þ PrRaθx þ CF Pe2 pffiffiffiffiffiffi
Da
pffiffiffiffiffiffi
Day CF Pe2 Dajvj
þ u 2 PrPe þ :
Da 2
The dimensionless vorticity form of the Navier-Stokes and energy equations for
the fluid sublayer is, respectively:
∂ω h i
Pe þ Pe2 ðuωÞx þ ðvωÞy ¼ PrPe ωxx þ ωyy þ PrRaθx ð2:12Þ
∂τ
∂θ ∂θ ∂θ ∂ ∂θ ∂ ∂θ
þ Peu þ Pev ¼ þ : ð2:13Þ
∂τ ∂x ∂y ∂x ∂x ∂y ∂y
2.4 Numerical Methods 19
where 5 < N < 6 for ϕ0 ~ 0.4. This equation can be used to allow for either the
channeling effect near a solid boundary or as is done in the present case, porosity
variation at the interface between the porous and fluid sublayers. For the results
reported in this monograph, N ¼ 5.
To summarize, the porous region is governed by the DBF equations with the
addition of the buoyancy, transient, and convection terms, while the fluid region is
governed by the Navier-Stokes equations. However, the single set of equations
developed is suitable for both regions, provided the porosity is varied, resulting in a
single problem domain. The resulting equation contains additional terms, wherein
the effects of porosity variation are realized in the Darcy number for flow and the
conductivity variation in the energy equation.
Finite volume methods are chosen to implement the numerical solution of the
governing equations. Finite volume methods provide better conservation of energy
and mass than traditional finite difference methods across the finite difference grid.
In many of the derivative terms, central differencing is employed, and for the
convective terms, upwind differencing is employed. Although quadratic upwind
differencing has been used [1], standard upwind differencing is less problematic
owing to inherent instabilities in regions of sharp gradients. A variable grid is
chosen over a constant grid to allow coarser overall grid except in desired regions to
achieve a grid-independent solution with fewer total grid points [51, 52]. Second
order difference equations are used for most equations with the exception of
boundary conditions where the grid points were clustered, thus, achieving second
order accuracy with first order equations. Complete details of the numerical solu-
tions are given by Dixon [53].
Chapter 3
Numerical Results
Boys and girls, as a rule, reach the age of eight or nine, perhaps
ten, before any event of importance interrupts the even tenor of their
lives. Then the assembly of the men, which when the harvest is over,
meets daily in the baraza, decides where the unyago is to be
celebrated in the current year. Since all the adjacent districts have
now taken their turn in bearing the expense of the ceremony, it is a
point of honour that our village should invite them this time. The
resolution is soon carried into effect; the moon is already on the
wane, and the celebration must take place before the new moon. The
unyago presents exactly the same features in all the tribes of this
region. The men erect a circle—larger or smaller as circumstances
may require—of simple grass huts in an open space near the village.
In this space the opening and closing ceremonies are performed; the
huts are intended for the candidates to live and sleep in. Such an
arena, with all its appurtenances in excellent preservation, was the
circle of something over fifty yards’ diameter which I was enabled to
photograph when visiting the echiputu at Akuchikomu. The charred
remains of a similar lisakasa, as the system of huts is called in Yao,
were to be seen near the road on this side of Akundonde’s—the relics
of a former festival.