Analysis of the Flow at the Interface of a Porous Media

Thamy C. Hayashi
1,a
, Isabel Malico
2,3,b
and J. F. C. Pereira
3,c

1
Federal University of Bahia, Department of Chemical Engineering, R. Aristides Novis, 2,
Federação, 40210-630, Salvador, BA, Brazil
2
University of Évora, Department of Physics, R. Romão Ramalho, 59, 7000-671 Évora, Portugal
3
IDMEC/IST, Technical University of Lisbon, Department of Mechanical Engineering, Av. Rovisco
Pais, 1049-001 Lisbon, Portugal
a
thamy@ufba.br,
b
imbm@uevora.pt,
c
jcfpereira@ist.utl.pt
Keywords: Porous media, Incompressible flow, Turbulence, CFD.
Abstract. The influence of inserting ceramic foam in a pipe with a 1:4 sudden expansion was
numerical investigated. The foam, with a thickness to diameter ratio of 0.60, was positioned at
different distances from the sudden pipe expansion wall. Three different porosities were analyzed
(10, 20 and 60 pores per inch) for pore Reynolds numbers in the range of 20-400, corresponding to
pipe Reynolds numbers of 2400 to 22000 in the pipe section upstream the sudden expansion.
Predictions of the sudden pipe expansion cavity assuming laminar flow within the foam yield the
penetration of the separated flow region into the foam. Considering turbulent flow in the porous
foam and the model of Pedras and Lemos [14] prevents this penetration. The numerical and physical
models used could not reproduce completely the foam influence on the separated turbulent flow
region between the sudden pipe expansion and the foam inlet.
Introduction
The presence of a foam insert in a flow affects the flow field upstream. This influence is not well
characterized when the foam is located inside a turbulent separated flow region such as in the
vicinity of a sudden pipe expansion. Questions may rise if the separated flow penetrates or not
inside the foam. Prakash et al. [1,2] studied by LDV and numerical simulations impinging jets in a
cylindrical enclosure with and without a porous layer. They conclude that further investigations are
required for the validation of turbulent flow models for porous medium in order to be able to predict
the feedback influence on incoming turbulent separated flow.
Practical applications of fluid flow through ceramic foams are characterized by pore Reynolds
numbers that give rise to questions about the laminar or turbulent nature of the flow. Hall and Hiatt
[3] used LDV to measure the turbulence characteristics in the interior of ceramic foams. Their
measurements were conducted through a perforation made in samples with 10, 20 and 30 ppi. These
authors observed the flow characteristics for pore Reynolds number ranging from 20 to 200 and
found that the flow was essentially laminar for the 20 and 30 ppi foams. However, for 10 ppi, the
measured turbulence intensity was 60-80% at a pore Reynolds number of 150. Seguin et al. [4]
conducted experiments with synthetic foams with pore sizes in the range of 10 to 100 ppi.
Concerning the onset of fluctuations, they observed a laminarisation effect in the synthetic foams,
and no turbulent regime was detected in the range of pore Reynolds number investigated, up to
Re
p
=1100.
Numerical calculations of the 3D turbulent flow structure at pore level are still impracticable. An
alternative way is the description of the flow within a porous medium from a macroscopic point of
view, in which governing equations are obtained by volume averaging the microscopic 3D flow
equations over a representative elementary volume of the fluid saturated porous structure. This
procedure heavily relies on the pressure loss model, which is well established for the case of packed
beds of spherical particles and consists of the tuned Ergun model in the case of ceramic foams [5].
The accuracy of the multidimensional effective velocity predictions or of the occurrence of flow
Defect and Diffusion Forum Vols. 283-286 (2009) pp 616-621
Online available since 2009/Mar/02 at www.scientific.net
© (2009) Trans Tech Publications, Switzerland
doi:10.4028/www.scientific.net/DDF.283-286.616
All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of TTP,
www.ttp.net. (ID: 132.248.30.3-02/12/10,01:01:27)
separation inside foams has seldom been investigated. Moreover, modeling the occurrence of
transitional or turbulent regimes inside porous media is a controversial issue, as discussed, for
example, by Mickeley et al. [6], Masuoka and Takatsu [7] and Nield [8]. Moreover, if the porous
medium is immersed in a transitional or turbulent flow, further difficulties arise owing to the lack of
ability of the models to deal with the flow conditions at the inlet and outlet boundaries of the porous
insert. To the best of the authors’ knowledge, such difficulties have not been addressed.
In this work, isothermal flow inside alumina (Al
2
O
3
) foams placed on a 1:4 sudden pipe
expansion is considered. Three different pore sizes were examined: 10, 20 or 60 ppi, corresponding,
respectively, to 5.1, 2.6 and 0.85 mm mean pore diameters. In Pereira et al. [9], velocity
measurements were taken in the recirculation flow region, close to the foam inlet. Numerical
predictions of the flow field were performed, in which either laminar or turbulent flow in the porous
region was assumed. The flow rates and pore Reynolds numbers considered in this study were
chosen to correspond to the range of typical porous ceramic burners and heat exchangers. Having
this in mind, the pore Reynolds numbers investigated ranged from 20 to 400, which corresponds to
pipe Reynolds numbers of 600≤Re
D
≤5500 (2400≤Re
d
≤22000 in the pipe section upstream the
sudden expansion).
The main objective of this work is to analyze the influence of the laminar and turbulent
assumptions over the porous medium and their backward influence on the turbulent recirculation
flow region upstream the foam.
Geometry
A schematic drawing of the flow configuration under study is shown in Figure 1. Compressed air
enters a pipe 1:4 (20 to 84 mm) sudden expansion, containing a 50 mm thick ceramic foam insert in
the 84 mm diameter section. Three Al
2
O
3
foam samples, with 10, 20 or 60 pores per inch, were
studied. The mean pore diameters are 5.1, 2.6, and 0.85 mm, respectively. The ceramic foam sample
could be mounted at different distances from the expansion wall, b (40, 60 and 80 mm).
Details on the velocity measurements can be seen in [9].

Figure 1 - Schematic drawing of the of the flow configuration.
umerical Model
Governing Equations. The ceramic foam was modeled as a single continuum, taking the averaged
forms of the continuity and momentum equations the form [10]:

( )
f
0 φ ⋅ = u ∇ ∇∇ ∇
(1)
( ) ( ) ( )
f f f f
2
f
f f
1
P φ φ ν φ
ρ ρ
⋅ = − + +
r
u u u ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇
(2)

b
Defect and Diffusion Forum Vols. 283-286 617
where ρ
f
, ν
f
, are the density and kinematic viscosity of the fluid phase, respectively, φ the void
fraction (or porosity) and
f
u and
f
P the local intrinsic fluid-phase volume averages of the
velocity and pressure, respectively.
In Eq. 2, r stands for the body force term, which quantifies the resistance imposed by the solid
matrix to the fluid flow. This term was represented by a Darcy-Forchheimer extended model in the
form of the modified Ergun model [5]. This has been heuristically extended to three-dimensional
flows ([11], cit. in [12]):

f E
1
2
C
K
K
µ
ρ =− − r u u u
(3)

where µ
f
stands for the dynamic viscosity of the fluid, C
E
for the Ergun coefficient and the
permeability, K, is given by the Carman-Kozeny equation [12]:

( )
3
2
p 2
180 1
K d
φ
φ
=

(4)

where d
p
stands for the equivalent particles diameter.
The above equations can be used to simulate the flow when no porous medium is present. In this
case, the averaged forms of the momentum and continuity equations revert to the equations for the
clear flow when the pressure loss, expressed by Eq. 3, is dropped and the porosity φ is assigned the
unity value.
For the flow rates considered, flow in the pipe is turbulent and, consequently, the flow inside the
ceramic foam layer may or may not be turbulent. The eddy viscosity for the fluid flow is obtained
with the low Reynolds k-ε turbulence model by Launder and Sharma [13].

Turbulence Model. Being the flow turbulent in the upstream pipe section, it will either
relaminarise or remain turbulent inside the foam. These two possibilities were considered. When the
flow is assumed to be turbulent, the model of Pedras and Lemos [14] was used. This model is a k-ε
turbulence models, modified in order to include the effect of the presence of the solid structures on
the flow; it is based on taking the volume integration, within the fluid phase in a representative
elementary volume, of the locally defined turbulence quantities. As a result of the volumetric
averaging of the microscopic equations for the turbulent kinetic energy per unit mass, k, and
dissipation rate of turbulent kinetic energy per unit mass, ε, extra terms representing the transport
and production of turbulence due to the presence of the solid matrix are obtained. By means of
numerically solving the flow field within a spatially periodic array of elliptic rods, and considering
both low and high Reynolds flow situations, they obtained the model constant C
k
. The macroscopic
transport equations for the turbulence kinetic energy and for its dissipation rate are expressed as:

( ) ( )
t f f
f f
k
f
f f
f k f
' ' : C
k k
k
K
φ
µ
ρ µ φ
σ
ρ ρ φ ρφ ε
( | |
⋅ = ⋅ + ( |
|
(
\ ¹ ¸ ¸
− ∇ + −
u
u
u u u
∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇
(5)
618 Diffusion in Solids and Liquids IV
( ) ( )
( )
( )
t f f
f f
2
f
f f
f
1 f 2 f k f f
C ' ' : C C
K k k
φ
µ
ρ ε µ φ ε
σ
ε
ε ε
ρ ρ φ
ε
ε ε
( | |
⋅ = ⋅ +
( |
|
(
\ ¹
¸ ¸
¦ ¹
¦ ¦
+ − ∇ + −
´ `
¦ ¦
¹ )
u
u
u u u
∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇ ∇
(6)
with
f f
f t f
2
' ' : 2
3
k
φ
ρ µ φρ − ∇ = − u u u D I
(7)
( )
2
f
t f f
C
k
φ
µ ρ
ε
µ
=
(8)

where
t
φ
µ is the dynamic eddy viscosity based on intrinsic averages of turbulence quantities, σ
k
, σ
ε
,
C
k
, C

, C

, and C
µ
are standard k-ε model constants.
All the calculations assumed fully developed turbulent pipe flow at the inlet, and the low
Reynolds k-ε model by Launder and Sharma [13] was used to predict the flow in the clear fluid pipe
sections.

umerical Method. A finite volume method and the SIMPLE algorithm [15] were used. The
convection terms were discretized with a deferred correction in which central differences contribute
80% and first order accurate upwind differences 20%. Higher order schemes for the discretization of
the convective terms (see e.g. Kobayashi and Pereira [16]) were not required, since a fine mesh was
used. Moreover, it is well known that this particular sudden pipe expansion flow problem presents
very low velocities in the regions of strong curvature of the streamlines while high velocities are
only found in regions of quasi one-dimensional flow. As a consequence, false diffusion is prevented
by using sufficiently fine meshes and this particular flow may be well predicted with the first order
upwind scheme and the k-ε turbulence model [17]. In order to study the dependence of the results on
the grid refinement, calculations of laminar flow in the pipe with a sudden expansion were obtained
using seven regular meshes with a coarse-to-fine refinement factor of 1.5. The results obtained with
the two finer meshes (230×116 and 344×173 grid points) were virtually identical and therefore all
the results presented for this geometry were obtained with the mesh with 230×116 grid nodes.
Results
Figures 2a and 2b show predicted streamlines and turbulence kinetic energy contours assuming,
respectively, laminar flow within the porous medium and using the turbulence model of Pedras and
Lemos [14]. No recirculation could be detected downstream the foam exit for all the calculation (50
mm thick, 10, 20 and 60 ppi samples).
From these two figures, it is observed that the k-ε model of Pedras and Lemos [14] predicts the
promotion of the turbulence, and, therefore, if this model is used, the results differ from the
predictions where the laminar model is used. In the results obtained with the model by Pedras and
Lemos [14] the enhancement of the diffusivity owing to the turbulence almost eliminates the
laminar separated flow region inside the foam at the expansion wall. Unfortunately, flow inside the
foam could not be visualized, nor could any measurements be taken as to be able to confirm the
existence of flow separation. The outflow from the foam resembles the streamlines shown in Figs.
2a and 2b.

Defect and Diffusion Forum Vols. 283-286 619

Figure 2 - Streamlines and turbulence kinetic energy contours predicted for flow within foam
positioned at pipe sudden expansion, (a) assuming laminar flow in the foam; (b) using the model of
Pedras and Lemos [14] (20 ppi, Re
p
=115, dashed line indicates foam outlet).

Calculations were also performed for several cases where the foam sample is located at a finite
distance from the sudden pipe expansion. It has been observed that this gives rise to the formation
of an axisymmetrical cavity between the expansion wall and the foam sample. In this region, the
flow displays the velocity characteristics of a typical cavity and experiments with smoke flow
visualization were not conclusive about penetration of the separated flow streamlines into the foam.
Results taken 10 mm upstream the foam inlet for a cavity lengths, b, equal to 60 mm are shown
in Fig. 3. They show the formation of a toroidal vortice flow in the cavity space.
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
0.000 0.015 0.030 0.045
r (m)
U
m
/
U
C
L
Laminar
Pedras and Lemos [14]
Pereira et al. [9]

Figure 3 - Comparison of predicted and measured mean axial velocity in a 60 mm long cavity flow.
Results taken 10 mm upstream foam inlet (50 mm thick 20 ppi foam, Re
p
=115).

Figure 3 shows a satisfactory agreement between the experiments and calculations of the
stream-wise velocity profile on a plane positioned 10 mm upstream the foam inlet for the impinging
jet flow around the centerline. However, in the separated flow region, both its size and the velocity
magnitudes differ significantly. The experimental results show a smaller region with negative axial
velocities close to the pipe wall, where the velocities are relatively low compared to the velocity at
the centerline jet-like region.
Summary
Calculations were conducted to investigate the influence of a ceramic foam insert on the flow inside
a pipe with a 1:4 sudden expansion. The main conclusions can be summarized as follows: i)
Predictions of the sudden pipe expansion cavity assuming laminar flow within the foam yield the
penetration of the separated flow region into the foam. The predictions obtained with the model by
Pedras and Lemos [14] do not show the penetration of the flow recirculation inside the porous
matrix. Predictions and measurements of the mean axial velocities for the cavity flow show very
620 Diffusion in Solids and Liquids IV
large qualitative differences close to the pipe wall. ii) The influence of assuming laminar or
turbulent flow within the foam over the flow upstream the foam is limited to the influence of the
pressure loss within the foam. Nevertheless, the use of laminar or turbulent assumptions has been
shown to have a large qualitative impact in the vicinity (within about five pore diameters distance)
of the foam inlet.
Acknowledgements
This work was developed within the framework of project POCI/EME/59510/2004, which is
financially supported by FCT-Fundação para a Ciência e a Tecnologia, programme POCI 2010
(29.82% of the funds from FEDER and 70.18% from OE).
References
[1] M. Prakash, O.F. Turan, Y. Li, J. Mahoney and G.R. Thorpe: Chem. Eng. Sci. Vol. 56 (2001),
p. 3855.
[2] M. Prakash, O.F. Turan, Y. Li, J. Mahoney and G.R. Thorpe: Chem. Eng. Sci. Vol. 56 (2001),
p. 3879.
[3] M.J. Hall and J.P. Hiatt: Exp. Fluids Vol. 20 (1996), p. 433.
[4] D. Seguin, A. Montillet, J. Comiti and F. Huet: Chem. Eng. Sci. Vol. 53 (1998), P. 3751.
[5] I.F. Macdonald, M.S. El-Sayed, K. Mow and F.A.L. Dullien: Ind. Eng. Chem. Fund. Vol. 18
(1979), p. 199.
[6] H.S. Mickeley, K.A. Smith and E.I. Korchak: Chem. Eng. Sci. Vol. 23 (1965), p. 237.
[7] T. Masuoka and Y. Takatsu: Int. J. Heat Mass Tran. Vol. 39 (1996), p. 2803.
[8] D.A. Nield: Int. J. Heat Mass Tran. Vol. 40 (1997), p. 2499.
[9] J.C.F. Pereira, I. Malico, T.C. Hayashi and J.M.F. Raposo: Exp. Therm. Fluid Sci. (Submitted)
[10] C.T. Hsu and P. Cheng: Int. J. Heat Mass Tran. Vol. 33 (1990), p. 1587.
[11] V. Stanek and J. Szekely: AIChE J. Vol. 20 (1974), p. 974.
[12] M. Kaviany: Principles of Heat Transfer in Porous Media (Springer-Verlag, New York, 1991).
[13] B.E. Launder and B.I. Sharma: Lett. Heat Mass Tran. Vol. 1 (1974), p. 131.
[14] M.H.J. Pedras and M.J.S de Lemos: J. Fluids Eng. Vol. 123 (2001), p. 941.
[15] S.V. Patankar: umerical Heat Transfer and Fluid Flow (Hemisphere, Washington DC, 1980).
[16] M.H. Kobayashi and J.C.F. Pereira: Commun. Numer. Meth. Eng. Vol. 12 (1996), p. 395.
[17] W.H. Stevenson and H.D. Thompson: AIAA J. Vol. 28 (1990), p. 276.
Defect and Diffusion Forum Vols. 283-286 621
Diffusion in Solids and Liquids IV
doi:10.4028/www.scientific.net/DDF.283-286

Analysis of the Flow at the Interface of a Porous Media
doi:10.4028/www.scientific.net/DDF.283-286.616

In this work. Compressed air enters a pipe 1:4 (20 to 84 mm) sudden expansion. were studied. 2.85 mm. further difficulties arise owing to the lack of ability of the models to deal with the flow conditions at the inlet and outlet boundaries of the porous insert. 20 or 60 ppi. corresponding. Three Al2O3 foam samples. modeling the occurrence of transitional or turbulent regimes inside porous media is a controversial issue.1.1.Defect and Diffusion Forum Vols. Masuoka and Takatsu [7] and Nield [8].6. Details on the velocity measurements can be seen in [9]. Three different pore sizes were examined: 10. In Pereira et al. 60 and 80 mm). The main objective of this work is to analyze the influence of the laminar and turbulent assumptions over the porous medium and their backward influence on the turbulent recirculation flow region upstream the foam.6 and 0.Schematic drawing of the of the flow configuration. Numerical predictions of the flow field were performed. for example. b (40. close to the foam inlet. with 10. isothermal flow inside alumina (Al2O3) foams placed on a 1:4 sudden pipe expansion is considered. 20 or 60 pores per inch. and 0. [6]. taking the averaged forms of the continuity and momentum equations the form [10]: ( )=0 1 ∇ ⋅ (φ u u ) = − ∇ ( φ ρ ∇⋅ φ u f f f f (1) P f ) + ν ∇ (φ u ) + ρ 2 f f r f (2) . To the best of the authors’ knowledge. such difficulties have not been addressed. the pore Reynolds numbers investigated ranged from 20 to 400. containing a 50 mm thick ceramic foam insert in the 84 mm diameter section. 2. respectively. umerical Model Governing Equations. The flow rates and pore Reynolds numbers considered in this study were chosen to correspond to the range of typical porous ceramic burners and heat exchangers. respectively. The ceramic foam sample could be mounted at different distances from the expansion wall. b Figure 1 . The ceramic foam was modeled as a single continuum. Moreover.85 mm mean pore diameters. in which either laminar or turbulent flow in the porous region was assumed. Geometry A schematic drawing of the flow configuration under study is shown in Figure 1. Moreover. as discussed. by Mickeley et al. if the porous medium is immersed in a transitional or turbulent flow. Having this in mind. to 5. The mean pore diameters are 5. 283-286 617 separation inside foams has seldom been investigated. velocity measurements were taken in the recirculation flow region. [9]. which corresponds to pipe Reynolds numbers of 600≤ReD≤5500 (2400≤Red≤22000 in the pipe section upstream the sudden expansion).

expressed by Eq. and considering both low and high Reynolds flow situations. respectively. CE for the Ergun coefficient and the permeability. The macroscopic transport equations for the turbulence kinetic energy and for its dissipation rate are expressed as: ρf ∇ ⋅ u k ( f ) = ∇ ⋅  µ + σ  f  µ tφ  k ∇ φ k   ( f )   − ρf u ' u ' : ∇ u + Ck ρf φ f k f u K − ρφ ε f (5) . the averaged forms of the momentum and continuity equations revert to the equations for the clear flow when the pressure loss. In this case. This model is a k-ε turbulence models. Being the flow turbulent in the upstream pipe section. modified in order to include the effect of the presence of the solid structures on the flow. When the flow is assumed to be turbulent. in [12]): r =− µf K u − CE K 1 2 ρ u u (3) where µf stands for the dynamic viscosity of the fluid. Turbulence Model. extra terms representing the transport and production of turbulence due to the presence of the solid matrix are obtained. the model of Pedras and Lemos [14] was used. As a result of the volumetric averaging of the microscopic equations for the turbulent kinetic energy per unit mass. ε. k. In Eq. By means of numerically solving the flow field within a spatially periodic array of elliptic rods. For the flow rates considered. respectively. they obtained the model constant Ck. consequently. K. which quantifies the resistance imposed by the solid matrix to the fluid flow. and dissipation rate of turbulent kinetic energy per unit mass. are the density and kinematic viscosity of the fluid phase. within the fluid phase in a representative elementary volume. φ the void fraction (or porosity) and u f and P f the local intrinsic fluid-phase volume averages of the velocity and pressure. 2. 3. is dropped and the porosity φ is assigned the unity value. it will either relaminarise or remain turbulent inside the foam. flow in the pipe is turbulent and. the flow inside the ceramic foam layer may or may not be turbulent. cit. This has been heuristically extended to three-dimensional flows ([11]. νf. The eddy viscosity for the fluid flow is obtained with the low Reynolds k-ε turbulence model by Launder and Sharma [13]. The above equations can be used to simulate the flow when no porous medium is present. This term was represented by a Darcy-Forchheimer extended model in the form of the modified Ergun model [5]. r stands for the body force term. of the locally defined turbulence quantities.618 Diffusion in Solids and Liquids IV where ρf. is given by the Carman-Kozeny equation [12]: K= φ3 180 (1 − φ ) 2 dp2 (4) where dp stands for the equivalent particles diameter. it is based on taking the volume integration. These two possibilities were considered.

false diffusion is prevented by using sufficiently fine meshes and this particular flow may be well predicted with the first order upwind scheme and the k-ε turbulence model [17]. A finite volume method and the SIMPLE algorithm [15] were used. Unfortunately. . 2a and 2b. flow inside the foam could not be visualized. σk. it is observed that the k-ε model of Pedras and Lemos [14] predicts the promotion of the turbulence. No recirculation could be detected downstream the foam exit for all the calculation (50 mm thick. Ck. Results Figures 2a and 2b show predicted streamlines and turbulence kinetic energy contours assuming. All the calculations assumed fully developed turbulent pipe flow at the inlet. The results obtained with the two finer meshes (230×116 and 344×173 grid points) were virtually identical and therefore all the results presented for this geometry were obtained with the mesh with 230×116 grid nodes. respectively. and Cµ are standard k-ε model constants.5. In the results obtained with the model by Pedras and Lemos [14] the enhancement of the diffusivity owing to the turbulence almost eliminates the laminar separated flow region inside the foam at the expansion wall. laminar flow within the porous medium and using the turbulence model of Pedras and Lemos [14].g.Defect and Diffusion Forum Vols. From these two figures. In order to study the dependence of the results on the grid refinement. and. since a fine mesh was used. the results differ from the predictions where the laminar model is used. Moreover. 10. The outflow from the foam resembles the streamlines shown in Figs. therefore. nor could any measurements be taken as to be able to confirm the existence of flow separation. if this model is used. calculations of laminar flow in the pipe with a sudden expansion were obtained using seven regular meshes with a coarse-to-fine refinement factor of 1. C1ε. 20 and 60 ppi samples). Higher order schemes for the discretization of the convective terms (see e. umerical Method. The convection terms were discretized with a deferred correction in which central differences contribute 80% and first order accurate upwind differences 20%. 283-286 619 ρf ∇ ⋅ u ε ( f ) = ∇ ⋅  µ + σ  f  µ tφ  ε ∇ φ ε   ( f )   + C1ε − ρ f u ' u ' : ∇ u with ( f ) εk f f f  f ε ε u  + C 2 ε ρ f φ C k − f K k   ( )  2    (6) 2 f f − ρ f u ' u ' : ∇ u = µ tφ 2 D − φρ f k I 3 (7) µ t φ = ρ f Cµ (k ) f 2 ε f (8) where µ tφ is the dynamic eddy viscosity based on intrinsic averages of turbulence quantities. As a consequence. Kobayashi and Pereira [16]) were not required. it is well known that this particular sudden pipe expansion flow problem presents very low velocities in the regions of strong curvature of the streamlines while high velocities are only found in regions of quasi one-dimensional flow. C2ε. and the low Reynolds k-ε model by Launder and Sharma [13] was used to predict the flow in the clear fluid pipe sections. σε.

6 0.2 1 0. Figure 3 shows a satisfactory agreement between the experiments and calculations of the stream-wise velocity profile on a plane positioned 10 mm upstream the foam inlet for the impinging jet flow around the centerline. 1. the flow displays the velocity characteristics of a typical cavity and experiments with smoke flow visualization were not conclusive about penetration of the separated flow streamlines into the foam. dashed line indicates foam outlet).000 Laminar Pedras and Lemos [14] Pereira et al. Calculations were also performed for several cases where the foam sample is located at a finite distance from the sudden pipe expansion. where the velocities are relatively low compared to the velocity at the centerline jet-like region.Comparison of predicted and measured mean axial velocity in a 60 mm long cavity flow. The predictions obtained with the model by Pedras and Lemos [14] do not show the penetration of the flow recirculation inside the porous matrix.Streamlines and turbulence kinetic energy contours predicted for flow within foam positioned at pipe sudden expansion. [9] Um/UCL 0. Results taken 10 mm upstream foam inlet (50 mm thick 20 ppi foam. The experimental results show a smaller region with negative axial velocities close to the pipe wall. in the separated flow region.030 0. (b) using the model of Pedras and Lemos [14] (20 ppi. However. 3. They show the formation of a toroidal vortice flow in the cavity space. In this region. b. (a) assuming laminar flow in the foam.4 0. Summary Calculations were conducted to investigate the influence of a ceramic foam insert on the flow inside a pipe with a 1:4 sudden expansion.4 0. It has been observed that this gives rise to the formation of an axisymmetrical cavity between the expansion wall and the foam sample.2 0 -0. equal to 60 mm are shown in Fig. Rep=115.045 Figure 3 .8 0. Results taken 10 mm upstream the foam inlet for a cavity lengths. The main conclusions can be summarized as follows: i) Predictions of the sudden pipe expansion cavity assuming laminar flow within the foam yield the penetration of the separated flow region into the foam.015 r (m) 0. Rep=115). both its size and the velocity magnitudes differ significantly.620 Diffusion in Solids and Liquids IV Figure 2 .2 -0. Predictions and measurements of the mean axial velocities for the cavity flow show very .

Korchak: Chem. Vol. Washington DC. Nevertheless. 433. 283-286 621 large qualitative differences close to the pipe wall. Eng. [14] M. 18 (1979). References [1] M. Hsu and P. Comiti and F. p. Szekely: AIChE J. p.J. 941. 3879.I. J. J. Pereira: Commun. Acknowledgements This work was developed within the framework of project POCI/EME/59510/2004. M.S. Sci. J. 2803. K. Kaviany: Principles of Heat Transfer in Porous Media (Springer-Verlag. Thompson: AIAA J. 12 (1996).S.T. Macdonald.F. Vol. 1991).F. [5] I. Vol. which is financially supported by FCT-Fundação para a Ciência e a Tecnologia. Fund. 56 (2001). Y. 39 (1996). 53 (1998). Hiatt: Exp. Montillet. [3] M. El-Sayed. Vol. [8] D. 131.R.C. 28 (1990). Nield: Int. p. Huet: Chem. Eng.P. p.H. O. Heat Mass Tran. p. Prakash.F. J. 395. Smith and E. Dullien: Ind. Mahoney and G. Hall and J.D. 199. Thorpe: Chem. Sci. Pereira.V. (Submitted) [10] C. p. A. 1587. Heat Mass Tran. Raposo: Exp. ii) The influence of assuming laminar or turbulent flow within the foam over the flow upstream the foam is limited to the influence of the pressure loss within the foam. Pedras and M. Vol. I. 40 (1997). Li. [15] S.C. P. programme POCI 2010 (29. Chem. Fluid Sci. Masuoka and Y. Vol. 1980). Hayashi and J. J. K.F. Vol. 23 (1965).I. Takatsu: Int. Prakash. O. [17] W.F. Stanek and J. p.M.18% from OE). . Vol. Eng.F. the use of laminar or turbulent assumptions has been shown to have a large qualitative impact in the vicinity (within about five pore diameters distance) of the foam inlet. 33 (1990). [11] V. p. Seguin.A. Turan. [2] M. Patankar: umerical Heat Transfer and Fluid Flow (Hemisphere. Meth. Launder and B. Vol. [9] J.A. Sci. Malico. [4] D. Eng. Cheng: Int. 56 (2001). J.H. Vol.Defect and Diffusion Forum Vols.C. 3855. Eng.E. New York.H. Heat Mass Tran. Sci. Kobayashi and J. 276. 2499. 1 (1974). p.S de Lemos: J. Eng. [13] B. 974. Sharma: Lett.L. Fluids Eng. Y. Vol. p. [16] M. Fluids Vol. 20 (1974). Heat Mass Tran. Mahoney and G. p.A. Li.R. Thorpe: Chem. Turan. [7] T.J.82% of the funds from FEDER and 70. Mickeley. p. T. 3751. [6] H. Mow and F. Numer. 20 (1996).J. p. Stevenson and H. [12] M. Vol. 123 (2001). Therm. 237. Vol.

net/DDF.Diffusion in Solids and Liquids IV doi:10.net/DDF.616 .283-286 Analysis of the Flow at the Interface of a Porous Media doi:10.scientific.4028/www.283-286.scientific.4028/www.

Sign up to vote on this title
UsefulNot useful