Professional Documents
Culture Documents
3897—3909, 1998
( 1997 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0009–2509(97)00176–6 0009—2509/98/$—See front matter
Abstract—This work deals with hydrodynamics in porous media beyond the end of the stable
laminar regime, at higher Reynolds numbers. Local measurements of current fluctuations were
carried out with electrochemical probes located at different positions in the porous media.
Owing to the use of the electrochemical transfer function, the spectrum of velocity gradient
fluctuations at the micro-electrodes and the velocity gradient fluctuating rate were determined.
In packed beds of particles, a stabilization of this fluctuation rate was observed at most
electrodes in the Reynolds number range covered. It is shown that this stabilization corres-
ponds to a locally turbulent flow. In reticulated media, no such stabilization was observed. The
characteristic length scales of the flow, i.e. the order of magnitude of the flow eddies dimensions,
were evaluated from the autocorrelation function of the velocity gradient fluctuations cal-
culated from the spectrum. They were compared to the pore diameter calculated from the
capillary model proposed by Comiti and Renaud [(1989), Chem Engng Sci. 44, 1539—1545]. The
stable values of the length scales obtained for high Reynolds numbers confirm the turbulent
nature of the flow regime. The flow regime transition is gradual from laminar to turbulent in the
entire bed, it is characterized with the pore Reynolds number, Re , based on the employed
p
capillary model: in packed beds presenting an isotropy in the plane perpendicular to the main
direction of the fluid flow, the laminar regime ends at Re "180 whereas a value of Re "900
p p
corresponds to the stabilization of the velocity gradient fluctuating rate at 90% of the
electrodes. Calculations, based on a pressure drop model related to the capillary representation,
show that the percentage of inertial effect on the pressure drop is then about 90% at this
Reynolds number in these packed beds. ( 1998 Elsevier Science Ltd. All rights reserved.
Keywords: Porous media; turbulent regime; characteristic length scales; spectral analysis;
capillary representation; velocity gradient fluctuation
References Porous media Measurements Remarks about the observation Criterion for flow regime
location of the turbulent regime characterization
is open to discussion, as can be seen in the different — Packed beds of square based, 5 mm in side, para-
criteria proposed. llelepipedal plates with different height to side
A study of the flow regimes over a wide range of ratios: e/a"0.21 and e/a"0.44. As explained be-
high Reynolds numbers in packed beds of spheres of fore (Seguin et al., submitted), according to the
different diameters should enable to get more accurate orientation of the packing plane of the plates and
information about phenomena that are not well the main direction of the fluid flow, they can be
understood yet. It should enable to improve the investigated either in the parallel or in the perpen-
knowledge of the evolution of the velocity gradient dicular configuration by using, respectively, a cy-
fluctuating rate at high Reynolds numbers and help to lindrical cell of diameter 0.06 m or a square based
determine, in particular, the exact nature of the flow parallelepipedal cell of side 0.055 m.
regime as well as the Reynolds number corresponding — Synthetic foams of different grades, i.e. number of
to the laminar—turbulent transition. This work also pores per inch: G10, G20, G45 and G100.
mostly focuses on other types of porous media, such
as packed beds of variously shaped particles and Local measurements in various locations of these
reticulated media, only investigated, to our know- porous media were carried out thanks to electro-
ledge, by Hall and Hiatt (1996). chemical micro-probes either located in networks at
The effect of the structure of those porous media the wall surface of the test column or inserted at the
will be enlightened owing to the testing of different surface of bed particles inside the porous media. In
dimensionless criteria for the characterization of the packed beds of spheres, internal electrodes were either
observed flow regime transition. In particular, the located at the surface of a single particle inserted in
pore Reynolds number, Re , based on the application the fixed bed or facing the center of a tetrahedral
p
of the capillary model proposed by Comiti and Re- assembling made with four spheres. A more complete
naud (1989) will be tested because it was shown to be description of the experimental technique and appar-
the most suitable criterion for the characterization of atus is provided in the first part of this study (Seguin
the end of the stable laminar regime in beds packed et al., 1998) as well as the capillary model em-
with particles presenting an isotropy in the plane ployed and the structural parameters of these porous
perpendicular to the flow. (Seguin et al., 1998). media.
P
micro-electrode and D is the diffusion coefficient of q0
the ferricyanide in the solution, then, the amplitude ¹ " J (q ) dq (11)
*/5 ss i i
DH( f )D of the transfer function may be expressed as 0
a function of the normalized frequency p: Following Taylor’s hypothesis about isotropic tur-
bulence, orders of magnitude of the characteristic
A B
d2 1@3
p"2n f e (4) length scales can be determined by multiplication of
DS 2 the time scales by the average fluid velocity, º , in the
p
x
pores (Reynolds, 1973).
1 IM The macroscopic length scale, i.e. the maximum
f if p)1: H(p)"H(0)" (5)
3 S1 diameter, D , of the flow eddies is then defined as:
.!9
where IM is the average limiting current at the elec- D "º q (12)
.!9 p 0
trode.
and the integral scale, i.e. the average diameter, D , of
av
f if 1)p)6: the flow eddies as:
D "º ¹ . (13)
H(p)"H(0) (1#0.049p2#0.0006p4)~1@2 (6) !7 p */5
The different steps of the signal processing are sum-
C D
4.416 1.7 1.44 1@2
f if p'6: H(p)"H(0) 1! # . marized in Fig. 1.
p Jp p
(7) 3. VELOCITY GRADIENT FLUCTUATING RATE
Considering that the flow in the vicinity of the 3.1. Evolution of the velocity gradient fluctuating rate
micro-electrodes inside the porous media can be Depending on the electrode location and on the
locally approximated as a pipe or channel flow, this porous medium investigated, two experimental evolu-
electrochemical function was also applied to measure- tions of the velocity gradient fluctuating rate were
ments at internal probes. observed when the Reynolds number was increased:
As usual, the following decomposition of the velo-
city gradient was used: — The velocity gradient fluctuating rate increased
sharply after the onset of fluctuations and then stabil-
S (t)"S #s(t) with s(t)"0 (8) ized. Figure 2 illustrates this typical evolution of the
x x
Flow regimes in various porous media—II 3901
Table 2. Stabilization of F)R in packed beds of particles: minimum pore Reynolds number corresponding to
s
a stabilization, pore Reynolds number corresponding to the stabilization at 90% of electrodes and average
fluctuating rate at the stabilization
Re
p
5 mm spheres 510
8 mm spheres 464, 542, 542
Porous medium Re Re Re
1!35 p i
Table 5. Ranges of stabilized macroscopic and integral length scales, D and D , average values and standard deviations in
.!9 !7
packed beds of spheres (number of determinations in parentheses
5 mm Wall column 1.6—3.6 2.4 (19) 0.7 0.6—1.7 1.0 (19) 0.34
spheres At a single sphere 0.9—4.5 1.9 (5) 1.3 0.4—1.5 0.8 (5) 0.4 1.88
In a tetrahedral assembling 2.1—2.3 2.2 (2) / 0.86—0.90 0.88 (2) /
8 mm Wall column 1.8—5.4 4.0 (14) 1.3 0.7—3.0 1.5 (13) 0.6
spheres At a single sphere 1.9—3.2 2.5 (3) 0.5 0.9—1.0 1.0 (3) 0.06 3.00
In a tetrahedral assembling 3.0—3.2 3.1 (2) / 1.0—1.1 1.05 (2) /
Table 6. Ranges of stabilized macroscopic and integral length scales, D and D , average values and standard deviations in
.!9 !7
packed beds of parallelepipedal plates. (number of determinations in parentheses)
e/a"0.21 // 0.8—2.4 1.75 (9) 0.5 0.32—1.1 0.68 (9) 0.24 0.91
o 1.55—4.1 3.0 (9) 1.0 0.62—1.5 1.1 (9) 0.44 1.5
e/a"0.44 // 1.2—3.0 2.0 (9) 0.7 0.44—1.0 0.64 (9) 0.2 1.51
o 1.5—4.0 2.39 (7) 0.85 0.6—2.0 0.98 (7) 0.46 1.56
Table 6. The dispersion of results is more important same order of magnitude as the pore diameters de-
than in packed beds of spheres, possibly due to fined in the capillary model employed.
a stronger variety of interstice dimensions in the me-
dia. It can be noted that the order of magnitude of the 5. RELATION TO THE PRESSURE DROP EVOLUTION
macroscopic length scales is in agreement with the Another interest of the capillary model is an accurate
pore dimensions. Indeed, in most cases, the pore modelling of the pressure drop inside the porous
diameter is very close to the smallest value of D media investigated owing to the expression proposed
.!9
obtained. by Comiti and Renaud (1989). This model, presented
In packed beds of plates e/a"0.21, the difference in the first part of this work (Seguin et al., 1998), gives
between the sizes of the aggregates in the two config- the following expression of the pressure drop inside
urations was important, showing the strong difference the porous medium:
of flow structure in those media. On the contrary, in
packed beds of plates e/a"0.44, the aggregate sizes in *P k2(1!e)3a3 q
" 7$ (Re #0.0121Re2) (14)
both configurations were quite close, confirming the H 2oe3 p p
similarity of the flows already observed in this work.
As explained in the first part of this study, whatever
4.2.3. ¸ength scales in synthetic foams. No stabiliz- the porous media investigated, a given value of the
ation of the characteristic length scales measured in pore Reynolds number always corresponds to the
synthetic foams was observed, the decrease of the same proportion of inertial contribution to the total
integral and macroscopic scales was continuous over pressure drop. Figure 14 shows the proportion of
the range of Reynolds numbers investigated, showing inertial effect to the pressure drop calculated with eq.
that no turbulent regime was obtained. This is consis- (14) at the characteristic pore Reynolds numbers cor-
tent with the absence of stabilization of the velocity responding to the onset of fluctuations and to the
gradient fluctuating rate observed above. stabilization of F)R at 90% of electrodes in packed
s
beds that present an isotropy in the plane perpendicu-
4.3. Conclusion lar to the main direction of the fluid flow. At the
The stabilization of the characteristic length scales Reynolds number corresponding to the stabilization
provides another argument of the existence of a turbu- of F)R at 90% of electrodes inside these packed bed,
s
lent regime inside packed beds at high Reynolds the inertial stress, i.e. the second term of eq. (14),
numbers, most results being consistent with those represents 91% of the pressure drop in the porous
obtained in the analysis of the velocity gradient fluctu- medium. The flow is then essentially dominated by
ating rate. Moreover, it is particularly interesting to inertial strength, the viscous effect is almost negligible.
note that the determined macroscopic scales have the This result is consistent with the turbulent nature of
3908 D. Seguin et al.
Fig. 14. Proportion of inertial stress to global pressure drop [calculated from eq. (14)] inside packed beds
versus the pore Reynolds number.
the flow observed in most parts of packed beds at this In packed beds of particles presenting an isotropy
Reynolds number. in the plane perpendicular to the main direction of the
fluid flow, the turbulent regime was observed, for 90%
6. CONCLUSION of electrodes, at a pore Reynolds number Re +900.
p
New information about flow regimes in various As noticed for the end of the stable laminar regime, in
porous media have been determined owing to the the first part of this study, the other dimensionless
spectral analysis of the current and velocity gradient criteria tested failed in characterizing this flow regime
fluctuations measured at electrochemical probes transition.
located at different positions in the media. This work Calculations, based on the application of the capil-
also shows the necessity of internal measurements in lary model of Comiti and Renaud (1989), show that, at
packed beds of particles, but not in synthetic foams. this Reynolds number, the inertial contribution rep-
In all packed beds of particles, the evolution of the resents almost 91% of the pressure drop, whereas it
velocity gradient fluctuating rate, F)R , shows a stabil- represents 70% at Re "180, corresponding to the
s p
ization at high Reynolds numbers for most electrodes. onset of fluctuations. This capillary representation
The spectral analysis of the velocity gradient fluctu- then enables an accurate characterization of flow re-
ations and the determination of the characteristic gime transition in packed beds of particles presenting
length scales of the flow show that the flow regime is an isotropy in the plane perpendicular to the main
then locally turbulent. The transition to the turbulent direction of the fluid flow.
flow regime in the whole packed bed is gradual, it
occurs at various Reynolds numbers according to the
location in the porous medium. This phenomenon can NOTATION
explain the gradual changes observed in pressure a side length of a parallelepiped plate, m
drop and mass transfer correlations reported in the a dynamic specific surface area, m~1
vd
literature. d pore diameter defined by Rode et al. (1994).
In synthetic foams, no turbulent regime was detec- d electrode diameter, m
e
ted in the Reynolds numbers range investigated, con- d pore diameter, "4e/(1!e)a , m
p vd
firming the ‘laminarising effect’ observed in the first d particle diameter, "6/a , m
1!35 74
part of this study (Seguin et al., 1998) concerning the D diffusion coefficient of the reacting species,
onset of fluctuations in the same porous media. m2 s~1
Flow regimes in various porous media—II 3909
D integral length scale of the velocity gradient, REFERENCES
!7
m Comiti, J. and Renaud, M. (1989) A new model for deter-
D macroscopic length scale of the velocity mining mean structure parameters of fixed beds from
.!9 pressure drop measurements: application to packed beds
gradient, m
with parallepipedal particles. Chem. Engng Sci. 44,
e thickness of a parallelepiped plate, m
1539—1545.
f frequency, s~1 Deslouis, C., Tribollet, B. and Viet, L. (1983) The correlation
F)R fluctuating rate of the limiting current, di- between momentum and mass transfer for a turbulent or
i
mensionless periodic flow in a circular pipe by electrochemical
F)R fluctuating rate of the velocity gradient, di- methods. Proceedings of the 4th International Conference
s on Physicochemical Hydrodynamic, Vol. 404, Ann. NY
mensionless Acad. Sci., New York, p. 471
F)R normalized turbulence intensity, Deslouis, C., Gil, O. and Tribollet, B. (1990) Frequency
U
("Ju2/ºM ), dimensionless response of electrochemical sensors to hydrodynamic fluc-
tuation. J. Fluid Mech. 215, 85—100.
H height of porous medium, m Dybbs, A. and Edwards, R. V. (1984) A new look at porous
H( f ) electrochemical transfer function media fluid mechanics. Darcy to turbulent. Fundamentals
IM average limiting current, A of ¹ransport Phenomena in Porous Media, eds J. Bear,
J normalized autocorrelation function, de- Y. Corapcioglu, NATO ASI Series, Series E, Martinus
ss Nishoff, Dordrecht.
fined by eq. (9) Hall, M. J. and Hiatt, J. P. (1996) Measurements of pore scale
Qv flow rate, l/h flows within and exiting ceramic foams, Exp Fluids 20,
Re interstitial Reynolds number, 433—440.
i
("º /e d o/k), dimensionless Jolls, K. R. and Hanratty, T. J. (1966) Transition to turbu-
0 1!35 lence for flow through a dumped bed of spheres. Chem.
Re pore Reynolds number, ("º d o/k), di-
p p p Engng Sci. 21, 1185—1190.
mensionless Lamine, A. S., Colli Seranno, M. T. and Wild, G. (1992)
Re@ pore Reynolds number defined by Hall and Hydrodynamics and heat transfer in packed beds with
p
Hiatt (1996), dimensionless liquid upflow. Chem. Engng Process. 31, 385—394.
Re particle Reynolds number, ("º d o/k), Latifi, M. A., Midoux, N. and Storck, A. (1989) The use of
1!35 0 1!35 micro-electrodes in the study of flow regimes in a packed
dimensionless bed reactor with single phase liquid flow. Chem. Engng
Re average Reynolds number, dimensionless Sci. 44, 2501—2508.
S (t), S instantaneous, average and fluctuating velo- Lebouché, M. (1968) Contribution à l’étude des mouvements
x x turbulents des liquides par la méthode polarographique.
and s(t) city gradient, s~1 Thèse d’état, Nancy.
u instantaneous fluctuating velocity compon- Ling, S. C. (1963) Heat transfer from a small isothermal
ent, m s~1 spanwise strip on an insulated boundary. J. Heat ¹ransfer
º experimental local fluid velocity, m s~1 44, 230—236.
Mickley, H. S., Smith, K. A. and Korchak, E. I. (1965) Fluid
ºM mean local velocity, m s~1 flow in packed beds. Chem. Engng Sci. 20, 237—246.
º superficial fluid velocity, m s~1 Nakoryakov, V. E., Budukov, A. P., Kashinsky, O. N. and
0
º fluid velocity in the pore, ("º q/e), m s~1 Geshev, P. I. (1986) Electrodiffusion method of investiga-
p 0 tion into the local structure of turbulent flows, ed. V. G.
t time, s
Gasenko, Novosibirsk.
¹ integral time scale of the velocity gradient, s
*/5 Nakoriakov, V. E., Kashinsky, O. N. and Kosmenko, B. K.
¼ power spectral density of the fluctuating cur-
** (1983) Electrochemical method for measuring turbulent
rent, A2s characteristics of gas-liquid flows. Measuring ¹echniques
¼ power spectral density of the velocity gradi- in Gas—¸iquid ¹wo Phase Flows, Iº¹AM Symposium,
ss Nancy, France, pp. 695—721.
ent, s~1
Reynolds, A. J. (1973) ¹urbulent Flows in Engineering. Wiley
London.
Greek letters Rode, S., Midoux, N., Latifi, M. A., Storck, A. and Saatdjian,
*P pressure drop in the porous medium, Pa E. (1994), Hydrodynamics of liquid flows in packed beds:
e bed porosity, dimensionless an experimental study using electrochemical shear rate
k dynamic viscosity of the fluid, Pa s sensors. Chem. Engng Sci. 49, 889—900.
Seguin, D. (1997) Contribution à l’étude du transfert de
o liquid density, kg m~3 matière et des régimes d’écoulement dans des milieux
p dimensionless frequency, defined by eq. (3) poreux de structures variées. Thèse de doctorat, Univer-
p standard deviation of Re. sité de Nantes, France.
n~1 Seguin, D., Montillet, A. and Comiti, J. (1998) Experimental
C S D
+n (Re!Re )2
" 1 , dimensionless characterization of flow regimes in various porous media.
n!1 I—limits of laminar flow regime. Chem. Engng Sci. 53,
q bed tortuosity, dimensionless 3751—3761.
Van der Merwe, D. F. and Gauvin, W. H. (1971) Velocity
q incremental time delay, s
i and turbulence measurements of air flow through
q0 macroscopic time scale of the velocity gradi- a packed bed. A.I.Ch.E. J. 17, 519—528.
ent, s