You are on page 1of 9

Pergamon (hemtc~tl L.,lin,'er:nq 5,1cmc. Vol 52. N o s 21 22. p p 41)45 41, ~,~,.

1997
c b.~97 Flse',ler Science I.td All rights rexer',ed
PriNted in Qircal Bril~lln
PII: S0009-2509(97)00247-9 ~ .~50~97sly ~ , t~t~)

Flow regimes, liquid holdups and two-phase


pressure drop for two-phase cocurrent
downflow and upflow through packed beds:
air/Newtonian and non-Newtonian liquid
systems
I. lliuta** and F. C. Thyrion'*
*Department of Chemical Engineering. Faculty of Industrial Chemistry. "'Politehnica'"
University of Bucharest, Polizu 1. 78126 Bucharest, Romania: :Chemical Engineering Institute,
Universit8 Catholique de Louvain, 1 Vole Minckelers, B-1348, Louvain-la-Neuve, Belgium

IAcccpted I July 1997)

Abstract The analysis of the hydrodynamics results of two-phase downflow and upflow
through fixed beds with porous particles shows that the flow regimes, two-phase pressure drop
and liquid holdups are strongly influenced by the presence of the non-Newtonian liquids. With
highly viscous non-Newtonian liquids the hydrodynamic characteristics (pressure drop, dy-
namic and total liquid holdups) of two-phase downflow are close to those of two-phase
upflow- the axial dispersion appears to be an important cause for the possible effects of the
hydrodynamics on the reactor performance. ,c 1997 Elsevier Science Ltd

Keywords: Two-phase flow: fixed beds: gas,'non-Ncwtonian liquid systems.

INTRODUCTION laboratory-scale packed-bed reactor was a glass col-


Much research into the operating behaviour of down- umn with an inside diameter of 0.051 m. The packed
flow and upflow fixed-bed reactors has been carried height was 0.92 m. The bed was made of porous
out over the last two decades. Despite the fact that in particles {SiO2/AI,Os:'Cr20.r-internal porosity:
the biochemical reactors many fluids do not exhibit 0.464) of 3.3 mm diameter ldpl, the porosity of the bed
Newtonian behaviour, a few papers concerning the being 0.356.
hydrodynamics of fixed-bed three-phase reactors un- Air was used as the gas phase and water or aqueous
der these conditions have been published. Unfortu- carboxymethyl-cellulose (CMC) solutions at different
nately, these studies are limited on the two-phase concentrations were used as the liquid phase. The
pressure drop for two-phase downflow (Larkins et al., CMC solutions were prepared by dissolving CMC
1961: Sai and Varma, 1987) and two-phase upflow powder lcarboxymethyl-cellulose sodium salt, low
{Srinivas and Chhabra, 19941 through packed beds viscosity) in water at 60 C. The rheological properties
with non-porous particles or liquid holdup for two- of the CMC solutions were measured with a Couette
phase downflow through non-porous particles (Sai viscometer (Haake, model VT 180) at shear rates (7)
and Varma, 1988). from 63 to 10('17s ~. The shear stress (V) was meas-
Taking into account these aspects, the main objec- ured with both increasing and decreasing shear rates
tive of this work is to contribute to the elucidation of and a mean value was used in evaluating the rheologi-
hydrodynamics of two-phase downtlow and upflow of cal model parameters. The flow behaviour of
air/non-Newtonian (pseudoplastic) liquid systems pseudoplastic CMC solutions was characterized by
through fixed beds with porous particles. the power-law model of Ostwald de Waale (F = k ;'").
The consistency index k and the flow behaviour index
EXPERIMENTAl, n (Table 1) were evaluated by plotting the logarithm of
the measured shear stress against the logarithm of
The description of the apparatus employed in this
shear rate: a linear regression then gave k and n from
study has been given previously (lliuta. 1996). The
the power-law model.
The liquid holdups were obtained by least-squares
fits of the experimental residence-time distribution
'Corresponding author. Tel.: 004013126879: e-mail: curves. Also. the dynamic liquid holdup (volume of
i._iliuta(a)chim,upb.ro. the liquid drainedvolume of the bed) was obtained by
4(145
4046 I. Iliuta and F. C. Thyrion
Table 1. Physical and rheological properties of non-Newto- only trickle and pulsing-flow regimes were observed.
nian liquids (22C) The transition between trickle and pulsing-flow
regimes is plotted in Fig. 1. A comparison of the
Non-Newtonian Pt. k x 10 3 experimental points shows that, for a i r - C M C liquid
liquids (kg..'ms ) (kg/m s2 ")
n systems, the transition occurs at relatively lower gas
superficial mass flow rate for a given liquid superficial
0.1% CMC 1000.30 4.96 0.936
0.5% CMC 1001.40 17.78 0.900 mass flow rate. For air water system, the dispersed
1% CMC 1004.67 55.99 0.849 bubble flow was observed at

L = 17kg/m z s a n d G ~ < 0 . 1 8 4 k g / m 2s.

draining the bed after the inflow was stopped, and the In the case of two-phase upflow of air-water and
residual liquid holdup (volume of the liquid retained air-CMC liquid systems, the transition between
on the packing after draining/volume of the bed) was bubble and pulse flow regimes is unclear. Instead, we
obtained from the difference between the theoretical noted a common zone of transition (0.092 < G <
external void volume of the bed and the volume of the 0.12 kg/m2 s) for air-water and air-CMC systems
drained liquid from the reactor previously completely (Fig. 2).
filled with liquid.
Residence-time distribution curves were obtained Liquid holdup
using the imperfect pulse method. According to this As in the case of two-phase flow through fixed beds
method, the electrical conductivity of the potassium with non-porous particles (Iliuta et al., 1996b), for
chloride traccr is simultaneously measured down- two-phase flow through porous particles total and
stream or upstream of the injection point at two dynamic liquid holdups increase with increasing
different locations at the bottom and the top of the CMC concentration due to the increase of the liquid-
bed. The signals of the sensors are amplified and side shear stress at the gas-liquid and liquid-solid
transmitted to a computer by a data acquisition sys- interface (Fig. 3). For two-phase upflow, the effect of
tem. CMC concentration is much less marked with porous
Assuming that tracer diffusion is considered to particles than with non-porous particles. This may be
create the long tails of the residence-time distribution due to the fact that the variation of two-phase pres-
curves, we used a transient diffusion model of the sure drop with CMC concentration is larger with
tracer in the porous particle coupled with the piston- non-porous particles (the effect of the liquid-solid
dispersion-exchange model (PDE) model to describe shear stress in the balance of the forces is more impor-
the liquid flow (lliuta et al., 1996a). tant in the case of non-porous particles).
The transition from the high to low interaction With two-phase downflow, the values of the dy-
regime (two-phase down flow) was determined by the namic and total liquid holdups for air-CMC systems
cessation of pressure fluctuations characteristic of the are much higher than the values for air-water systems.
pulse-flow regime or by visual observations of the bed The same phenomenon is observed in the case of
for dispersed bubble-flow regimes. With two-phase two-phase upflow, but the difference is much lower
upflow, the transition from bubble flow to pulse flow and is much less marked in fixed bed with porous
was determined or by visual observations. The pres- particles than with non-porous particles.
sure drop measurements were made with a pressure For air-0.5 and 1.0% CMC systems, the values of
transducer connected to two pressure taps located the total liquid holdup and dynamic liquid holdup for
near the bottom and the top of the packed bed. two-phase downflow are close to those for two-phase
In order to compare the values of the liquid hold- upflow (Table 2). This is due to the fact that the
ups and two-phase pressure drop with literature cor- liquid-side shear stress appears to be very important
relations, we considered the mean values, calculated in the balance between driving forces and the resist-
for each pair of gas and liquid flow rate. Under trickle- ance.
flow regime (two-phase downflow) and pulsing-flow The difference between total and dynamic liquid
conditions (two-phase downflow and upflow), since the holdups (static external liquid holdup) strongly de-
liquid holdup inside a pulse is much larger than outside pends on the type of packing, the dimension of the
a pulse, liquid holdups and two-phase pressure drop in particles, the nature of the liquid and is a result of
the column vary between an upper and lower limit. a balance between the gravity forces and the capillary
This effect was clearly visible in the standard deviation forces. In the present work, the static external liquid
of the repetitive measurements. holdup was found to be independent of the liquid and
gas flow rates for both two-phase downflow and up-
RESULTS A N D D I S C U S S I O N flow. In the case of two-phase downflow, this is due to
the fact that the external wetting efficiency (the static
Flow regimes external liquid holdup is dependent upon the wetted
In the case of air-water systems trickle, pulsing and surface) is close to 1 at liquid superficial velocities
dispersed bubble-flow regimes were observed. In the higher than 0.005 m/s due to increased radial liquid
case of two-phase flow of a i r - C M C liquid systems spreading.
Air Newtonian anti non-Newtonian liquid ~stcmx 4047

0.8 , - - ~ r

0.7 ¸

0.6

05l I

% I
E
Q4-
Pulsing flow

d ~

0.3 i-

o2! X
0

0.1~ +
4-

Trickle flow
0 i t I I

4 6 8 10 12 14
L, kg.rn2s "~

Fig. I. Transition between trickling and pulsing llow regimes (two-phase downtlow): air water system (.~3)
noCMC, lx)0.1% CMCJ*)0.5% CMC.( - ) 1.0%,CMC.

Comparison with literature correlations for air-0.5% CMC and a poor estimate for air 0.1
At the outset, it is useful to explore the possibility of and 1.0'¼, CMC systems. Hopefully, the correlation of
correlating the present data obtained with air:non- karachi et al. (19911, developed for gas-Newtonian
Newtonian liquid systems with the empirical expres- liquid systems, seems to be satisfactory for the entire
sions available in the literature. Unfortunately, only in range of CMC concentration. For two-phase upflow,
the case of two-phase downflow, Sai and Varma unfortunately, no correlation for the total liquid hold-
(1988) have presented correlations for dynamic and up in the case of gas/non-Newtonian liquid systems
total liquid holdups. exists in the literature. The total liquid holdup data
Likewise. it is interesting to compare the experi- were correlated by using the expression of Larachi
mental data obtained in the case of two-phase flow of et ul. (1991), established for two-phase downflow and
air/non-Newtonian liquid systems with correlations gas Newtonian liquid systems.
available at this moment for two-phase flow with
gas'Newtonian liquid systems (Ellman et at., 1990: Dynamic liquid holdup. The predicted values of the
Larachi et al., 1991). dynamic liquid holdup for two-phase downflow
Thus, the present results are analysed in terms of through porous particles calculated from the expres-
the previously reported correlations established for sion of Ellman et al. (1990), which was established for
non-porous and porous packings and various gas- gas Newtonian liquid systems, agree well with
Newtonian liquid (Table 3), based on the use of the our data, as it can be seen in Table 4. The correla-
modified Reynolds number (lliuta et al., 1996b). Also, tion of Sai and Varma (1988) predicts lower dynamic
in the case of two-phase downflow the results are liquid holdups than our values, in the case
analysed using the correlations of Sai and Varma of air-CMC systems the high interaction regime
(1988). Table 3. correlation of Ellman et al. (1990) gives a fair esti-
mate of the dynamic liquid holdup for two-phase
Total liquid holdup. The comparisons between the upJtm~.
predictions of the two literature correlations and ex-
perimental data are summarized in Table 4. With Two-phase pressure drop
two-phase downflow, the correlation of Sai and Varma For both two-phase downflow and upilow through
(1988) gives a fair estimate of the total liquid holdup fixed beds with porous particles, two-phase pressure
4048 I. lliuta and F. C. Thyrion

lo2l
\
Pulse flow
Bubble flow
\
\
\

10~

!
I Spray flow

10°1 i , i i i i , iI , . . . . . . iI

10 -2 10 "1 10 0

G, kg.m2s 1

Fig. 2. Hydrodynamic regimes for two-phase upflow: (', "..',..)observed transition zone, (---) transition lines
from Turpin and Hutington (1967).

drop increases with increasing CMC concentration Thus, as in the case of the liquid holdups, in this
due to the increase of the liquid-side shear stress at the work two-phase pressure drop data for air/non-New-
gas-liquid and liquid-solid interface (Fig. 4). As ex- tonian liquid systems were correlated using expres-
pected, in the case of air-CMC systems two-phase sions established for gas-Newtonian liquid systems
pressure drop increases with increasing liquid flow (Ellman et al., 1988; Larachi et al., 1991), based on
rate and with increasing gas flow rate, such trends are the use of the modified Reynolds number (Table 5).
in agreement with the observations made in the case In these correlations, the gravitational effects are
of air/Newtonian liquid systems. negligible, the Weber number incorporates the inter-
facial forces between the liquid and gas phases, the
Comparison with literature correlations. Only Sai modified Lockhart-Martinelli parameter incorporates
and Varma (1987) and Srinivas and Chhabra (1994) the inertial forces of the gas and liquid while
presented correlations for two-phase pressure drop in the Reynolds number incorporates the viscous forces
the case of two-phase downflow and upflow of inside the liquid.
air/non-Newtonian liquid systems through fixed beds Concerning the gravitational effects on the two-
with non-porous particles. Unfortunately, these cor- phase pressure drop, one can note with air-0.5 and
relations are based on the Lockhart-Martinelli 1.0% CMC systems, two-phase pressure drop
parameter and require in situ measurements of the measurements for downflow (high interaction regime)
single-phase pressure drop. It is possible to use an and upflow operation give close results (Table 6), which
Ergun's-type equation for single-phase pressure drop, means that the gravity forces may be neglected com-
but the empirical coefficients of this correlation have pared to inertial forces and liquid-side shear stress.
to be determined experimentally; a priori calculations The scatters between predictions and experiments
can be very inaccurate. Moreover, in the case of two- are summarized in Table 7. The correlation of Ellman
phase downflow of air/non-Newtonian liquid systems, et al. (1988) gives better predictions for both two-
we cannot measure the liquid-phase pressure drop phase downflow and upflow (with highly viscous
due to the occurrence of flooding in the case of single non-Newtonian liquids) through fixed beds with por-
liquid flow. ous particles.
Air.,Newtonian and non-Ncwtonian liquid systems 4049

Q.
"0
0
t-

.0
E
C
E3

U.UUJ 05 V~(3, m/s


(a) VSL, m/s

r~
10
0
t--

.O"

.0
E

r~

u uuz 0 05 V,5(3, m/s

(b) VSL, m/s

Fig. 3. Effects of gas and liquid velocities (vs~; and VSLJon the dynamic liquid holdup for various gas-liquid
systems, (a) two-phase downflow, (b) two-phase upflow, and for air-water system (1) no CMC. (2) 0.1%
CMC. (3) 0.5% CMC. (4) 1.0% CMC.
4050 I. lliuta and F. C. Thyrion
Table 2. Values of the dynamic liquid holdup for fixed bed with porous particles [VsL = 0.005 m/s)

Dynamic liquid holdup, (m2/m 3)

Two-phase downflow Two-phase upflow

Vs¢; air-CMC air-CMC air-CMC air..CMC air CMC air CMC


(m/s) air-water 0.1% 0.5% 1.0% air water 0.1% 0.5% 1.0%

0.028 0.144 0.182 0.212 0.225 0.226 0.230 0.232 0.236


0.085 0.139 0.167 I).179 0.197 0.186 0.191 0.198 0.213
0.140 0.133 0.158 0.165 0.178 0.171 0.178 0.183 0.186
0.228 0.126 0.141 0.149 0.162 0.145 0.150 0.156 0.166
0.410 0.094 0.113 0.130 0.146 0.118 0.120 0.138 0.153

Table 3. Liquid holdup correlations used in this study

Correlations established for gas/Newtonian liquid systems

Stiegel and Shah (1977) ~h.., = 1.47~.Re°L'll Red °'14(a.dp)-I),41


q'
Ellman et al. (1990) ~:l,.a = ~:× 10;, where ~ = 0.001 - --

¢ = X~," Re[ We~.\


1
q~, m, p, q, r, s = constants
Larachi et al. (1991) ,%., = t:(1 - 10 ')
= 1.22 We~.TM " oo1~,(X(;'. ~s, +o~x)SJRe~-~±.~..4~)

Correlations established for gas/non-Newtonian liquid systems

Sai and Varma 11988) /Re,~ ~


~:t,.a= 5.83el2(a, x10 3),.51__=/1 kO.53
\ Re~;/

~;L., = 3-42':'"2(a, x 10-3)""s( Ret'~ L'k°''~


" \ Re~ J

Table 4. Scatter between experimental and predicted total and dynamic liquid holdups

Mean absolute relative error (,~), (%)

Two-phase downflow Two-phase upflow

Total liquid Dynamic


Total liquid holdup Dynamic liquid holdup holdup liquid holdup

Gas --liquid Larachi Sai and Varma EIIman Sai and Varma Ellman
system correlation correlation correlation correlation correlation

Air-water 3.50 -- 9.75 -- 4.39* 28.60


Air-0.1% CMC 8.90 18.50 9.50 21.80 10.10' 11).17
Air 0.5% CMC 6.70 9.00 10.80 13.50 12.40' 5.24
Air 1.0% C M C 6.10 20.30 4.80 14.50 ll.90' 8.29

* Stiegel and Shah 11977) correlation.


t Larachi et al. 11991) correlation.

CONCLUSIONS liquid h o l d u p are strongly influenced by the presence


The analysis of the h y d r o d y n a m i c s results of two- of n o n - N e w t o n i a n liquids.
phase downflow a n d upflow t h r o u g h fixed beds with The h y d r o d y n a m i c characteristics (pressure drop,
porous particles shows that the flow regimes, two- d y n a m i c and total liquid holdups) of two-phase up-
phase pressure d r o p a n d the different parts of the flow for the air-0.5 and 1.0% C M C systems are close
Air,'Newtonian and non-Newtonian liquid systems 4051

x 104

0~

0.005% ~ ~ ~ 0 3 02 01
04
(a) VSL, m/s 0 0.5 VSG, m/s

4
xl0
7~

6~
5\

<3

1,
0.015
\
0 01

(b) VSL,m/s 0 0.5 VSG.m/s


Fig. 4. Effects of gas and liquid velocities on the two-phase pressure drop for various gas liquid systems (a)
two-phase downflow, (b) two-phase upflow and for air-water system (I) no CMC, (2) 0.1% CMC, f3) 0.5°/.
CMC, (4) 1.0% CMC.

Table 5. Two-phase pressure drop correlations used in this stud)'

Re~
Ellman et al. (1988) .l;-c;~;= 200 ( X . ~ ) - , . 2 + 85 (Xc;:)- o .~. where
(0.001 + ReL1,5 )

Larachi et al. (1991) fL(~c; = 31.3(-t- 3.9) + -


17.3(_~
-
0.6)) where fl = X(;(ReL WeL)"'2~
3o5
4052 I. lliuta and F. C Thyrion
Table 6. Two-phase pressure drop for tixed bed with porous particles (rsL = 0.(X)5m/s)

lwo-phase pressure drop (mbarl

Two-phase downflow Two-phase upflow

I/s6 Air CMC Air-CMC Air CMC Air CMC Air-CMC Air-CMC
(re:s) Air water 0.1% 0.5% 1.0% Air-water 0.1% 0.5% 1.0%

0.028 3.00 9.28 56.40 160.13 11613 144.40 182.17 278.90


0.085 21.85 28.14 106.70 220.00* 135.00 16(I.13 198.00 317.00
0.140 47.00 65.85 141.30" 276.4(1" 147.60 179.00 223.00 348.00
0.228 84.71 103.56 194.70" 336.13" 175.80 204.15 251.28 406.00
0.410 141.27 179.00" 263.00* 411.50" 204.13 267.00 333.00 478.90

* Pulsing-flow regime.

Table 7. Scatter between experimental and predicted two-phase pressure drop

Mean absolute relative error (fit (%)

Two-phase downflow Two-phase upflow

Ellman Larachi Ellman Larachi


Gas-liquid system correlation correlation correlation* correlation

Air -water 20.00 240.00 49.30 31.50


Air 0.1% CMC 34.41 158.00 32.45 13.50
Air-0.5% CMC 22.30 20.81 11.71 58.60
Air 1.0% CMC 24.41 225.0t) 13.(16 79.27

* Developed for high interaction regime (two-phase downflow}.

to those of two-phase downflow (high interaction L liquid superficial mass flow rate, kg/m2 s
regimej--the axial dispersion appears to be an impor- m number of experimental points
tant cause for the possible effects of liquid and gas n flow behaviour index
flow rate on the reactor performance. AP pressure drop, Pa
Good agreement between the predicted (given by Re• gas Reynolds number, ( = dpG."It6)
the correlations obtained for gas-Newtonian liquid Re£ liquid Reynolds number ( = dpL/l~Lj for
systems) and the measured values of the two-phase Newtonian liquids and Re£l = p£t.SL 2-, dp,
, ; k)
pressure drop, total and dynamic liquid holdup for for non-Newtonian liquids
the gas/non-Newtonian liquid systems was demon- vs6 superficial velocity of the gas, m/s
strated in this study. vsL superficial velocity of the liquid, m s
X,; modified Lockhart--Martinelli parameter,

NO'IATION =z 2j
ap specific area of particle {surface of the par- y hydrodynamic parameter ( = eL,,, CL.a,./L(m)
ticle/volume of particlel, m2/m 3 We£ liquid Weber number, ( = L 2 dp/p£,aL)
a:, specific area of the packing (surface of the
particles/volume of the bed), m2/m 3 Greek letters
dh hydraulic diameter of the packed bed. ¢ effective shear rate, 1.,'s
[=( 16r,3 \°'33 7 F shear stress, Pa
\9n(l __ c)2) dpJ 6 mean absolute relative error
m , /
dp diameter of a sphere having the same ex- [ = 1.,m]Li. , I(.v=~., - 3,,=.;),'~.ox~.~l]
ternal surface area as the particle, m E bed porosity
fec, c~ two-phase friction factor, EL liquid holdup
[ = (AP/tt)Lc, d~,pc,/2G2] viscosity, kg/m s
G gas superficial mass flow rate, kg/m2s ,Heft effective viscosity of non-Newtonian liquids,
k flow consistency index, kg/m s 2 " kg/m s
Air/Newtonian and non-Newtonian liquid systems 4053
density, kg/m 3 lliuta, I., Thyrion, F. C. and Muntean, O. (1996b)
interfacial tension, kg/s 2 Hydrodynamic characteristics of two-phase flow
through fixed beds: air/Newtonian and non-New-
Subscripts tonian liquids. Chem. Enftn9 Sci. 51, 4987~,995.
calc calculated Larachi, F., Laurent, A., Midoux, N. and Wild, G.
d dynamic {1991) Experimental study of a trickle-bed reactors
operating at high pressure: two-phase pressurc
exp experimental
drop and liquid saturation. Chem. En~lmt Sci. 46.
G gas 1233-1246.
L liquid Larkins, R. P. White, P. R. and Jeffrey, D. W. 11961)
LGG gas liquid/gas Two-phase concurrent flow in packed beds.
t total A.I.Ch.E.J. 7, 231 239.
Sai. P. S. T. and Varma, Y. B. G., (1987) Pressure drop
in gas-liquid downflow through packed beds.
A.I.Ch.E.J. 33. 2027 2035.
REFERENCES Sai, P. S. T. and Varma. Y. B. G. (1988) Flow pattern
Ellman, M. J., Midoux, N., Laurent, A. and Charpen- of the phases and liquid saturation in gas-liquid
tier, J. C. (1988) A new, improved pressure drop concurrent downftow through packed beds. Can. J.
correlation for trickle-bed reactors. Chem. Engn,q Chem. Engng 66, 353-360.
Sci. 43. 2201-2206. Srinivas, K. V. and Chhabra, R. P. (1994l Pressure
Ellman, M. J., Midoux, N., Wild, G., Laurent, A. and drop in two phase cocurrent upward flow in packed
Charpentier, J. C. (1990) A new, improved liquid beds: air/non-Newtonian liquid systems. Can. J.
hold-up correlation for trickle-bed reactors. Chem. Chem. Enynq 72, 1085 1091.
Enqng Sci. 45, 1677 1684. Stiegel, G. J. and Shah, Y. T. 11977) Backmixing and
lliuta, I. (1996) Hydrodynamics and mass transfer in liquid holdup in a gas liquid cocurrent upflow
multiphase fixed bed reactors. Ph.D. thesis, Univer- packed column. Ind. Ending Chem. Process Des. Dev.
sitO Catholique de Louvain, Belgium. 16, 37-43.
lliuta, I., Thyrion, F. C., Muntean, O. and Giot, M. Turpin, J. L. and Huntington. R. L. (1967) Prediction
(1996a) Residence time distribution of the liquid in of pressure drop for two-phase, two-component
gas-liquid cocurrent upflow fixed-bed reactors. concurrent flow in packed beds. A.I.Ch.E.J. 13,
Chem. Engng Sei. 51, 4579-4593. 1196-1202.

You might also like