Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

On the Influence of Liquid Elasticity on Mixing in a Vessel

Agitated by a Combined Ribbon-Screw Impeller


V. V. CHAVAN, M. ARUMUGAM, and J. ULBRECHT
Department of Chemicol Engineering
University of Salford, Salford M5 4WT, England

Coyle et al. (1970) published a report on the influence 45 cm. The rest of the dimensions were as follows:
of geometry and pseudoplasticity upon the performance
of helical ribbon impellers. In this note, we indicate how
t/d1= 1.03; H / d 1 = 1.30; Z/d1 = 1.15; sr/dl = 0.57
their results can be extended for liquids showing a visco- wl/dl = 0.12; d2/dl = 0.29; and wdd2 = 0.4
elastic behavior. This extension will be found particularly
useful for the design of batch-operated or continuous Further details of the experimental procedure can be
stirred reactors for polymerization in bulk or in solvent. found in Arumugam's dissertation (1974).
Coyle et al. (1970) measured the time a fluid element Several measurements were made to obtain the time
takes for one circulation in the vertical plane using vis- the particle takes to complete one circulation in the ver-
cous Newtonian and some inelastic liquids and employ- tical plane (axial circulation time e,) and in the horizontal
ing the particle-follower technique. They obtained identi- plane (angular circulation time 0,'). Although the indi-
cal circulation times for both the Newtonian and pseudo- vidual observations are subject to random variations, the
plastic liquids if the geometry and rotational speed re- average values of e, and e', are in a way related to the
mained constant. These workers have also measured the overall axial flow and the overall angular flow, respec-
mixing-time employing dye-dispersion technique and tively. These averages are plotted against the rotational
found that the mixing time also does not depend upon speeds in Figures 2 and 3. In each case the circulation
the fluid properties and that its value is three times time was found to be inversely proportional to the rota-
higher than that of the corresponding circulation time. tional speed of the impeller. Apart from the inverse pro-
In an investigation comparing different impellers, Nagata portionality between the circulation times and the rota-
et al. (1972) also report some data on these impellers tional speed, two other conclusions are apparent from
with conclusions similar to those of Coyle et al. Figures 2 and 3. These are that (1) the average axial
Here the influence of elasticity on the performance of circulation is unaffected by the shear-thinning properties
this mixer was investigated. The kinematics were studied and ( 2 ) the viscoelastic anomaly brings about a significant
by following a small polystyrene particle (density about suppression of the axial circulation rate but simultaneously
lg/cm3, characteristic dimension lmm) . The liquids used
were translucent (density about lg/cm3) aqueous solu-
tions of Carbopol, CMC, PAA, and a mixture of Glycerine
and PAA solutions. The rheological properties of these
liquids were measured by Weissenberg Rheogoniometer.
In this viscometer, where the liquid sample is placed
between a flat plate and a small angle cone, both the
torque T and the total axial thrust F between the cone
and the plate are measured by sensitive transducers dur-
ing the shearing of the liquid. The torque T is related to
the variable viscosity 9 of the liquid by
7 = 3T/%R3j (1)
where + = d a is the shear rate in the conical gap ( a
being the gap angle and o the angular velocity of rota-
tion) and R is the radius of the cone-and-plate setup.
The axial thrust F can be related to the primary normal
stress coefficient of the liquid u1 which is the measure of
the liquid's elasticity
'TI = 2F/aR2 4 (2)
Both 7 and u1 were found to depend strongly upon the
shear rate y in the range 0.2s-1 < y < 50s-', and the
data were therefore interpreted by the following interpola-
tion formulae:
7 = k( + and u1 = h( y ) m - 2 (3)
The values of k, h, n, and m are given in Table 1 for all
the liquids used.
The transparent vessel made of perspex was used for
experiments. The impeller diameter (d1, in Figure 1 ) was

Correspondence concerning the note should be addressed to J. Ulbrecht, Fig. I. A combined ribbon-screw impeller.
AlChE Journal Wol. 27, No. 3)
May, 1975 Page 673
TABLE1. LIQUIDS
USED
k h n m
Aqueous solution of Batch no. (dyne/cmz s-n) ( dyne/cm2 s - n ) (-1 (--)
Edifas B 1500-CMC 1 298 - 0.49 -
2 305 - 0.47 -
Carbopol934 1 165 - 0.35 -
2 280 - 0.23 -
Separan APSO-PAA 1 220 140 0.31 0.80
2 260 240 0.17 0.73
PAA + Glycerine 1 100 280 0.38 0.65
2 260 1,500 0.27 0.51

PAA 1
0 Carbop1 2

D carbop1 1
PAA + Glycerlne I

I I L

0.05 0.1 1.0 5 .o

N s- N s-

Fig. 2. Axial circulation time. Fig. 3. Angular circulation timer.

a substantial increase of the angular circulation. In the


extreme cake (PAA dissolved in aqueous glycerine), the
value of NOc is almost equal to one which means that for
all practical purposes the liquid in the vessel exhibits an
almost solid-like circulation with the same angular velocity
as the impeller. On the other hand, the axial circulation
times are increased by a factor of between 2 and 6 for the
viscoelastic liquids. In all these measurements the Reyn-
olds number was less than ten.
The mixing time data were obtained using Iz-NazSz03
reaction. The line average of the color intensity was mea-
sured at a spot midway between the screw and the ribbon
and halfway between the surface and the bottom by a
photometric method. Since there were no stagnant zones
in the vessel, the point of measurement did not matter
much. The endpoint was selected so that the intensity
was 90% of the final expected intensity. The results are
shown in Figure 4. Not surprisingly the mixing time em
was again found to be inversely proportional to the rota- 0.01 0.1 1.0 2.0

tional speed N . The shear-thinning properties were found N 5-l


to have no influence upon the mixing time, but it took
Fig. 4. Mixing times.
much lancer to mix the viscoelastic liquids (PAA and
PAA with glycerine) than the inelastic ones. Although
the actual value of the experimentally determined mixing
time has little significance because of its dependence viscoelastic fluids will be about twice as long as in inelastic
upon the method used, the comparison of dimensionless liquids of comparable consistency.
mixing times obtained under the same experimental con- The movement of the suspended particle allowed US
ditions for different liquids will siqnify the influence of also to assess the flow pattern in the vessel. The primary
the material properties of the liquids investigated. Thus it flow pattern in a vessel agitated by a combined ribbon-
may be concluded that the dimensionless mixing times in screw impeller (Figure 1) is similar to that obtained with

Page 614 May, 1975 AlChE Journal (Vol. 21, No. 3)


a ribbon impeller. For the ribbon pumping downwards, n = material parameter, Equation (3)
the fluid adjacent to the wall of the tank travels axially R = radius, Equation (1)
downwards with a superimposed angular velocity com- S1 = pitch of the ribbon, Figure 1
ponent. At the bottom it turns inward and then flows Sz = pitch of the screw, Figure 1
upwards along the shaft again with some angular motion. t = tank diameter
At the surface it turns radially outwards toward the rib- T = torque, Equation (1)
bon. W1 = width of the ribbon blade
In highly elastic liquids, however, the axial component W2 = width of the screw blade
of velocity was very much suppressed in favor of the
Greek Letters
angular component. From time to time, the particle was
drawn radially into a recirculating vortex and was released a = angle of the conical gap
again. This phenomenon was observed for other impellers + = shear rate
as well and the work was recently summarized by Ulbrecht o = angular velocity
(1974). Consequences of such a flow pattern with not 1) = viscosity, Equation (1)
much intermixing between vortices in a polymerization e, = circulation time (axial)
reactor might be very serious not only for the molecular 8, = circulation time (angular)
weight distribution of the polymer produced but also for Om = mixing time
the potential side reactions. u1 = primary normal stress coefficient, Equation (2)

NOTATION LITERATURE CITED

Ci = circulation number Arumugam, M., M.Sc. dissertation, Univ. Salford, Salford,


dl = ribbon diameter, Figure 1 England ( 1974).
d2 = screw diameter, Figure 1 Coyle, C. K., H. E. Hirschland, B. J. Michel, and J. Y. Old-
F = axial thrust, Equation (2) shue, Mixing in viscous liquids, AIChE J., 16, 906 ( 1970).
Nagata, S., M. Nishikawa, T. Katsube, and K. Takaishi, Mix-
H = height of the liquid column, Figure 1 ing of highly viscous liquids, Intern. Chem. Eng., 12, 175
h = material parameter, Equation (3) (1972).
k = material parameter, Equation (3) Ulbrecht, I., Mixing of viscoelastic fluids by mechanical agita-
I = length of the impeller, Figure 1 tion, The Chemical Eng., No. 286, 347 (June, 1974).
m = material parameter, Equation ( 3 ) Manuscript received December 17, 1974; revision received February
N = rotational speed 28 and accepted March 3,1975.

Magnetofluidynamic Separation of a Binary Gas


E. P. DEL CASAL and JOHN J. McAVOY JR.
The Boeing Company, Seattle, Washington

The purpose of this technical note is to show the feasi- tromagnetically induced drag is exerted on the gas. The
- bility of separating the constituents of a gas mixture drag in turn generates a pressure gradient and thus also
through magnetofluidynamic means. This may be done a density gradient. Now, since the gas consists of an ion-
when the component that is to be separated has a sig- ized constituent and a neutral constituent, only the former
nificantly different ionization potential than the rest of is constrained by the magnetic field so that when the den-
the constituents. Although the analysis presented here may sity gradient is formed, the neutral constituent shall sep-
be applied to multicomponent mixtures, the discussion is arate from the ionized constituent by diffusion. The amount
limited to a binary system. The gas mixture may be pre- and degree of separation is dependent on the molecular
heated to a temperature range where the constituent with diffusivity of the binary mixture. The diffusivity is in turn
the lower ionization potential achieves a certain degree of dependent on the temperature and pressure.
ionization whereas the other remains un-ionized because The electromagnetically induced drag is generated when
of its higher ionization potential. When this mixture is al- the electrons and ions are forced to cross the magnetic field
lowed to flow in a magnetic field, the magnetic field acts via the pumping process, thereby causing a motional emf
selectively on the ions and electrons of the ionized con- Uelecx B to appear. In turn, a current J = aUeteeX B is
stituent. If in addition the field is set oblique to the flow, generated producing a force J X B that acts primarily on
there should occur a separation of the constituents. the electrons and opposes their flow across the field lines
The separation process as envisioned in this system con- (Cowling, 1957). The force J X B is the drag that is in-
sists of forcing the virgin gas into a constraining magnetic duced on the gas and is transmitted through collisions with
fieId where, by a process of selective pumping, a net elec- ions and neutral atoms. To achieve a desired amount of

AlChE Journal (Vol. 21, No. 3) May, 1975 Page 615

You might also like