ChemicalEngineering Science, 1967,Vol. 22, pp. 1689-1699.

Pergamon Press Ltd., Oxford. Printed in Great Britain.

Model studies on mixers in the viscous flow region

C. J. HOOGENDOORNand A. P. DEN HAKTOG

Koninklijke/Shell-Laboratorium, Amsterdam
(Shell Research N.V.)

(Received 3 May 1967)

Abstract-The batch mixing of viscous liquids with rotating agitators was studied in vessels of 0.24 and
1.8-m dia. The quantities measured were power input and mixing time. For mixing time measurement
a decolouration and a thermal-response technique were used, both yielding reproducible quantitative
data. Seven types of mixers were tested: a flat turbine, inclined-blade paddles, a helical screw, a
helical ribbon, a propeller and an anchor agitator.
For some agitators, in the region of truly viscous flow, the mixing time was inversely proportional
to stirrer speed. The mixers could be compared in terms of two dimensionless numbers based on
vessel diameter, liquid viscosity and density, mixing time and power input. The power required to
reach a certain mixing time was lowest for the helical screw in a draught tube, with inclined blades or
propellers in a draught tube a close second. The turbine and anchor mixers were unsatisfactory for
viscous mixing.

1. INTR~DOCTI~N fulfUed, we assume that-on a comparative basis

STIRRED vessels are widely used in the chemical
industry for mixing very viscous liquids, often
polymers or polymer solutions. Batchwise homo-
genization of such liquids may be required in
reactor vessels or in product blending vessels.
For this duty different types of rotating mixers
are in use, in many cases simple propellers, turbines
or anchor agitators. NAGATA et al. [1] and GRAY 2.1 Mixers tested
[2] have shown that a helical-ribbon agitator can
be very effective.
This diversity in types of mixers induced us to viscous fluids :
make a comparative study of agitators for viscous
mixing. Literature data on the subject [1, 2, 31
have one or more of the following shortcomings:
(a) mixing time is only vaguely defined;
(b) power consumption is not measured;
(c) each paper discusses only a few mixers,
the types chosen being different for each
author.
We therefore aimed at measuring a well-defined
mixing time as well as power consumption for a
large number of typical agitators. Our study was
restricted to the batch-mixing of Newtonian liquids
with equal density and viscosity. Although in
actual practice these conditions will often not be
-the results presented here will still hold.
Most of the work was done using small-scale
models, but for some of the mixers we verified the
results thus obtained by experiments in a large
vessel.
2. EXPW~MENTAL
Our teats were made with the following types of agitators,
all of which are applied in commercial mixing vessels for
(1) six-blade Rushton turbine [Fig. l(a)] with a d/D
ratio of l/3. It was used without batlles or with
four bafllles of 0.1 D;
(2) three inclined-blade paddles (fan-type turbines) on
one shaft [Fig. l(b)], each with two blades d/D=
O-50, blade width 0.25 d, blade angle 45”. pumping
direction downwards ;
(3) same as (2), but mounted in a draught tube of
051 D internal diameter [Fig. l(b)]. A few tests
were also done using only a single inclined-blade
stirrer in this draught tube;
(4) helical screw agitator [Fig. l(c)], single flight, pitch
1.15 d, d/D=O-4, pumping downwards;
(5) same as (4), but with a draught tube of special
design; internal diameter 0.46 D [see Fig. l(c)];
(6) helical-ribbon agitator IFig. l(d)], twin flight,
d/D=O*96, pitch 0.61 d, blade width 0.087 D;
(7) marine-type propeller A IFig. l(e)], three blades,
djD=O-47, pitch 1.25 d; pumping downwards in a
draught tube of 0.51 D intemal diameter;

1689
 C. J. Hoooe~~oo~~ and A. P.
(8) marine-type propeller B [Fig. l(e)], three blades,
d/D=O50, pitch 1.01 d; pumping downwards in a
draught tube of 0.51 D internal diameter:
(9 anchor agitator Fig. l(f)]; blade width &=0*083D,
gap width s=O*O167 D, blade thickness 0.0167 D.
#-
a b
FIG. 1. Types of mixers tested.
(a) Six-blade turbine
(b) Three inclined-blade paddles
(c) Helical screw
2.2 Test set-up
All the mixers were tested in a 024-m dia. vessel. The
liquid filling height was 1.0 D for agitators (l), (2) and (3).
l-25 D for (4) and (5), l-05 D for (6) and 1.0 for (7), (8) and
(9). A few tests with mixers (1) and (2) were done in a 1-8-m
dia. vessel.
The liquids were water-glycerine and water-molasses
mixtures with viscosities of 1 to 500 P in the small vessel
and a 50 P oil in the large vessel.
3. TEST METHODS
3.1 Determination of mi.xing time
In the literature, several methods to determine
the mixing time are described and the results so
1690
DEN HARnx,
obtained by the various authors [l, 2, 31 differ con-
siderably. This may be due to differences both in
measuring technique and in definition of mixing
(d> Helical ribbon
(e) Propeller in draught tube
(f) Anchor.
time. We investigated this by applying different
techniques and using only one definition.
In these techniques homogenization in a stirred
vessel is generally followed by a small amount of
liquid tagged with a tracer being added to the
vessel contents.
Different tracers are used, dependent on the
method to be applied, viz.
(i) the conductivity or pH method;
(ii) the thermal method;
 hkdel studies on mixers in the viscous3ow region

(iii) the colour addition method; by less than +(100-p)% of the total concentration
(t~~~tu~) step.
(iv) the decoiouration method.
With methods (i) and (ii) it is, however, only
In method (i) the conductivity or pH is higher possible to find 8,, which holds for the degree of
after each test and eventually the entire vesse1 homogeneity reached at the locations actually
contents have to be renewed. This is not the case measured.
in (ii), which we used in most of our tests. A small From the thermal response curves obtained by
quantity (l-2%) of heated (2040°C) liquid is method (ii) we derived mixing times 0, 5, 8,, and
added; during homoge~tion the tem~rature is e 95. Here e,5 (6 90, 8,s) is the time beyond which
measured with the~ocouples located at several at all measuring locations the differences between
positions in the vessels. This yields quantitative the temperatures measured and the final tempera-
information on the progress of mixing at these ture will remain smaller than f25 % (IO%, 5 %)
locations. of the total temperature step (i.e. the difference
To follow the mixing process visually throughout between the initial and final temperatures of the
the vessel, methods (iii) and (iv) are preferred. batch).
Addition of a dye (iii) to a clear liquid renders it . . .
This IS illustrated for 8,, and &, by some
possible to follow the initial stages very clearly. typical thermal response curves in Fig. 2. An
However, in the final stages, nearly the whole advantage of the thermal method is that for the
vessel contents are coloured and further changes positions measured it gives a uniquely defined
are then difficult to see. GRAY f2], who used this qu~titative value of the mixing time.
method, in general found lower mixing times than
NAGATAet al. [l] and ourselves; probably due to
this effect.
For good observation of the final stages it is
better to use a d~olouration method (iv). For
this purpose a fast chemical reaction in the presence Upper thermocouple
of an indicator can be used, for instance, the acid-
base reaction with phenolphthalein or the reaction
AT
between iodine and sodium thiosulfate in the ;ar,
presence of a starch solution. The experiment is
started with a coloured batch of liquid, which
gradually clears as mixing proceeds, stagnant zones
appearing as coloured patches while the bulk is
already clear.
In the decolouration method, say with a starch
solution, it is first necessary to add the iodine to
the liquid to colour it. Obviously, the iodine must
be homogen~usly mixed with the liquid; more
uniformly, in fact, than can be done with the stirrer
to be tested, because otherwise possible stagnant
zones are not coloured and will remain undetected FIG, 2. Turbine with bat&s. Thermal response
curvesat Re=476.
in the subsequent decolouration stage.
To define mixing time it is necessary to state the
required degree of homogeneity. In general, one With the decolouration method it is not possible
would like to define the mixing time er,p as the to reach equally well-defined values for the mixing
time beyond which the concentration of the tracer time. To approach this as closely as possible we
(or the temperature) in r% of the batch volume added a well-measured excess (33, 11 or 5 y0 above
differs from the final concentration (or temperature) stoichiometric) of decolourizing agent to the batch.

1691
 c. J. HCNIGENDWRNand A. P. DEN HIARTOG

Decolouration then occurred at places where the by the thermal method, the influence of the Schmidt
concentration differed by only + 25, -I-10 or + 5 % number would seem also to be negligible. This
from the final step. In this way it was also possible means that the model rule to be fulfilled is that Re
to find mixing times 0,5, es,, and eg5, being the is the same in the model and in the full-scale mixer.
time required for the bulk of the liquid to be In that case, the product of stirrer speed and
decolourized. There were, however, parts of the mixing time, no, will be completely fixed and the
liquid near fixed walls that often were not de- mixers can be studied by means of small models.
colourized and which were neglected. However, Generally, in such model experiments a given
large stagnant zones were taken into account. Reynolds number is reached with much less viscous
This renders the method more subjective. liquids than those occurring in actual practice.
It must be said that the BP-value found,
approaches the desired value t?,,P, with r=p, rather 4. RESULTS
closely, because, to bring the concentration in the 4.1 Turbine with bafles
bulk of the fluid to within +(100-p)% of the This widely used agitator has been thoroughly
final value, the tracer has to be spread over at tested. Figure 3 gives the experimental results for
least p % of the vessel volume. no,,. No differences were found between mixing
As will be shown in the discussion of the results, times for large-scale and small-scale equipment, as
the value for 075 obtained by the decolouration determined by the thermal method, which proves
technique was in good agreement with that from the the applicability of the model rules. Also, it is
thermal method, except in cases where large stagnant evident that the thermal and the decolouration
zones occurred. This shows that in general the differ- method give no significant differences in mixing time
ent techniques yield the same mixing time results if a at Re> 170. In the range of Re = 170-4000, n0 is
well-defined, unique mixing time definition is used. almost inversely proportional to the Re number.
3.2 Model rules
Mixing times can most simply be determined by
l Thermal, 1.8 m vessel
measurements in scale models. If the model is a
0 Thermal, 0.24m vessel
true copy of the full-scale mixer, the flow behaviour h Decolouration, 0.24m vessel
of Newtonian liquids is for both cases determined
in the same way by the Reynolds number:
Re =-. Pnd2
tl
When a vortex occurs, the Froude number is
also a governing factor:
F,,!!?.
9
Since in the experiments at low Reynolds number,
the region of main interest here, no vortex of any
importance was observed, the effect of Froude
number can be neglected.
The determination of mixing time by the de-
colouration method may, in principle, be influenced
by the Schmidt number:

Fro. 3. Turbine with batlIes. n89,~as a function of

Revnolds number for vessels 0.24 m and 1.8 m in
But since there was little difference between the diameter using the thermal and the decolouration
mixing times obtained by the decolouration and method.

1692
 Model studies on mixers in the viscous flow region
Figure 4 gives the results for no,, and n&,,; the
differences are not very large. At Re < 170 there is
a large scatter in the results, with very long mixing
times. Moreover, slightly shorter mixing times are
found by the thermal than by the decolouration
method. These phenomena appear to be related to
the occurrence of stagnant zones behind the baffles
in the upper region of the vessel, which in this
range of low Re numbers could be observed with
the decolouration method, but went undetected in
the thermal method. The n0 figures, given at the
top of Fig. 3 for the decolouration method, corres-
pond with the times required for these regions to
be decolourized also. These times are extremely
long and therefore the baffled turbine is not
adequate for obtaining complete homogeneity with
Re< 170.
FIG. 4. Turbine with and without baffles. n&s and
n&o as a function of Reynolds number for vessels
0.24 m and l-8 m in diameter using the thermal
method.
4.2 Turbine without b&Yes
To avoid dead zones behind baffles we have, in
a number of runs, omitted the baffles. In general,
mixing times were shorter (Fig. 4), but there was
another limitation. At Re < 150 very stable toroidal
1693
vortices were formed under and over the stirrer
tip. As a result, the time required for complete
homogenization increased sharply for low Reynolds
numbers. Consequently, this set-up is unsuitable
at Re< 150.
4.3 Three inclined-blade paddles on one shaft
Figure 5 shows the results obtained with this
mixer set-up. Here, too, nO,, increased sharply
with decreasing Re number in the low Reynolds
number range (300 to 40). In this region the
results of the thermal method showed a wide
scatter, both for the large-scale and for the small-
scale equipment. The decolouration method gave
mixing times equal to the longest times found by
temperature measurement.
l Thermal Mm vessel
o,Q Thermal 0.24 m vessel
aa Dccolourot~on. 0.24m vessql
. 0 A Without draught tube
Q4 With drought tube
5
With draughttube
Re
FIG. 5. Three
inclined-blade paddles with and with-
out a draught tube. r&s as a function of Reynolds
number for vessels O-24 m and 1.8 m in diameter
using the thermal and the decolouration method.
In the experiments with the decolouration
method, mixing was seen to proceed somewhat
compartmentwise; each paddle was surrounded
by a mixing zone with an intense radial flow. The
tagged liquid supplied at the surface was trans-
ported slowly downward due to only a small axial
flow and mixed up with the bulk of the liquid. This
phenomenon was especially marked in the region
around the lower paddle. Consequently, for the
 C. J. H~OQEND~ORNand A. P. DEN HARTOG

thermal method the position of the thermocouple 4.5 Helical screw

is very important. As the higher “thermal” values
of n6 agreed fairly well with the results of the
decolouration method, the latter were hence
assumed to be correct.
For Re< 20 a region was found in which,
according to the thermal method, nO,s was almost
constant. It could be seen that mixing took place
throughout the vessel, axial flow being pre-
dominant. However, now some toroidal vortices
appeared, which vanished very slowly, yielding
long mixing times by the decolouration method.
The results for a helical-screw mixer, given in
Fig. 7, show a strong increase in nt9, 5 with decreas-
ing Reynolds number. This agitator, too, creates a
partly stagnant zone in the form of a narrow and
high annular region around the screw.
‘\\ 0
‘,Screw
0 ‘\ -.

4.4 Three inclined-blade paddles with draught tube

In order to increase axial transport, the paddles
were surrounded by a draught tube. As a result,
the nf3,, values became distinctly lower, as shown
by Fig. 5. At low Re numbers they became con- Re
stant again. The thermal response curves and FIG. 7. Helical screw (with and without a draught
visual observation revealed circulation throughout tube) and ribbon. n&5 as a function of Reynolds
number, using the decolouration method.
the vessel, without considerable dead zones occurr-
ing. Homogenization proceeded everywhere and
0,5 was reached after about two to three circula- 4.6 Helical screw with draught tube
tions. This is shown by some typical thermal In view of the improvement resulting from the
response curves (Fig. 6). In Fig. 6 they are com- use of a draught tube with the inclined-blade
pared with the case where no draught tube was paddles it was also tested here, though in a some-
employed. what modified configuration. It effected a good

I r;\ r
I II
I II - Re=2+8
--- Re=lO

Re =393

I I I I J
0 too 200 300 400 500 0 100 200 300 400 500

n0 n8
FIG. 6 (a, b). Three inclined-blade stirrers with (a) and without (b) a draught tube. Thermal response
curves at various Reynolds numbers.

1694
 Model studies on mixers in the viscousflow region

circulation throughout the vessel and thus promoted
rapid homogenization (Fig. 7). The value of no,5
(in this case about 140) was again independent of
Re.
4.7 Helical-ribbon agitator
The results plotted in Fig. 7 show that for the
helical-ribbon agitator also no,, is independent of
Re. NAGATA et al. [I] and others came to the same
conclusion. We found an ntI,, value of 65;
NAGATA et al. [l] quote a value of 33 for nfI,,,
using a slightly different ribbon. GRAY [2] found
no =27 for this configuration, but he used the
colour addition method, which generally produces
unrealistically short mixing times.
the formation of a dead zone at the inside and out-
side of the tube. This part of the liquid circulated
without passing through the impeller plane, where
actual mixing occurs.
Figure 9 compares the results for propeller B in
a draught tube with those for three inclined-blade
paddles in a draught tube. At low Re numbers
fie,s values were equal.
As only a single propeller was used, we also
made a number of tests with a single inclined-
blade paddle in a draught tube. At low Re numbers
this agitator gave higher nf375 values than the
propeller (Fig. 9).

4.8 Marine-type propellers

Without a draught tube, large toroidal vortices
were formed in the lower Reynolds number range,
giving long mixing times (see Fig. 8) for propeller
B. With a draught tube, mixing was much better;
only a small toroidal vortex developed at the out-
side of the lower edge of the tube. Increasing the
clearance between propeller and draught tube again
led to very long mixing times (propeller A), due to
with drought tube

. Prop. A
With drought tube
0 Prop 0 I
c. Prop. B Without drought tube

FIG. 9. One inclined-blade paddle in draught tube

compared with three inclined-blade paddles and
propeller B, both in a draught tube. ~2075as a func-
tion of Reynolds number, using the decolouration
method.

4.9 Anchor agitator

Fro. 8. Marine propellers A and B with and without
a drau@ tube. 1~075as a function of Reynolds
number, using the decolouration method.
Mixing with this agitator was very unsatisfactory.
no75 increased strongly with decreasing Re number
(Fig. 10). The liquid rotated with the anchor, large
zones of secondary flow occurred outside the
agitator and no mixing was effected at low Re
numbers.

1695
 C. J. HOCK+ENDOORN
and A. P. DEN HAFUOG

FIG. 10. Anchor agitator. n&s as a function of

Reynolds number, using the decolouration
method.
FIG. 12. Power number curves.

4.10 Results of po wer measurements 5. COMPARISONOF MIXERS

Figures 11 and 12 give the curves of power
number, P,, vs. Re obtained from our measure-
ments. For all the mixers P, in the laminar region
is inversely proportional to the Re number. In
this region baffles have no affect on PO, but the
use of a draught tube considerably increases it.
The region of purely laminar flow is seen to extend
to Re numbers of about 10-100. The mixing time
measurements for the turbine agitator were made
at RezlOO (i.e. in the transition region between
laminar and turbulent flow) but even then this
stirrer was inadequate.
To decide which is the best mixer we must know
the power, P,,,, required to reach a specified mixing
time.
We have seen that for a number of mixers, in the
viscous region, the product ntJ was independent of
Re, while the power number in this region was
inversely proportional to Re. So:
nO=C,
and
P CZ
*=pnd2lrl-

After elimination of n, these equations give:

e2Pm=const
qz *
For a given mixer d/D is constant, from which it
follows that:

FIG. 11. Power number curves.
“Pm
3
=const -
The value of the constant is dependent on the
type of mixer and the vessel geometry. This means
that for the above-mentioned mixers the power per

1696
 Model studies on mixers in the viscous flow$egion

unit volume required to obtain a certain mixing
time is proportional to the viscosity and indepen-
dent of the size of the vessel. Furthermore, the
mixing time at a given viscosity is inversely pro-
portional to the square root of the power input per
unit volume.
However, for other types of mixers n0 is depen-
dent upon Re. Hence, if a mixer is to be chosen
which requires the lowest power input to reach a
specified mixing time for a liquid with viscosity r)
in a vessel with diameter D, the Reynolds number
should be taken into account. For this case it is
convenient to have a Reynolds number from
which the stirring speed and stirrer size, being
often very different for the various types of mixers,
have been eliminated. An adequate number is the
following group.
On this basis, the results for the mixers tested
have been compared in Fig. 13. It follows directly
that the mixers with a draught tube, such as the
three inclined-blade paddles (3), the screw (5) and
propeller B (8), require the least power. The
helical ribbon (6) is the next best. Propeller A (7),
having a large clearance between tip and draught
tube, is much less efficient.
These conclusions hold for the viscous region,
where
In this region the anchor (9), turbine (1) and
inclined-blade without draught tube (2) are com-
pletely inadequate and the screw without draught
tube (4) is poorer than the one with a draught tube.
GRAY [2] found that the screw without a draught
tube was better, but his data are in the range
Our data co&m his finding for that flow region.
For [p(D/8, &D/q]> 10 the trends are different,
but this is not the truly viscous region. Here the
turbine, propeller A and inclined-blade paddles
without a draught tube are satisfactory as well.
6. DISCUSSION
OF MIXINGMECHANISM
Visual observation during the process of de-
colouration-aided by a movie film made to
illustrate typical flow patterns-enabled us to study
the mode of action of the various agitators. On

z-

I$=7 2 la
9
6-
\
4-

?_ \\A \
‘I
,A6 6 \3aW\ \\
\
\

3,3inclined-blade paddles+drau httube

30.1 inclined-blade paddle t draug a t tube

:~~%draught tube

_,%er At draught tube 60. Propeller I3

RPropelier B-kdraught h&e 9. Anchor
I I l11111l I I Ill1111 1 I1111”’ ’ ’
‘4b_s f 4 6 at!p 2 4 6 SIQ z 468, .?

Fm. 13. Comparison of mixex performance.

1697
 C. 3. HOOGENDOORNand A. P. DEN HARTOG
the basis of this study it can be said that in general
the requirements for a good mixer are as follows:
(i) There should be a good axial flow
throughout the vessel (top-to-bottom
mixing).
(ii) All streamlines should pass through the
stirrer plane, since mixing in fact occurs
only near the agitator. Closed streamlines
outside this plane give stagnant zones.
(iii) Laminar wall layers should be disturbed
as much as possible.
Regarding point (i) simple agitators like the flat
turbine and the anchor are unsatisfactory because
they create apart from a rotational, mainly a radial
and only a small axial flow, which considerably
reduces the range of effectiveness. The same-
although to a lesser extent-is true for the inclined-
blade and propeller agitators when used without a
draught tube. The helical-ribbon mixer gives a
reasonable top-to-bottom flow, the stirrers with a
draught tube being the best in this respect.
Closed streamlines (point 2) outside the stirrer
plane were observed with all the agitators applied
without a draught tube. They formed stagnant
zones, varying from simple toroidal vortices with
turbine, inclined-blade and propeller mixers to
complicated geometries with the helical screw. A
draught tube takes the place of a stagnant zone,
which was clearly illustrated for propeller A. It
had a large gap between tip and draught tube. As
a result, streamlines closed around the tube,
passing through this gap without being mixed.
In the film the mixing action of three inclined-
blade paddles in a draught tube was illustrated by
introducing a small, dark, horizontal band of dye
into the annular space outside the draught tube.
When the agitator was started a nearly parabolic
velocity profile in this annulus became apparent
[Fig. 14(a)].
Reversing the pumping direction of the agitator
completely restored the horizontal band to its
original position [Fig, 14&l], as long as parts of
the dye band had not entered the draught tube.
When, however, parts were sucked into the draught
tube [Fig. 14(c)], they were diffused over a larger
volume [Fig. 14(d)].
1698
This leads to the following picture of the mixing
process. Owing to differences in velocity and/or
length along the flow paths a residence time
distribution is created around the mean circulation
time. Thus portions of liquid from different places
in the vessel meet near the agitator, where they are
mixed. This means dispersion outside and diffusion
inside the stirrer plane.
When a draught tube is employed the velocity
profile in the annulus is parabolic. Approximate
calculations for this case, based on the above
picture, yielded pulse response curves of the type
given in Fig. 6(a).
A completely different question is how the mixing
process, given by e,, changes withp. In some cases
we measured 6, 5 and 8, 5. It was found that mixing
could fairly well be represented by
&=l-exp(-COP).
Thus, from the &,-values given, any other value
for 6, can be derived.
7. CONCLUSIONS
The decolouration and the thermal method yield
the same mixing times except in cases where large
stagnant zones occur. The former technique is
more suitable for use on a small scale, giving a
good idea of the flow pattern as well. The latter is
more convenient for obtaining reproducible quan-
titative data for both small and large-scale work.
Mixers can be compared on the basis of two
dimensionless numbers, only incorporating vessel
diameter, liquid viscosity and density, mixing time
and required power; thus covering all scaling
effects.
It appears that, for viscous mixing, agitators
with a draught tube are the most satisfactory types
of equipment. The much-used anchor and flat
turbines are definitely inadequate. A helical-ribbon
agitator seems the most attractive proposition
when next to efficient mixing good heat transfer to
the vessel wall is important.
From our study the requirements of a good
mixer are found to be: good top-to-bottom mixing,
all streamlines passing through the stirrer plane
C d
 Model studies on mixers in the viscousflow region

+.nd disturbance of laminar wall layers. If these 9 acceleration of gravity

iequirements are fulfilled the product of stirrer pm power input
speed and mixing time is independent of Re number pcl power number
Cl,2 constants
in the viscous region. rl dynamic viscosity
These conclusions hold for batch mixing of 0 batch mixing time
Newtonian liquids differing only a small amount liquid density
in density and viscosity. F temperature
time
b; temperature difference in thermal method
Acknowledgement-Theauthors would like to thank Messrs AT
E. Verschuurand J. A. Wesselinghfor valuablecontributions m final temperature difference in thermal
to the experimentalwork. method

Indices to 8
NOTATION 75 value of p
d stirrer diameter 90 value ofj
D vessel diameter 95 value of p
5% diffusion coefficient p degree of homogeneity as definedon p. 1691
n stirrer speed r degree of homogeneity as definedon p. 1691

REFERENCES

[l] NAGATA S., YOK~GAMA T. and YARAGIMOTOM., A study on the mixingof high-viscosityliquid. Chem. Engng Japan
1957 21 278.
[2] GRAY J. B., Batch mixingof viscousliquids. Chem. JZngngProg. 1963 59 55.
[3] JENSENW. P. and TALIQN R. T. JR., Design of viscous-materialmixers, Paper 10.8.A.I.Ch.E.4. Chem. E. Joint Meeting,
London 1965.

R&uut&I_e mklangeageen discontinude liquidesvisqueux a l’aide d’agitateurs rotatifs a et6 etudie

dans des reservoirs d’un diam&re de 0,24 et de 1,8 m. Les grandeurs mesur6es sont l’energie absorb&
et la dur&e de melangeage. Pour la mesure de la dun% de m&tngeage on a utilis6 un pro&de de
d&coloration ainsi qu’une technique de reponse thermique, fo urn&ant tous deux de-s domrees quantita-
tives reproductibles. L’etude a compris sept types d’agitateurs, a savoir: un turboagitateur, un
agitateur a palettes inclir&s, un agitateur a vis toumante, un agitateur a bande helicoidale, un
agitateur il h&lice et un agitateur a ancre.
Avec quelques-uns de ces agitateurs, dans la region de 1’6coulement visqueux vrai, la d&e de
m&ngeage est inversement proportionnalle a la vitesse #agitation. Les melangeurs ont pu Ctre
compares a l’aide de deux nombres sans dimension bases sur le diam&re du reservoir, la viscosite du
liquide et sa densite, la dur6e de melangeage et l’energie absorb&z. La puissance n6cessaire pour
atteindre une certaine dun& de melangeage est la plus faible pour l’agitateur a vis toumante month
dans un tube, suivi de p&s de l’agitateur a palettes inclm6e.s ou l’agitateur a helice dans le tube. Le
turboagitateur et l’agitateur 51ancre ne conviement pas pour le melangeage de liquides visqueux.

Zusammenfassuug-Nichtkontinuierliches Vermischen viskoser Fliissigkeiten mittels rotierender

Rtihrer wurde in GefsSsen mit 0,24 und 1,8 m Durchmesser untersucht. Die aufgenommene Leistung
sowie die Mischdauer wurden gemessen. Zur Messung der Mischdauer wurden ein Entfarbungs-
verfahren sowie eine auf den Temperaturanstieg beruhende Methode benutzt, die beide quantitative
Damn mit ausreichender Reproduzierbarheit liefem. Die Untersuchung umfasste sieben Riihrertypen:
Turbmenriihrer, Schritgblatttigelrlihrer, Schraubenrtihrer, Bandrtihrer, Propellerrtihrer und Anker-
riihrer.
Bei einigen Rtihrem war im Gebiet des rein plastischen Fliessens die Mischdauer der Rtihrgesch-
windigkeit umgekehrt proportional. Mit Hilfe von zwei dimensionslosen Zahlen, die auf Gef&s-
durchmesser, ViskositB;t und Dichte der Fltissigkeit, Mischdauer und Leistungsaufnahme basieren,
liessen sich die unterschiedlichen Rtihrer vergleichen. Die zur Erreichung einer bestimmten Misch-
dauer beniitigte Leistung war fti den Schraubem-tihrer in einem Rohr am geringsten, ftir den
Schr&gblat-und Propellerruhrer im Rohr am zweitbesten. Die Turbmen- und Ankerrtihrem erwiesen
sich fur das Mischen viskoser Fliissigkeiten als ungeeignet.

1699