Chem. Eng. Technol.

15 (1992) 313-318 313

Mixing Times in Stirred Suspensions *

Matthias Kraume**

In stirred systems, the presence of solid particles leads to a pronounced lengthening of mix-
ing time, sometimes to over 10 times that for the single-phase state. The mixing behaviour
is strongly heterogeneous, since the slurry is mixed comparatively rapidly, while markedly
slower homogenization occurs in the solid-free zone. This is the consequence of different
fluid velocities in the two regions. For particle settling velocities in excess of 5 cm/s, mixing
times assume maximum values on reaching the state of complete suspension. By contrast,
at lower settling velocities, maximum mixing times occur before suspension is complete.
Mixing times are influenced only be the state of suspension and not by the mode of its
generation. Consequently, for the fulfilment of the 90% suspended slurry height criterion,
mixing times are independent of stirrer speed, solid concentration, type of agitator or
diameter ratio d/D. The effects of particle diameter, viscosity and equipment dimensions
on mixing time, when the 90% slurry height criterion is fulfilled, are reflected by the ratio
of the vessel diameter to a representative liquid velocity.

1 Introduction
Mixing of solid/liquid systems is a process frequently en-
countered in industrial practice. In the special case of
chemical reactors, in addition to suspending the solids, it is
often necessary to mix a liquid component into the contents
of the vessel. If the time constants of chemical reaction and
mixing process attain the same orders of magnitude, this
can favour parallel and consecutive reactions, which is
reflected in the corresponding losses of selectivity. A series
of precipitation crystallizations [I] provide typical examples
of these problems. Consequently, an optimum operation of
stirred solid/liquid reactors requires the knowledge of the
effects of solid particles on the mixing behaviour in stirred
vessels.
The physical processes occurring during the suspension of
solids have not yet been completely elucidated. Thus, for
example, a reliable calculation of the minimum suspension
speed is not yet possible [2], and this raises serious prob-
lems, particularly with regard to scale-up. In this connec-
tion, the investigation of homogenization in stirred suspen-
sions supersedes the assessment of the mixing behaviour
alone in offering the opportunity to analyze the fluid dy-
namics in this equipment.
2 Experimental Equipment and Procedure
The experiments were performed using three geometrically
similar tanks of different diameters (0.1 m, 0.33 m, 1 m),
which were operated at a constant liquid depth ( H / D = 1)
and with a constant stirrer clearance of the bottom of
* Paper presented at the GVC-Jahrestreffen, October 3 - 5 , 1990 in
Stuttgart.
** Dr.-Ing. M. Kraume, BASF AG, Technische Entwicklung, D-6700
LudwigshafedRh.
h / D = 0.17 (see Fig. 1). The vessels were equipped with four
baffles, 0.1 D wide and immersed to a depth of 0.8 D.
Propellers and pitched six-blade turbines were investigated
in addition to six-blade disk turbines of various diameter
ratios d/D. Three glass bead fractions with mean diameters
of 0.09 mm, 0.37 mm and 1.5 mm, were used in different
concentrations as the solid. The continuous phase was
either water or aqueous polyvinyl pyrrolidone solutions
with viscosities of 12 and 50 mPa s. Thus, the particle set-
tling velocity ranged between 0.22 and 20 cm/s.
Mixing time measurements were carried out by a
decolorization method [3], in which the reactor contents are
first coloured by alkali and phenolphthalein and subse-
quently decolorized again by adding an excess of acid. The
time from the addition of acid to the disappearance of the
last trace of colour inside the vessel corresponds to the mix-
ing time. A 2% excess of acid was used, corresponding,
under the selected conditions, to a degree of mixing of 98%.
Upon the conclusion of each mixing time measurement, the
actual excess was determined precisely by titration, in order
D Varied oarameterp:
*
D ~ 0 . ...
1 l m
d/ D 0.19 _._0.6
Disk turbine i propeller / pitched-blade turbine
Viscosity: 1 mPas, 12 mPas, 50 mPas
' Glass beads: 0.09 mm, 0.37 mm, 1.5 mm
Solids concentration: 1.6 ... 16 %
State of suspension
Measured auantities
Mixing time
Suspension speed
Fig. 1. Mixer geometry, measured quantities and varied parameters.

C VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1992 0930-7516/92/0510-0313 $03.50+ ,2510

314
to allow a correction [3] to compensate for the influence of
any possible deviations in concentration. Since the addition
site exerts a marked influence on mixing time [4], the acid
was always fed at the same point on the liquid surface in the
immediate vicinity of stirrer shaft.
In addition to the mixing time measurement, minimum sus-
pension speed was also determined visually. In this case,
both the criterion of complete suspension (Is criterion) and
that of suspended slurry height (Hslurry = 0.9 D )were ap-
plied to measure the speed of suspension. Furthermore, the
slurry height was determined visually for the various condi-
tions.
3 Experimental Results
For the understanding of the measurement results, it is
necessary to know the states of suspension occurring in a
stirred vessel, as illustrated in Fig. 2: as long as the stirrer
speed is inadquate, all the solids rest on the bottom of the
vessel (a). With rising stirrer speed, the particles are increas-
ingly lifted up by the circulating flow and suspended up to
a certain height (b). When the applied power input is suffi-
cient to prevent a relatively long-lasting settlement of par-
ticles, the state of the so-called “complete suspension” is at-
tainted (c). The measurement of the required stirrer speed
is normally performed using the I s criterion. This requires
that the solid particles should not settle on the bottom for
longer than Is. A further increase in stirrer speed causes the
fairly clearly defined interface between clear liquid and
slurry to rise. When the slurry height amounts to 90% of the
total liquid height, the slurry height criterion is fulfilled (d).
Finally, a still further increase in stirrer speed distributes the
solids throughout the entire liquid volume (e).
3. I Influence of Stirrer Speed
It is sensible to use the mixing time of the corresponding
single-phase system as a standard of comparison for the
qualitative and quantitative assessment of the effect of
solids on homogenization in stirred vessels. There are
numerous experimental results and equations available
[S - 71 relating to the blending of miscible liquids which are
generally well confirmed by our measurements. Under the
conditions set in this investigation, the product of stirrer
speed and mixing time always assumes a constant value,
dependent on stirrer geometry.
Fig. 2. Occurrent states of suspension
Chem. Eng. Technol. 15 (1992) 313-318
I
0 200 400 600 800 1000 1200 1400 llmin 1800
Stirrer speed n -
Fig. 3. Mixing time t M as a function of stirrer speed and solids con-
centration, at q = 1 mPa s.
Due to the presence of solids, there is a considerable varia-
tion in the mixing behaviour inside the apparatus. Fig. 3
shows the mixing time tM as a function of stirrer speed n
for the operation in absence of solids, as well as for solid
contents between 1.6 and 16 vo1.-Yo. The represented rela-
tionships are characteristic of the effects of the dispersed
phase on the total mixing process. The curves can be divided
into three regions:
as long as the stirrer speed is too low to suspend the par-
ticles, virtually no effect of the solid can be observed,
i.e. the mixing time coincides with that of a single-phase
system.
With incipient suspension of solids and simultaneous
formation of the slurry, mixing time becomes longer
than for the single-phase operation while stirrer speed
increases. Here, the final decolorization always occurs
inside the particle-free liquid layer. Depending on the
values of the other influencing parameters, the absolute
mixing time can increase substantially despite increasing
stirrer speed until it attains a maximum.
After the maximum mixing time has been exceeded, an
increase in stirrer speed leads to a marked drop in the
mixing time, which once again closely approaches that
of the single-phase system.
The reason for the observed behaviour lies in the variations
of fluid dynamics in the vessel, which are caused by the
dispersed phase. As long as the solid phase settles on the
bottom of the vessel, it can have a slight effect at the most,
by slightly modifying the contour of the circulating flow.
However, if the particles are suspended, a complete change
occurs in the flow field inside the whole apparatus. As rep-
resented diagrammatically in Fig. 4, there is an uneven dis-
tribution of solids within the slurry [8, 91 and a non-steady
circulating flow. The mean velocity of this flow is reduced
by the suspended solids in relation to the single-phase
operation. The upwards directed liquid velocity at the
boundary between the slurry and the clear liquid above is
approximately proportional to the particle swarm settling
velocity. Above the slurry, the liquid velocities are therefore
lower than in the single-phase state, as confirmed by veloci-
Chem. Eng. Technol. 15 (1992) 313 - 318 315

ty measurements [ 101. Since mixing times are determined by

the flow rates, the consequences of the altered fluid dynam-
ics are as follows:

- Complete mixing of the slurry is relatively rapid

(recognizable in the experiments by decolorization).
However, in comparison with the single-phase system,
the mixing time is longer because of the reduced circula-
tion velocity. These two findings are confirmed by the
results of Raghav Rao and Joshi [ l l ] , who measured

-
local mixing times in a slurry.
Reduction of velocity in the solid-free zone is con-
siderably more important, so that, in this region, com-
100 200 500 1 lmin

Stirrer speed at max mixing time n, -2000

plete mixing always occurs last. The approach of the
mixing times, found in the region of higher stirrer speed,
for two-phase operation to that for single-phase (see
Fig. 3) results from the increasing height of the slurry,
which leads to a reduction in the volume of the clear liq-
uid. Consequently, the paths traversed there by the liq-
uid flow also become shorter, and thus the mixing time
becomes shorter as well. If the solids are distributed
throughout the entire volume of the liquid, the region
especially critical for complete homogenization disap-
pears and the mixing time approaches that of a single-
phase system.
Fig. 5. Stirrer speed at maximum mixing time n, versus that at com-
plete suspension njs.
If the settling velocity of solid particles is less than 5 cm/s,
the maximum mixing time is displaced into the region of in-
complete suspension (see Fig. 6), since a stable slurry is al-
ready formed in this state. A further increase in stirrer
speed causes an extension of the slurry height, and thus a
decrease of the upper vessel region, which is critical for
homogenization, and this is followed by a reduction in the
homogenization time.

Therefore, the maximum mixing time occurs when the stir-
rer speed is just sufficient to produce a uniform slurry, since
then the region of the liquid free from solids is at its largest.
Fig. 3 also shows mixing times tM,js, measured in the state
of complete suspension. The corresponding suspension
speed njs is the same as that associated with maximum mix-
ing time n,. This identity of complete suspension and max-
imum mixing times is observed in all experiments in which
the particle settling velocity exceeds approximately 5 cm/s.
Fig. 5 confirms this agreement for different types of
agitators, concentrations of solids, particle sizes and vessel
diameters.
For energy reasons, stirred suspension vessels are in practice
frequently dimensioned in such a way that complete suspen-
sion is just guaranteed. According to the presented results,
this procedure must be altered when designing chemical
reactors, if the increased mixing times lead to losses of
selectivity.
Fig. 6 illustrates the dramatic influence of solids on mixing
time. Thus, the maximum mixing times are at least 10 times
longer than those for the single-phase system. It must
always be stressed here, however, that this increase occurs
only in the liquid free from solids. This heterogeneity of the
mixing process can be detected by the measurement tech-
nique only if homogenization is observed in the entire ap-
paratus, as is the case with the decolorization method. By
contrast, local measurements, for example, by means of
probes, lead to inconclusive results. Thus, Raghav Rao and
Joshi [l 11, who measured mixing times in stirred suspen-
sions by using a probe attached near the agitator, obtained
much shorter mixing times since, due to the positioning of
the probe, the critical region near the surface was not
covered. Therefore, these results do not reproduce the
homogenization in the whole liquid content of the vessel.
250
5

200 '-
n
1
I
150

Clear liquid
(Region 01 final-
g
.- 100
decolorlzatlonl 0)

Slurry 3 50
ghl-depndent
ntrallon of so11ds)

0
0 200 400 600 800 lirnin 1200

Stirrer speed n -
Fig. 6. Mixing time tM as a function of stirrer speed and solids con-
Fig. 4. Flow field inside a stirred suspension. centration, at q = 12 mPa s.
316 Chem. Eng. Technol. 15 (1992) 313-318

3.2 Relationship between the State of Suspension

and Mixing Time
In the following, the relationship between the mixing time
and the state of suspension, already noted in Figs 3 and 6,
is examined in more detail. Mixing times are obtained from
the liquid flow path and from the associated velocity. In the
case of a stirred suspension, it is necessary to use path
length and velocity in the liquid layer free from solids,
which is critical with respect to complete mixing. The criti-
cal path is thus proportional to the height of the single-
phase region. The flow rate in this zone depends directly on
the particle settling velocity since, in order to suspend
solids, fluid velocities in the interface must exceed this set-
tling velocity. Therefore, no information is required on the
stirrer speed necessary to generate the corresponding flow
field; only the velocity field, corresponding to a specific
state of suspension, is considered.
3.2.1 Mixing Times for Complete Suspension
Fig. 7 shows mixing times for the state of complete suspen-
sion tM,js which were measured in a vessel using a propeller
and a disk turbine for different solids concentrations and
viscosities. For the propeller, the mixing times tM,jsare in-
dependent of solids concentration, although the required
minimum suspension speed varies by approximately a fac-
tor of 2. A simultaneous observation of the associated
slurry heights shows that these are almost identical. Since
both the mixing time and the critical path are constant, the
characteristic liquid velocity must likewise assume a fixed
value which, however, decreases with increasing viscosity.
In the boundary region between the slurry and the single-
phase liquid, the velocity of the upwards directed flow is
proportional to the particle settling velocity. This propor-
tionality must therefore also exist between the mean liquid
velocity in the region free from solids and the settling
velocity. The influence of viscosity on mixing time thus
results from the dependence of particle settling velocity on
viscosity.
When a disk turbine is used, the mixing time increases with
solids concentration. This is attributed to the decrease in the
s I
I
I
I D=033rn
d/D=03 032
h/D - 0 17 I
I 1 I
1 2 5 10 % 20
Solids concentration 9" -
Fig. 8. Mixing times tM,90vofor fulfilment of 90% slurry height crite-
rion as functions of solids concentration pv, for different particle sizes
and agitators.
slurry height, occurring with increasing solids concentra-
tion, when complete suspension has been achieved, which
causes the critical path and thus also the mixing time to in-
crease.
The dependence of slurry height on solids concentration
cannot at present be described by physically based model-
ling. Since the knowledge of the critical path is a prere-
quisite for the estimation of mixing times, the latter cannot
be described for the state of complete suspension by a
physically based expression.
3.2.2 Mixing Times for 90% Suspended Slurry Height
It is already clear from Figs 3 and 6 that, when the 90%
suspended slurry height criterion is fulfilled, the mixing
times tM,90% are independent of solids concentration
despite increasing suspension speed. By contrast, as Fig. 8
shows, the mixing time decreases with increasing particle
diameter and thus with rising particle settling velocity. This
is also to be expected in accordance with the proportionality
between liquid velocity in the solid-free zone and settling
velocity.

I +,' D=033rn
d / D = 0 3 032
d, = 0 3 7 m m
h/O = 0 17

1 2 5 10 % 20

1 2 5 10 % 20 1 2 5 I 0 % 20 Solids concentration 'P, -

Solids concentration Ip, - Solids concentration 9" - Fig. 9 . Mixing times tM,wvo for fulfilment of 90% slurry height crite-
Fig. 7. Mixing times tM,js at complete suspension as functions of rion as functions of solids concentration pv, for different agitators
solids concentration pv for different agitators and viscosities. and viscosities.
Chem. Eng. Technol. 15 (1992) 313-318 317

Since, due to the selected suspension criterion, the height of

the single-phase liquid is equally independent of the type of
agitator as the velocity distribution occurring there, the dif-
ferent agitators lead to identical mixing times.

An analogous interrelationship arises with respect to a

change in the liquid viscosity. The mixing times become lon-
1
ger with increasing viscosity and thus with decreasing set-
tling velocity (see Fig. 9). As in Fig. 8, these measured
values once again illustrate the independence of the mixing
time tM,90vO from solids concentration and type of agitator
and thus, overall, from the stirrer speed.
0.2 05 1 2 s 5

The independence of the mixing time tM,90q0 from the
diameter ratio d / D , as shown in Fig. 10, follows from the
fact that, upon fulfilment of the slurry height criterion, liq-
uid velocities, decisive for homogenization, depend only on
the particle settling velocity. Since, for the selected particle
concentration of 10% by volume, the expansion of the
slurry is virtually independent of the diameter ratio and
type of agitator in the case of complete suspension, the mix-
ing times tM,jsof this suspension state are also of the same
order of magnitude. The larger height of the region free
from solids, when suspension is complete, causes a marked-
ly longer mixing time than that required for the fulfilment
of the slurry height criterion.
The effect of the vessel diameter on mixing time tM results
from the change in the critical path, which must be tra-
versed by the liquid flow. Since this distance is directly pro-
portional to the equipment diameter, the mixing times
become longer with increasing vessel diameter for identical
particle settling velocities.
4 Physical Description of Mixing Times
The physical interpretation of mixing times based on the ex-
tension of the solid-free region and on the mean liquid
velocity prevailing there, permits these times to be de-
scribed. However, for this purpose, the slurry height rele-
vant to the respective stirrer speed must be known, which is
D -
-3
Fig. 11. Representation of mean mixing time for fulfilment of 90%
slurry height criterion.
not generally the case. This precisely defined state is
established when the 90% suspended slurry height criterion
is applied, so that modelling can produce reasonable
results. The flow field and the far-reaching distribution of
solids throughout the vessel suggest an analogy to a two-
phase propeller loop reactor. The circulation velocity of the
liquid phase necessary to suspend the solids is known from
the flow conditions there [12]. Transferring to the condi-
tions in the stirred vessel, it is possible to derive a mean min-
imum velocity of the circulating flow of t )
I
However, in contrast to the loop apparatus, there are large
local differences in the concentration distribution in stirred
vessels [8, 91 and, particularly at the height of the interface
between the slurry and the clear liquid, the concentration of
solids is markedly reduced. Consequently, the particle
velocity w, is to be used in Eq. (1) instead of the swarm set-
tling velocity, and the solids concentration is to be
eliminated. The following proportionality is therefore ob-
tained for the mean circulation velocity in the zone free
from solids:

100 D-033m
0 0
Since the extension of the clear liquid layer is proportional
S
0 Complete suspension to the vessel diameter, the following must be valid:
! 50
0
0
U 0 hlD = 0 17
DtSk turbine
Pitched blade turbine 0
c Propeller (3)
a,
5
m
; 20 0 -8 0 . . .. ..
This relationship is illustrated in Fig. 1 1 by a plot of the
2 0 0 @ 90% slurry height'
mean mixing times fMM,wvo,measured for different solids
concentrations and all the parameters investigated, against
10
01 02 03 04 05 06 D / ( ~ D W ~ The
) " ~graph
. shows the derived direct propor-
Diameter ratio dlD - tionality between the two quantities, since all the values can
be approximated by a regression straight line with a slope
Fig. 10. Mixing time tM for complete suspension and fulfilment of
90% slurry height criterion as a function of diameter ratio d/D for dif-
ferent agitators. 1) List of symbols at the end of the paper.
318 Chem. Eng. Technol. 15 (1992) 313-318

of unity. The mixing times in the case of fulfilment of the mixing time for complete suspension
slurry height criterion can be described by the following mixing time for 90% slurry height criterion
equation: mean circulation velocity
settling velocity of single particle
settling velocity of swarm
33 D solids concentration
(4) liquid density
solid density
difference in density
viscosity
Received: June 27. 1991 [CET 4051 drag coefficient

References
Symbols used
Gbsele, W., Kind, M., Chem.-Ing.-Tech. 63 (1991) pp. 59-62.
vessel diameter Kraume, M., Zehner, P., Chem.-Ing.-Tech. 60 (1988) pp.
agitator diameter 822 - 829.
particle diameter Hiby, J.W., Chem.-Ing.-Tech. 51 (1979) pp. 704-709.
gravitational acceleration Stein, W.A., Chem.-Ing.-Tech. 59 (1987) pp. 750-751.
liquid depth Henzler, H.-J., VDI-Forschungsh. (1978) No. 587.
depth of immersion of agitator baffles Zlokamik, M., Chem.-Ing.-Tech.39 (1967) pp. 539- 548.
slurry height Joshi, J.B., Pandit, A.B., Sharma, M.M., Chem. Eng Sci. 37
stirrer clearance (1982) pp. 813-844.
stirrer speed Niesmak, G., Thesis, Univ. Braunschweig 1982.
stirrer speed at maximum mixing time Einenkel, W.-D., VDI-Forschungsh. (1979) No. 595.
minimum stirrer speed for complete suspen- Geisler, R . , Mersmann, A., Proc. 6th European Conf. on Mix-
sion ing, Pavia 1988, pp. 267-272.
width of agitator baffles Raghav Rao, K.S.M.S., Joshi, J.B., Chem. Eng J.
distance between baffles and wall (Loughborough, Engl.) 39 (1988) pp. 111 - 124.
mixing time for 98% mixing quality Zehner, P., Fortschrittsber. VDI, Reihe 3 (1988) No. 160.