Grace 1982

This article was downloaded by: [University of Ottawa]
On: 13 March 2013, At: 09:01

Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK
Chemical Engineering Communications

Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gcec20
DISPERSION PHENOMENA IN HIGH VISCOSITY IMMISCIBLE

FLUID SYSTEMS AND APPLICATION OF STATIC MIXERS AS
DISPERSION DEVICES IN SUCH SYSTEMS
a
HAROLD P. GRACE
a
Engineering Technology Laboratory, E. I. DuPont de Nemours & Co., Inc., Wilmington, DE,
19898
Version of record first published: 03 Apr 2007.
To cite this article: HAROLD P. GRACE (1982): DISPERSION PHENOMENA IN HIGH VISCOSITY IMMISCIBLE FLUID SYSTEMS AND
APPLICATION OF STATIC MIXERS AS DISPERSION DEVICES IN SUCH SYSTEMS, Chemical Engineering Communications, 14:3-6,
225-277
To link to this article: http://dx.doi.org/10.1080/00986448208911047
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Chem. Eng. Commun. Vol. 14, pp. 225-277 Q Gordon and Breach. Science Publishers. Inc.. 1982
0098-6445/82/1403-0225$06.50/0 Printed in the U.S.A.
DISPERSION PHENOMENA IN HIGH VISCOSITY

IMMISCIBLE FLUID SYSTEMS AND
APPLICATION O F STATIC MIXERS AS
DISPERSION DEVICES IN SUCH SYSTEMS
HAROLD P. GRACE?
Engineering Technology Lnborarory
E. I. DuPont de Nemours & Co., Inc.
Downloaded by [University of Ottawa] at 09:01 13 March 2013
Wilmingron. DE 19898
(Received March 3. 1981)
I will discuss dispersion phenomena in high viscosity immiscible fluid systems and
application of static mixers as dispersion devices in such systems in the following four
steps:
(i) Single drop deformation and breakup in rotational shear fields (Couette).
' (ii) Single drop deformation and breakup in irrotational (extensional) shear fields.
(iii) Immiscible jet stream breakup upon injection into laminar flowing stream.
(iv) Application of (i)-(iii) to interpretation and prediction of dispersion perform-
ance at high viscosity in static mixers, at both low and high viscosity ratios.
The information presented here is based on studies in model fluid systems of two
phase liquids of high viscosity under creeping flow conditions, i.e., where the contin-
uous phase viscosity was from 50 to 3000 poise, and very nearly Newtonian at the
shears involved, the particle Reynolds number was in the range from to the
interfacial tension varied from a fraction of a dyne to 25 dynes/cm, and the dispersed
phase to continuous phase viscosity ratio varied from to 950.
Areas of application pertain to high viscosity polymer melts o r solutions intended
for fiber, film o r bulk plastic production. These include the dispersion of immiscible
t This paper was presented by Dr. H. P. Grace, at the time Senior Research Fellow with E. I. DuPont de
Nemours Co., at the "Third Engineering Foundation Conference on Mixing" held in Andover (N.H.) in
August 1971. Although the paper was never published, many copies of the presentation (some of them of
third and fourth generation) have been in circulation for years and the work has been often quoted. It was
felt, therefore, that by publishing Dr. Grace's work, we shall meet the needs of those who have been trying
to get a copy of this work. Since Dr. Grace retired from active service, Dr. David G.R.Shon. Engineering
Consultant with DuPont, retrieved the manuscript and the figures and reissued both in March 1981. Suh-
sequently, the paper was edited for publication in ChemicalEngineering Communiculions. When editing the
manuscript. I changed mainly the references to slides and to movie sequences which, unfortunately, are no
longcr available. Some slides were combined with the text and some had to he left out since the photo-
graphs shown on them were not of good quality. Some figures had to be redrawn either because the data
points were not legible o r because the axis description was missing. I have also filled-in thc literature refer-
ences which were missing in the original manuscript. I have preserved the original terminology and nota-
tion. The paper was not subject to our normal peer review hut the high demand for copies of this presenta-
tion, which has persisted for over ten years, testifies to the value of Dr. Grace's work. -1. I. Ulhrecht.
Editor
226 H.P. GRACE
Y
Inner bowl I
FIGURE 1 Simple shear
droplets of antistatic o r antisoiling agents of much lower viscosity, the dispersion of

droplets of color concentrates originally prepared in an immiscible polymer of lower
viscosity, the breakup and dispersion of unwanted immiscible gel structures of higher
viscosity resulting from cross-linking o r polymer degradation, and finally the blend-
ing of immiscible polymer systems to form controlled two phase structures of unique
physical, optical or electrical properties.
Let's start out by considering single drop deformation and breakup in rotational
shear fields, say a uniform Couette shear field. Figure 1 shows such a rotational shear
field with an initially spherical drop deformed to an ellipsoidal shape, where the drop
deformation D equals length minus breadth over length plus breadth.
L-B
Drop deformation, D = - (1)
L+B
Viscosity ratio, to be referred to frequently, is defined as dispersed phase viscosity
over continuous phase viscosity, and f@) is a theoretical function of this viscosity
r l varies only from 1 to 1.2 for viscosity ratios from 0
ratio suggested by ~ a ~ l owhich
to infinity.
Viscosity ratio, p = p o / p , f@) =

19 + 16
16p + 16
For small deformations, Taylor1 showed theoretically that the deformation D is
equal to E, where E is a ratio of viscous to interfacial tension forces necessary for
burst.
DISPERSION PHENOMENA 227
This will be termed the reduced shear required for breakup, and is shown here as the
shear rate, Gn, times drop radius, r d , times continuous phase viscosity, pe, times func-
tion, A p ) , over the interfacial tension, y. Experimentally, as on' demonstrated,
under certain conditions, that the deformation at burst equals E a t burst is in the
range of 0.5 to 0.6 for viscosity ratios of 0.1-1.0.
where
At very high viscosity ratios, ~ a ~ l opredicted

r' a limiting deformation without burst
in simple shear
Drop deformation and burst have been studied in a Couette device, the principles
of which are shown in Figure 2. The viscous fluid in the annulus is floated on a heav-
ier low viscosity fluid to reduce end effects, single drops being inserted at the point of
shear reversal for ease of observation. Variable speed drive on both inner and outer
bowls provided uniform shear gradients from 0.001 reciprocal second to 100 recipro-
cal seconds.
The plot in Figure 3 shows that, at small deformation, the relationship of deforma-
tion to viscous forces deforming the drop is linear in accordance with small deforma-
tion theory, the slope providing a direct measure of effective interfacial tension. Whereas
at larger realistic shear gradients, the curve is n o longer linear. The value of the inter-
facial tension computed from the linear portions of such deformation curve agrees
well with independently measured values provided the latter are extrapolated to a
zero quiescent contact time at the interface.
Figure 4 shows this typical effect which results from the very slow diffusion in such
viscous systems and a finite but very small mutual solubility. Thus, under shear the
interface is dynamic, fresh components are in contact, and interfacial tension acting
is that between the pure components for an extended period for high viscosity
systems.
Figure 5 shows the very great effect of viscosity ratio on E, the ratio of viscous to
interfacial forces required for drop burst in Couette shear. The correlation shown
here spans the range of continuous phase viscosity from 45 t o 2800 poise and interfa-
cial tensions from a fraction of dyne/cm to about 25 dynes/cm. This general correla-
tion allows prediction of the critical bursting shear gradient, GB,required to break a
drop of radius, r d , at a continuous phase viscosity, p,, and a n interfacial tension, y,
for any viscosity ratio from to 3.5 in uniform rotational shear. It can be seen
H.P. GRACE
Outer
bowl
drop of
dispersed
Uniform shear field 2

w Observer
Viscous
continuous-phase?
fluid
,,
'. .': ,..
Heavier immiscible
41b
.. . .
, 5
.: . . .. . . - .
; \
,
L','
Drop of immiscible
dispersed-phase
fluid
fluid of low viscosity Lamp

FIGURE 2 Principles of Couette device.
from the curve that for actual fracture of the drop, a minimum occurs in the range of
viscosity ratios from 0.1-1.0, whereas at either lower viscosity ratios o r higher viscos-
ity ratios the value of the ordinate, and therefore the critical shear gradient required
to break, increases very rapidly. The horizontal portion of this curve, below a viscos-
ity ratio of 0.1, represents tip streaming breakup [when small drops are shed off
O.~I/L
0.2
Linear Region
Grd/Lcf(p)
rJ=-
v
/Lo ~3.7
c
poise
p ~ /y=6.8dynes/cm
g . =0.0037
Velocity Grodient Across Drop,Grd,cm/sec
FIGURE 3 Drop deformation.

Vorite 125 1 Si 1OOO System
vc - a315 poise. uo - 10.5 poise. voIpc - 3.751101-~
O Value from Small Deformation Measurements in Couette Shear
0 Values from du Nuoy Ring Tensiometer Measuremants
Measured Value
01 Interfacial
Tension, 7. -
dynes I cm
2 -
0 I I I
0 100 2W MO 400
Quiescent Contact Time to Measurement, min
FIGURE 4 Effect of quiescent contact time on measured interfacial tension.
from the two tips of a mother drop which assumes a sigmoid shape. ~ d .and ] does not
represent actual fracture of the drop as is required in short-term dispersion. I t can be
seen that, at very low o r very high viscosity ratios, it becomes several hundred times
more difficult to break a drop by uniform rotational shear. Because of this and a
large interfacial tension, dispersion of a gas in a high viscosity liquid becomes a very
00 pC= 45.5 poise

AA p, = 502 poise
VA p, = 2.815 poise
Drop breaking by
Viscosity ~ a t i o ,p = p D / p C
FIGURE 5 Effect of viscosity ratio.

H.P. GRACE
FIGURE 6 Critical shear gradient.
difficult operation. Also, of course, it is not possible to break drops in such a simple
shear field if the viscosity ratio exceeds 3.5.
Figure 6 shows some similar data reported previously by Mason and co-workers.'
Most of the data points for actual fracture of the drop were in the viscosity range
from a 0.1 to 1. No data were taken at viscosity ratios above 1. All the data at vis-
cosity ratios below 0.1 were for a tip streaming breakup rather than actual fracture of
the drop. As reported, Mason drew a straight line of constant values through these
points which we do not consider to be valid for actual drop fracture.
Figure 7 shows similar data reported by Karam and ~ e l l i n ~ e rhe. ~curve of these
data is quite similar in several respects to that of our data. That is, there is a min-
,001 .01 0.1 1.o 40

viscosity Ratio, p , /p,
FIGURE 7 Critical shear gradient. Karam and Eellinger.

DISPERSION PHENOMENA
imum in a viscosity ratio range of 0.1 to 1 with a rapid increase in what they term the
reduced shear rate (our term EB),at either low o r high viscosity ratios. However, the
two significant differences are, first, that we were able to actually fracture drops at
viscosity ratios as low as whereas Karam and Bellinger indicated this was not
possible below perhaps 5 X lo-'; this we feel was due to limitations of their equip-
ment and the continuous phase viscosities used which were much lower than those we
used. The second principal difference is the numerical value at the minimum of this
curve. Our data indicate it is in the region of 0.6 which checks with Mason's available
data. In contrast, Karam and Bellinger show a minimum a t a value of reduced shear
rate of about 2. This we feel is due to a difference in the interfacial tension value used,
i.e., whereas we have used a value of dynamic interfacial tension between fresh com-
ponents under shear, the value used by Karam and Bellinger is a long-term equilib-
rium value between mutually saturated components which could be expected to be

1/2 to 1/3 of the value actually representative of the dynamic shear conditions. We
feel the value of dynamic interfacial tension between pure components is more repre-
sentative of a n actual dispersion condition where two components are brought to-
gether and must be very rapidly dispersed in continuous equipment, under very short
contact times.
Figure 8 shows that the shape of our correlation of critical shear gradient versus
viscosity ratio corresponds to the shape of the curve of drop deformation versus vis-
cosity ratio in many respects. That is, a minimum in curve of required drop deforma-
tion for breakup occurs in the viscosity range from 0.1 to 1, and at either higher o r
lower viscosity ratios the drop deformation required for drop breakup becomes
greater and greater. Of course, on the high viscosity ratio side, upon reaching a vis-
cosity ratio of 3-1/2 a discontinuity occurs and suddenly only a limited deformation
can be achieved under uniform rotational shear, with no breakup occurring.
Figure 9 shows deformation data expressed as the drop draw ratio necessary for
break at the critical shear gradient, i.e., the length required for breakup over the orig-
I I I I I I
C U)
-
0
.- a
U)
,
0 E Z
a o n
- For Continuous-Phase Viscosity p, = 2.815 Poise
g x0 ,
& -
. z m
; .-
_Ir
0.4-
0.2- -
Viscosity Rotio, pD/pc,

Dimensionless

232 H.P.GRACE
Drop Draw -Ratio 45.5 poise cont.- phase viscosity
Necessary for A 502 l1 ,I
Breok at E, A2,815 " I, I# I
20
11 I U ~ I rn 1 ' ' 1 ' 1

10" id' 16' id" 46' lo' i 10
Viscosity Ratio, p,/pc ,dimensionless
inal drop diameter, plotted versus viscosity ratio. This again for continuous phase
viscosities from 45 to 2800 poise and for interfacial tensions from a fraction of a dyne/
cm t o about 25 dynes/cm. The general correlation is similar, with a minimum occur-
ring in a viscosity ratio range of 0.2-1.0, and a rapid rise in either direction from this
minimum, with required drop draw ratio approaching one hundred at either very low
viscosity ratios o r viscosity ratios approaching 3.5. For breakup to occur at the criti-
cal value of reduced shear, both a critical drop draw ratio and a critical time under
shear must be attained.
Figure LO shows a dimensionless correlation of the data in terms of a reduced burst
time versus viscosity ratio, from which the time, is, required under critical shear can
be predicted. Again this correlation covers continuous phase viscosity from 45 to
2800 poise and interfacial tensions from a fraction of a dyne/cm to 25 dyneshm.
The correlations of the past several plots allow us to predict, for a lo6 fold range of
viscosity ratio, the required critical shear, the required drop draw ratio, and the re-
Viscosity Rotio, ~ O / , L L ~dimensionless

,

quired shear time for breakup of a given drop size in a uniform rotational shear field.
However, in a practical dispersion problem we have a distribution of feed drop sizes,
and therefore in order to equal the critical value of reduced shear and assure breakup
of the smaller feed drops we will in effect exceed the critical shear for the larger size
drops.
Figure I I shows the effect of exceeding the critical shear on drop draw ratio at
burst, for four different viscosity ratios. It can be seen that, as the ratio of actual to
critical value of reduced shear (i.e., E/EB) increases, the required drop draw ratio
which must be reached before burst will occur increases quite rapidly.
-
-
-
E-TO-CONTINUOUS
E VISCOSITY -
-
A 0.107
0 I .69 (lo)-* -
0 1.79(10)-~
1.78 (lor4
-
--
VORITE 125 CONTINUOUS PHASE --
p,= .2,815 POISE
-
DISPERSE PHASE
INTERFACIAL TENSION
-
y = 5.9-6.3 dynes/cm
-
FIGURE 11 Effect of exceeding En on drop draw ratio at burst

234 H.P. GRACE
v,
I-
Z A 0.107
W
5 0 1.69 (lo)-'
w
a 0 1.79 ( l o r 3
LL
LL o 1.78
VORITE 125 CONTINUOUS PHASE

pc = 2,815POISE
DISPERSE PHASE
y = 5.9 -6.3 DYNES /CM
FIGURE 12 Effect of exceeding Ea o n number of drop fragments at burst.
Figure 12 shows the effect of exceeding the critical shear on the number of drop
fragments produced when burst does occur. It can be seen that, when the ratio of ac-
tual to critical value of reduced shear is one, the number of fragments produced de-
pends only on the viscosity ratio and the type of burst occurring; also that, as the
ratio of E to EBincreases, the number of drop fragments increases very rapidly reach-
ing values on the order of 10,000 fragments a t reduced shears on the order of 100
times the critical shear, with the breakup behavior becoming much more dependent
on ratio of E/EB than on viscosity ratio.
Figure 13 shows the effect of exceeding the critical value of reduced shear on the
reduced burst time, t ~required
, for drop breakup. The time under shear required for
breakup decreases quite rapidly as the ratio of actual t o critical values of reduced
shear exceeds one. However, this decrease is less than proportional to the increase in
value of the ratio E/Ee. Consequently, the energy input required is increasing quite
rapidly.
t VORITE 125 CONTINUOUS PHASE

p~ a 2,815 POISE
DISPERSE PHASE y a 5.9-6.3 DYNES /CM 1
FKGURE 13 Effect of exceeding Eg o n reduced burst time.
Figure 14 shows that the effect of exceeding the critical shear gradient on fragment
size distribution is to broaden the size distribution. Shown are histograms both by
number and by volume of the fragment size distribution for breakup at the critical
value of reduced shear, at 2.5 times this critical value, and at approximately 4 times
the critical value.
Figure 15 shows, for a viscosity ratio of 0.1, the effect of exceeding the critical
value of reduced shear on drop fragment size in a somewhat different manner. Shown
236 H.P.GRACE
uc 502 poise: uD I uc 3.99 (101-~;
T - 2 7 dynes I cm
Dibutyl phthalate at EB ,r.1, Dibutyl phthalate at 247 E Dlbutyl phthalate at 3.85 Eg

Total drop fragments: 4 1 ! Total drop lragmcnts: 111 Total drop fragments: 774
.-
VI
c
1
------ 1 1 ,
a,
Drop Fragment Size. 5 of original drop diameter
FIGURE 14 Effect of exceeding Es on drop fragment size distribution.
here is the ratio of fragment to original drop diameter plotted versus ratio of actual to
critical value of reduced shear at burst. The lower curve is for the ratio of volume sur-
face average fragment size to original drop size, the upper curve is for ratio of maxi-
mum fragment diameter to the original drop size. In both cases the ratios decrease
rapidly at first as the critical value of reduced shear is exceeded, but tend to level out
to such an extent that it generally does not appear worthwhile to operate at reduced
shears exceeding the critical value by perhaps more than 4 to 10 times.
Figure 16 shows similar data for a much lower viscosity ratio of approximately
2 X lo-'. Again, it can be seen that the curves tend to level out beyond values of E/Es
in the range of 4 to 10, indicating that for minimum energy consumption per unit of
new surface, operation beyond this point is generally undesirable. This indicates that
to achieve a large drop size reduction ratio, shear should be applied stagewise, with
sufficient low shear holdup between stages to permit breakup of already deformed
drops, in order to minimize energy consumption and heat generation. Thus, for a
given system, the product of stage shear rate and shear time is indicated to be a con-
stant, with shear rate low and shear time long a t feed end, and shear rate high and
shear time short at discharge end of a preferred dispersion device.
Drops of viscosity ratio greater than 3.5 cannot be broken in simple rotational
shear, but rather require an irrotational shear for breakup. For this reason I would
now like to consider single drop deformation, extension, and breakup in irrotational
(i.e., extensional) shear fields.
Figure 17 shows schematically the four roll device which we have used to experi-
mentally examine these phenomena. The four rolls are driven in such a way as to
produce a hyperbolic shear field in which the velocities are defined as shown in Fig-
ure 17, where G' is the irrotational shear rate which is extensional along the x3 - x3
axis and compressional along the x2 - x2 axis. Near the observation point, 0, G' is
(Maximum)
L A 0.107 6 . 2 9 dyneslcm
,I
r v 1 . 7 8 ( 1 0 ~ ' ~5 . 9 5 (Average)
pc = 2815 poise
-VORITE 125 / Si 30,000 SYSTEM

-- II n
Si SO
II
FIGURE 15 Effect of exceeding Es on fragment-size-to-original-size ratio
equal t o a n apparatus constant times the roll speed. For the configuration of the four
roll device used, this was experimentally demonstrated to be true within f 2 cm of
the point zero by measurements on tracer particles.
Two separate variable speed drives were employed, one driving rolls one and two
and a second driving rolls three and four, with vernier control of the ratio such that a
drop could be held at the center of the hyperbolic shear field. Irrotational shear rate
from 0.001 reciprocal second to about 25 reciprocal seconds could again be covered.
The viscous fluid was floated on a low viscosity fluid in order to reduce end effects,
and air clutches were installed on the two drives such that the rolls could be started
and stopped nearly instantaneously.
The plot in Figure 18 shows data, expressed in the form of a reduced irrotational
shear required for continuous drop extension and breakup versus viscosity ratio. For
continuous phase viscosities of 45 t o 503 poise and for various interfacial tensions
from 1 to 6 dynes/cm, these data fall on a curve which shows a minimum at viscosity
238 H.P. GRACE
VORITE 125 / S i 5 0 0 SYSTEM

pC = 2,815 POISE p =1.79
FIGURE 16 Effect of exceeding EB on fragment-size-to-original-size ratio.
ratios between I and 5 , with a slow increase in required values of reduced shear for
breakup at high viscosity ratios out to values in the order of 900, and a similar slow
increase in required value at low viscosity ratios down t o values as small as lo-'. The
few data points taken by aso on' and by ~ a ~ l oinr a' similar four-roll device are also
shown. This curve shows that the value of reduced shear required for drop breakup at
the minimum of the curve is about 0.2, this value being only one-third of the value of
0.6 observed at the similar observed minimum for the case of rotational shear. This can
be seen in Figure 19 which shows the reduced shear required for drop breakup in ro-
tational shear and in irrotational shear, both versus viscosity ratio. The ratio of re-
duced shear required for breakup in irrotational shear as compared to that required
in rotational shear becomes even smaller as the viscosity ratio moves away from one
in either direction. Thus for equal shear rates, irrotational shear appears to produce
more effective breakup and dispersion than rotational shear, even a t low viscosity ra-
tios. Part but not all of this apparent advantage may be a mirage, in that the correla-
x - xlaxisl to plane of paper Et through origin0
1
Near Point 0 G' = (Apparatus Constant) (Roll Speed)

FIGURE 17 lrrotational shear field of four-roll device,
a
3
1
0 0 pc= i l p yd=6dyne/cm
P
m 0 p c =4 5 . 5 ~ yd=2-6dyne/cm
lo2- 0 p c = 502.5 p yd= 2.5-5 dyne/cm
$ 2 MASON 4-Roll Data A p C =45-50p y ~4.8-7.5dyne/cm
b
a,
=
-0 TAYLOR 4-RollDota v
p C =9 0 - i 0 0 p y = 8 - 2 3 dyne/cm
*
c
- a,
10-
o E V
6 is
u
c2 Viscosity Ratio, p,
FIGURE 18 Drop break-up in hyperbolic shear field.

240 H.P. GRACE
0 - loa, k
GBrd~J(plIr. 100
Rotational shear
Viscous Forces
(couettel
Interfacial Forces
at
Burst l rrotational shear
1 (4-roll1
0.1
-------
loe6 10-I lo1 lo2 lo3
Viscosity Ratio, p - vD I pc
FIGURE 19 Comparison ofeffect of viscosity ratio on critical shear in rotational and irrotational shear
fields.
tion for both rotational and irrotational shear incorporate the measured value of
simple rotational shear viscosity for the continuous and dispersed phase viscosities.
Trouton has shown that for three dimensional irrotational shear the fluid viscosity is
three times the simple shear viscosity. For the two dimensional shear field of our
four-roll device a similar analysis indicates that the irrotational viscosity is really
twice the measured simple shear value, i.e. at least for Newtonian fluid. While this
would not affect the viscosity ratio which is the abscissa of these correlations, it
would double the values of the ordinate for the irrotational case if the irrotational
viscosity were available and could be substituted, bringing the values of reduced
shear a t minimum of curves nearer to each other. Still unaccounted for, however,
would be at least a 50% difference between rotational and irrotational shear rate re-
-
Continuous Phase. Pale 4; . 45.5 poise
Dispersed Phase 0 Silicone Xll fluids; yo 4 - 6 dynes I cm
- 0 Polybutene a) & -
32 yo 2 - 3 dynes I cm
hbx. Stable
LO
A Paraplex G-dl ' yo -
1.0 dynes I cm
L-BIL.0,
dimensionless 0.6 -
Dispersed lo Continuous Phase Viscosity Ratio, pO 1 uC, dimensionless
FIGURE 20 Maximum stable drop deformation-irrotational shear

DISPERSION PHENOMENA 24 1
1. M)
Drop 0. 98
Deformation.
O.%
L-B I L+B
0. w
0.92
0 0 Deformed drops resume original
shape upon removal of shear
0.90
B 8 Deformed drops break up upon
0.88 removal of shear
0.86
0.01 0. 1 1.0 10 100 loo0
Viscosity Ratio. pD I pc
FIGURE 21 Drop deformation for burst at critical shear-irrotational shear.
quirements for drop breakup, with the latter having the advantage for dispersion
purposes. A recent rheological examination of non-Newtonian polymeric materials
by allm man' indicates that the irrotational shear viscosity is much less sensitive to
high shear rates than is simple shear viscosity. This observation suggests an addi-
tional advantage in the use of irrotational shear for dispersion purposes.
LB I Zrd
Drop Draw-Ratio
Necessary for
Break at EB
10
Icouette)
Viscosity Ratio, pD I vc, dimensionless
FIGURE 22 Comparison of effect of viscosity ratio on drop draw ratio for burst at critical shear-rota-
tional and irrotational shear.
242 H.P.GRACE
Figure 20 shows the maximum stable drop deformation under irrotational shear as
a function of viscosity ratio. These data are of interest because no drop breakup can
occur until both shear stress and shear time are such that the maximum stable drop de-
formation is exceeded. Exceeding this deformation, however, does not insure drop
breakup, but merely makes it possible if a proper combination of shear stress and
shear time is imposed beyond this limit. The curve in Figure 20 resembles the shape
of those in the previous figures, with a minimum near a viscosity ratio of one.
Figure 21 defines the critical drop deformation that must be reached before irrota-
tional shear is removed, if the drop is t o break after cessation of shear, rather than
contract to its original spherical shape. This figure shows two experimental curves. One
for the maximum drop deformation from which a drop recovers unbroken upon ces-
sation of critical shear. The other represents the minimum observed drop deforma-
tion resulting in drop breakup upon removal of the critical shear. These critical de-
formation values are large compared with the maximum stable drop deformation,
particularly a t high viscosity ratios, and pass through a minimum near a viscosity
ratio of 0.1.
Figure 22 shows a comparison of the critical drop draw ratio for breakup in both
rotational shear and irrotational shear. The minimum critical draw ratio in both
cases in approximately 5 and occurs at viscosity ratios in the neighborhood of 0.1.
Figure 23 shows a correlation for drop breakup in hyperbolic shear fields of critical
magnitude which may be used to estimate the critical shear time necessary to insure
drop breakup upon application and removal of the critical irrotational shear.
In Figure 24 this correlation of reduced critical burst time for irrotational shear is
compared with that previously determined for rotational shear. In the region of over-
lapping viscosity ratios, approximate agreement is indicated. Since the critical value
of reduced shear actually required is shown to be lower for irrotational shear, with
equal times under critical shear being required t o assure breakup, a lower total
energy expenditure for breakup under irrotational shear is indicated.
Reduced
Critical
Burst Time.
dimensionless
Reduced burst time vs. viscosity ratio
0 -
N, 45.5 poise: y - 4 - 6 dynes I cm
-
0 pc 45.5 poise; r m 3 -4dynes I cm
A Pr - 45.5 poise: y - 1.0 dynes I cm
Viscosity Ratio, I P , dimensionless

D c
FIGURE 23 Effect of viscosity ratio on reduced critical burst time for irrotational shear.
lam I-
lrrotational shear R R d
(4-roll1 \ 1
Reduced k 0
Burst Time
F
at Es.
$1 1 ad\
dimensionless
lo1
Vlscositj Ratio. )ID I vc, dimensionless
FIGURE 24 Comparison of effect of viscosity ratio on reduced burst time at En for rotational and irro-
tational shear.
Maintenance of the critical shear rate to the breakup point is generally not attrac-
tive from an energy consumption standpoint. Large deformations and long shear
times are required because of the damping action of.shear upon the growth of the in-
terfacial Rayleigh waves that eventually cause breakup. However, one additional
path exists for proceeding from the maximum stable deformation under shear to the
drop breakup point. This route involves maintaining the critical shear rate so as to
exceed the maximum stable deformation by an incremental amount, followed by a
Time from Start of Shear, sec
FIGURE 25 Four-roll run-shear removal and no break.

(Pale 4/Si 30.000. p d p , = 6.66, yo = 4.3 dynes/cm. E / E e = 1.07)
244 H.P.GRACE
programmed reduction in shear rate s o as to hold a constant drop deformation until
the developing Rayleigh waves grow to the point of drop breakup. In exploratory
studies of this alternate method, breakup has occurred at deformations less than the
critical deformation required for breakup with complete removal of shear. Total
breakup times for both methods, however, have been comparable.
Figure 25 shows the shear-deformation history deduced from a movie sequence in
the four-roll device in which the maximum stable drop deformation was exceeded,
but the critical drop deformation for breakup not reached at the time the critical
shear gradient is removed entirely. For this case the drop is able to contract, because
of interfacial tension forces, to its original shape before the Rayleigh waves can grow
to an amplitude such as to fragment the extended drop.
In contrast, Figure 26 shows similar shear-deformation history in which the maxi-
mum stable drop deformation is incrementally exceeded but the critical drop defor-
mation not reached and in which the shear instead of being abruptly reduced t o zero,
is decreased at a programmed rate such that the drop extension is held constant until
the growing Rayleigh waves result in fracture of the drop. It can be seen that the
shear required t o maintain this constant extension decreases rapidly to a very low
value as Rayleigh waves grow and pinch region develops. As yet we d o not have a
good measure of a comparative total energy consumption for breakage by this route
as compared to the route of exceeding the critical shear gradient followed by com-
plete removal of all shear.
In order to produce a two-phase liquid dispersion, the dispersed phase must be in-
troduced into the flowing continuous phase at some point, either upstream of the
FIGURE 26 Special four-roll run-program shear reduction for constant deformations lo burst.
continuous dispersion device o r in the continuous dispersion device itself. One conven-
ient means of doing this is to introduce the disperse phase as a jet stream in the pipe-
line feeding the continuous phase to the continuous dispersion device, and perhaps
by this means to take advantage of natural or artificially induced breakup of the jet
stream in order to assist dispersion in the continuous dispersion device. With this in
mind, we shall now discuss immiscible jet stream breakup upon injection into a lam-
inar flowing system of a continuous phase.
Existing theory for jet breakup originated with Lord ~ a ~ l e iwas ~ hextended
, ~ by
~ o m o t i k a , 'and has been generalized recently by Meister and ~cheele,'but relates
only to the ideal case of a jet entering a primary fluid in such a way that both are in
rectilinear flow. For this case the amplitude of the interfacial wave, t, is equal to an
exponential function of an initial perturbation at the nozzle, 0, where a is the growth
rate, r is the time from nozzle discharge, z is the distance downstream of nozzle dis-
charge, and k is the wave number.
c = to exp(at + ikz) (6)
= 60 exp(; + ikz) 2a
where k = -A
The size of the initial perturbation, 0, is generally unknown and can only be calcu-
lated from observations of the breakup distance Zg, and Rayleigh wave growth rate.
For the case of interest of a viscous liquid jet entering a very viscous liquid, the wave
number k has been shown to be inversely proportional to the jet diameter and a com-
plex function of the disperse to continuous viscosity ratio, which has been given graph-
ically by Meister and ~ c h e e l e . ~
The maximum wave growth rate, am,has been given as shown in Eq. (a), where 4 is
a complex function of viscosity ratio and is also given graphically by Meister and
Scheele. These relationships have all been derived for rectilinear flow and their appli-
cation t o the real case of an injection nozzle within a flowing pipeline with velocity
gradients between the jet stream and pipeline flow at the nozzle tip has not been
reported.
Figure 27 shows that with the jet on the pipe centerline, the ratio of equilibrium jet
diameter t o the nozzle diameter can be simply correlated with the ratio of nozzle to
pipeline flow velocities over a wide range of additive ratios and/or nozzle sizes, and
that a differential momentum balance supports these data as long as the ratio of noz-
zle to tube diameter is small and the viscosity ratio is also small. Thus, the equilib-
rium jet diameter and therefore the Rayleigh wavelength and major fragment size on
breakup become predictable using existing theory.
H.P. GRACE
-
- I i I
DJ/DNz JVN/2VToV WHERE DN/DT<cl 8

-
lor
-- pD/pC<< I
.. p C = l0 8 5 p p ~13p
=
- pD/pC=
0.012 y ~ 8 . dynes/cm
5
1.o :-
.
0.1- I # 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 , , #,#I
8
0.01 0.1 1.0 10.0
FIGURE 27 Observed injected jet diameter-centerline injection.
Figure 28 shows that for a laminar jet stream introduced through a nozzle o n a
centerline of the bulk phase transport pipeline, the wavelength of the Rayleigh waves
which develop and result in natural breakup can be satisfactorily predicted for a con-
siderable range of jet stream diameters less than the nozzle diameter using Tomotika
theory as generalized by Meister and Scheele, but that an increasingly important em-
pirical correction must be applied in predicting Rayleigh wavelengths as the equilib-
rium jet diameter exceeds the nozzle diameter. The curves in this figure also show
that the resulting size of the major breakup fragment remains a nearly constant 2 to
2.2 times the equilibrium jet diameter or a nearly constant 30-35% of the Rayleigh
wavelength, and can be satisfactorily related to the observed jet diameters and Ray-
leigh wavelength by the following expressions:
For forced jet breakup

--- ~EOICTEORAYLEIGH IlUMOTIKA THEORYI f D 3

7% OBSERVED RPHGE OF RALEIGH WAVELENGTH
8
0 AY OIAMOF MAJOR FRAGMENTS ON -am 8
010 STREAM BREPXUP
o ODD MZJ o m owo amo oow cnm ooeo mso om 010 om a~x,o m o~so 0 =
FINAL JET DIAMmER 0, .CM
FIGURE 28 Comparison of observed and predicted data-Raleigh wavelength and maximum fragment
diame~erfor centerline jet breakup.
These can be derived from a conservation of mass on the assumption that the mass of
the major fragment is equivalent to that of a one wavelength-long cylinder of the jet.
This appears so even when the observed and predicted wavelengths d o not agree as
with a large expansion at the nozzle. Natural breakup, under system conditions con-
sidered, resulted in mostly uniform size major fragments, between which were in-
terspersed a few (i.e. Eq. 1-3) satellite fragments of much smaller size. Under prac-
tical operating conditions with highly viscous systems, distances from nozzle to
natural jet stream breakup may become very large, and means of promoting jet breakup
at o r near the nozzle are of interest. For low viscosity liquid/gas systems, axial
vibration of nozzle has been widely reported to result in such premature jet stream
breakup into a nearly monodispersed fragment size if the nozzle vibration frequency,
f, is adjusted to equal the jet stream velocity divided by the dominant Rayleigh wave-
length for jet breakup. We have found that a similar result can be obtained even with
these very viscous liquid-liquid systems either by actual axial vibration of the nozzle
or by pulsing the flow of the nozzle feed stream near the theoretically predicted fre-
quency. An amplitude of the order of the equilibrium jet diameter for conditions con-
sidered, is necessary. For the usual operating conditions with such viscous systems
frequencies of 0.1 to 10 cycles/sec are required, as compared with frequencies in the
kilocycle range for nonviscous liquid jets in a gas, and the resulting fragment size is
necessarily much larger.
Moving the injection nozzle off the centerline of the pipeline and toward the wall
results in the jet stream being subjected to a net velocity gradient which increases in
magnitude as the nozzle approaches the pipeline wall. This results in a type of Ray-
leigh wave development and ultimate jet breakup which has not been described be-
fore, and which is both sensitive to the exact shear conditions and not readily predict-
able from existing theory. In the extreme of placing the nozzle at the wall, our
248 H.P. GRACE
movies show this phenomena to comprise the growth of Rayleigh waves on the high
velocity size of the sheared jet stream, with the crests of these non-symmetrical Ray-
leigh waves eventually forming white caps, and developing into separate finger
streams, which migrate toward the centerline while at the same time developing
further Rayleigh waves which result in their eventual breakup. Thus;the original jet
stream progressively divides itself into many smaller streams which d o the actual
breaking up.
Figure 29 shows that a correlation also exists between ratio ofjet to nozzle diame-
ter, and the ratio of nozzle to pipeline average flow velocities for conditions of addi-
tive introduction at pipeline wall, but of a different form from that resulting with cen-
terline operation. The relationship shown in this figure for injection at wall is a cube
root relationship in contrast to the square root relationship developed for the center-
line jet, and is entirely empirical.

Figure 30 shows that the Rayleigh wavelength of the growing perturbations for a
jet nozzle near the pipeline wall is significantly greater than predicted by theory and
in addition is very sensitive to the existing values of wall shear rate. Jets introduced
off-center, but not at the wall, were found to show the same effect, but with sensitiv-
ity to wall shear rate decreasing with increasing distance from the pipe wall. The dis-
tribution of breakup fragments under either off-center o r near wall conditions was
found to be very wide, but the size of the largest fragment remained 30-35% of the
FIGURE 29 Observed injected jet diameter-wall injection

i ' l ~ , ' l ' l . l ~ , ~ l ~ l
1.5 - -
1.4- -
1.3- WALL SR. 0125 scc-' -
0 WALL SR = 0.25 sec-'
z 1.2-
u sec-I
WALL SR ~ 0 . 5 -
; - 0.0457 crn diarn. nozzle 0 WALL S.R a 1.0 sac'' -
= 1.085 poise A WALL S.R a 20 su-' -
o 0.682 c m d i a m . t u b e
6 0.9- pg = 13poise -
J
-
-
-
-
-
-
0.k . o.& ' o.;o ' a;2 ' a;4 ' O.k ' a;e
.
'
EQUILIBRIUM JET DIAM. Dj CM
FIGURE 30 Comparison of observed and predicted data-Raleigh wavelength and maximum fragment
diameler-wall injection.
observed Rayleigh wavelength for the unbroken jet stream, and therefore was always
larger than would have been obtained for the same flow conditions with the jet intro-
duced at the center of the pipeline. The data of this figure and the previous figure
allow prediction of equilibrium jet diameter, Rayleigh wavelength and maximum
fragment size for a considerable range of conditions with injection nozzle at the
transport line wall.
Figure 31 shows that, for these same conditions, distances from nozzle to natural
breakup of the jet stream can be correlated with the product of pipeline velocity and
equilibrium jet diameter, although considerable scatter occurs. Such a correlation is
expected based on the growth rate equation for Rayleigh waves, on the assumption
that the initial perturbation is constant and very small, i.e. of molecular size, and on
the assumption that breakup occurs when the Rayleigh wave amplitude grows to
equal thejet stream diameter. The use of the theoretical growth rate equation and the
average data of this slide to calculate the size of the initial perturbation yields a value
of 2 X lo-' cm, which supports this view. Whereas Rayleigh wavelengths is largely
independent of system properties, breakup distance or time is sensitive to system
250 H.P.GRACE
l40r I I I I I I
I pc 1.085 poise
p~8 13 poise
0 NOZZLE CENTER OF' TUBE
NOZZLE AT WALL OF TUBE
FIGURE 3 1 Observed jet breakup distance correlation
properties and can be expected to be directly proportional to continuous phase viscosity

and inversely proportional to interfacial tension for other than the system a t hand.
Data in this figure for conditions of jet stream introduced a t the pipeline wall indi-
cates that the same general relationship exists, between distance from nozzle to jet
breakup point, and the product of jet diameter and pipeline velocity, as with the jet
stream introduced at the centerline. Since pipeline velocity a t the introduction point
very near the wall was less than 10% of the centerline velocity, this means that time
required for Rayleigh wave growth to the point of jet breakup was more than 10
times as great when a jet stream of the same equilibrium diameter was located near
the pipeline wall rather than on centerline. Thus, these data also show that a net ve-
locity gradient across the diameter of the injected jet &am inhibits the rate of
growth of Rayleigh waves in the injected stream of immiscible fluid, and prolongs the
time between injection nozzle and natural breakup point. If the pipeline velocity and
percent disperse phase are kept constant such that the ratio of average nozzle velocity
to average pipe velocity equals one for a nozzle near the pipeline wall, the equilibrium
jet diameter for a nozzle near the pipeline wall will be 40% greater than for the same
nozzle located at the center of the pipeline and under these practically comparable
conditions distance from nozzle t o natural breakup will be 40% less for additive in-
troduction at the centerline rather than the wall. Also Rayleigh wavelength, and
therefore size of the largest fragment, will be a minimum for additive introduction to
centerline.
Now we can get t o the heart of the matter, i.e. the application of the basic data al-
ready discussed to the interpretation and prediction of dispersion performance at
high viscosity in static mixers, for both low and high viscosity ratios.
A travelling microscope device was set up to observe and follow the dispersion oc-
curring in transparent static mixer sections under continuous flow and dispersed
phase addition conditions. Static mixers were oriented either vertically o r horizon-
tally, dispersed and continuous phases were provided in a fixed ratio by coupled posi-
tive displacement pumps, and dispersed phase was introduced at the center of the
pipeline upstream of the static mixer by a single nozzle arranged so injection flow to
it could be pulsed at the Rayleigh wave frequency if desired, in order to produce arti-
ficial breakup at the introduction nozzle tip as well as a continuous unbroken jet
stream.
Most measurements have been made in the Kenics static mixer and some in a 2 X 2
multiflex mixer.
For systems of low viscosity ratio the dispersed drop size produced by passage
through the static mixers was generally found inversely proportional to the nominal
wall shear rate for the static mixer tube.
Figure 32 shows a typical result for a system of 1000 poise continuous phase viscos-
ity, and 13 poise dispersed phase viscosity in terms of the volume surface average di-
ameter of fragment produced versus nominal wall shear rate. Curves are shown both
for conditions of the dispersed phase injection stream entering the mixer as an un-
broken jet, and for injected feed entering the mixer as the fragments produced by arti-
ficial breakup of the jet stream ahead of entrance to the mixer. As it can be seen,
breakup of the injected jet stream before it enters the mixer is highly desirable.
Figure 33 shows similar data for the maximum fragment diameter versus nominal
wall shear rate for the same mixer and system. Results are quite similar, the same ad-
vantage being apparent for case of a premature natural or artificially-induced breakup
of dispersed phase jet stream.
Figure 34 shows the effect on static mixer performance of changing system pa-
rameters. The dotted and broken lines show the performance for a system of viscosity
ratio of 3.2 X lo-' and a dynamic interfacial tension of 8.5 dynes/cm. Performance
of the same mixer is shown by the solid line for a much lower system viscosity of
4.6 X 10'~and the somewhat higher dynamic interfacial tension of 14 dynedcm. This
shift in fluid system parameter requires a 15-20 fold increase in the operating shear
rate of Kenics mixer unit in order to produce the same average fragment diameter
with either a n unbroken or a broken jet feed to the mixer. Referring back to our gen-
eral correlation for breakup in rotational shear, Figure 5, and keeping in mind the
change in viscosity ratio and interfacial tension between the systems in the plot in
Figure 34, we can determine that values of the dimensionless reduced shear quantity,
G~r,+~f(p)/y must increase from 1 to 10 because of the reduction in viscosity ratio.
With interfacial tensions simultaneously increased 1.7 fold, the value of the bursting
shear gradient, GB, would be predicted to increase 17 fold for equal breakup, since
the continuous phase viscosity and radius of the drop remained unchanged while the
H.P. GRACE
G- 41'/ POLYBUTENE BLEND SYSTEM
MIXER:$
. I.D.
2 0 ' ELEMENTS
3% INJECTION
a-UNBROKEN JET
CONTINUOUS PHASE - PARA G- 41
10 1
INJECTED PHASE. - I
.6 Pg 16 + .4 Pg 8
I d
01 0.1 1.0 - 1 100

WALL SHEAR RATE scc
(ENTERING MIXER)
FIGURE 32 Kenics dispersion performance-average drop size vs. wall shear
theoretical function of viscosity ratio varies little. This prediction is in reasonable

agreement with the observed results shown in Figure 34.
Figure 35 shows comparative performance data for the same change in system
conditions in the mixer as before, but presented in terms of the maximum fragment
rather than the average fragment diameter. The 19-28 fold increase in shear re-
quired to produce the same drop fragment size, while not in as good agreement as
when considering average drop size, is in reasonable agreement with the predicted
value of 17 fold.
Figure 36 shows a plot of the calculated effective shear rate in a Kenics mixer vs.
the nominal wall shear rate, and indicates the effective shear rate is some six times the
nominal wall shear for conditions of rotational shear. Effective shear was arrived at
G-41/ POLYBUTENE BLEND SYSTEM
MIXER:^" I D
2 0 ELEMENTS
3% INJECTION
UNBROKEN JET FEED

DROP FEED FROM NATURALLY BROKEN JET \
CONTINUOUS PHASE
INJECTED PHASE --
PARA 6 - 4 1
- 6 P g 8 + .4 Pg 16
lo1 I I I d
.01 0.1 1 .o 100
WALL SHEAR R A T E rec"
(ENTERING M I X E R )
FIGURE 33 Kenics dispersion performance-maximum drop size vs. wall shear.
by using our basic correlation of rotational shear required for breakup, with substitu-
tion of experimentally observed values of average drop fragment size so as to back
calculate what the effective shear must have been in order to produce these observed
drop sizes.
O u r basic work on single drop breakup has shown that staging of shear intensity
and the inclusion of low shear relaxation zones between successive stages of increas-
ing shear intensity could be expected to increase the energy utilization of a dispersion
device for immiscible fluids, thereby reducing heat generation. The data shown in
Figure 37 support this conclusion by showing that improved performance (i.e. re-
duced dispersed drop size) is possible by staging the shear intensities, and can be
further increased by including a low shear relaxation zone between shear stages, i.e.
254 H.P. GRACE
V)
zz
8
U
P CONT. PHASE 6-41, & =11085 poise
3 %-INJECTION DISPERSE PHASE
d
"3
I- REF. I OATA: DISF! PHASE -2
S a Y B U T E N E BLEND pd& = 1.2[IO)
5
4
\\ \ y D = 8 5 YE=3.8 dyneslcm
a
I&
-- UNBROKEN JET FEED
v
----- DROP FEED FROM NATURALLY
10 r BROKEN JET
W
v
9
'\ \ NEW DATA: OlSP PHASE 50c3 DC-7.W S l t K O N ~
~ p C = 4 . 6 ( 1 ~ l ' y0=14.4
4 yc=60dyneslcm
a
rn
P:
t . UNBROKEN JET FEED
2
0
1Oqlm1n
MRUFUT
o DROP FEED FROM NATURALLY
BROKEN JET
>
0.1 . I I
a1 1.0 10
NOMINAL WALL SHEAR RATE OF 114-IN. MIXER. BVID, sec-'
FIGURE 34 Effect o f changing system parameters on Kenics dispersion performance-prediction.
"3
-
CONT. PHASE 6-41 pc 2 l.OB5poise
PU 3 I.INJECTION DISPERSE P W E
a
3:
5
-10 -
o DATA REPORT FA 13188 DISC PHASE 2
I-
t POLYBUTENE BLEND rdrc=12[lei
W
3 \
\
yo =a5 yE ~ 3 . 8dyntslcm
0
4 \\ \ --UNBROKEN JET FEED
=
L
\\ \ ---- DROP N O FROM NATURALLY
102 \\, \ BROKEN JET
NEW DATA DISC PHASE 5 0 0 CC-200SILICONE
z \ \\ g y .4.6[R)) yo 114.4 y ~ = h O d y n n l c m
X
u
a
'\
w UNBROKEN JET FEE0
t .
10q l m ~ n o DROP FEED FROM NATURALLY
THRUWT BROKEN JET
10 0
QOI 01 1.0 10
NOMINAL W A U SHEAR RATE Cf V4-IN. MIXER, 8 V I D.
FIGURE 35 EFTect of changing system parameters on Kenics dispersion performance-prediction.

3% injection
7-21s
Wall Shear Rate - set.-'

FIGURE 36 Kenics mixer effective shear for dispersion
-
pC= 1.085 poise; h=
13 poise; ydynamic
~ 8 . dyneslcm
5
Unbroken jet feed of 3% disperse phase
0 1/4: -20 stage mixer
8 %,, -20 8 %"-21 stage mixers in series
% -20. y4"-21, 8 ye'"l7 mixers in series
Y4"-21. 12.75 cc relaxation zone a
5$" -17 mixer in series
I I I
1 .o 10 100
Total Hydraulic Power Expended
in Mixer Assembly,
(ap) (Q), in. Ib/min
FIGURE 37 Kenics mixer effect of staged shear
256 H.P. G R A C E
between the 1/4 and 1/8 in. mixers for this case. The preferred frequency and holdup
volumes of such relaxation zones can be estimated for individual situations by use of
the general drop breakup correlations presented previously.
Figure 38 shows that even with a relaxation zone between stages, the proposed me-
thod of scaling performance from one fluid system to another is valid for viscosity ra-
tios less than unity. Thus, the same shift in fluid system parameters as shown pre-
viously for a single Kenics mixer, again requires a 16-19 fold increase in shear with
staged mixers in order to produce a dispersion of equivalent drop size. This observa-
tion compares well with the 17 fold increase predicted from basic breakup
correlations.
Figure 39 compares the dispersion performance of various combinations of stand-
ard Kenics static mixers in terms of volume medium diameter and maximum drop
size for a viscosity ratio of 2 X lo-', the interfacial tension of 6 dynes/cm, 5.9% injec-
tion of dispersed phase, with a total throughput of 15 Ib/hr. Produced drop sizes de-
crease nearly logarithmically with increases in mixer pressure drop o r hydraulic
power expended, to drop sizes equal to o r smaller than those produced in rotational
mixers under similar conditions. For equal drop sizes, necessary hydraulic power ex-
penditure in the static mixers appears to be an order of magnitude less than the rota-
tional power requirement in driven dispersion devices. While the curves in this plot
show a generally logarithmic relationship, significant scatter of data points exists.
Those representing staged combinations of static mixers separated by a low shear re-
laxation zone between stages show a minimum ratio between the maximum and vo-
lume median drop diameter, i.e. a narrower drop size distribution.
WNT. PHASE G-41 p c = 1,085pose

3% INJECTION DISPERSE PHASE
REF. I DATA: DISF! PHASE -2

WLYBUTENE BLEND 14\/14*1.2(10)
MAX FRAGMENT SIZE-UNBROKEN

- - . FEED
JET . ---
----- VOC;SURF: AVG. FRAGMENT SIZE -
UNBROKEN JET FEED
NEW DATA: DISF! MAS 50a DC- 2 0 0 SILICWE
p&:4.6(10)-' YD 10.4.4 YE= 6.0dyeslcm
a MAX.FREGMENT SIZE -UNBROKEN
JET FEED
I o WL:SURF. AVG. FRAG. SIZE -
10 q/ rnin
THfURlT UNBROKEN JET FEED
1 .o I I
1.0 I0 102
NOMINAL WALL SHEAR RATE \TE FlML1/8-IN. MIXER. BVID, scc-'
FIGURE 38 Effect o f changing system parameters on Kenics dispersion performance with staged
mixers-prediction.
ill Single U2-21
(a Two U P 2 1 i n series
U) One 1/2-21 + 102 cc holdup + One 1/2-21
(4) One U4-21
(5) Three 1/2-21 in series
100 7 (6) One U2-21 + One 1/4-21
QJ One U P 2 1 + 96 cc holdup + One 1/4-21
(8) Two 1/2-21 + One 1/4-21
(9) Two U2-21 + 22cc holdup +One 1\4-21
o(4) (10) One U2-21 + One U8-21
Volumemedian diam.
I (Cou lter counter)
1 I I I I I
10- 10-
raulic Power Expended i n Mixer Assembly, Hp
lo2 lo3
Total Mixer Pressure Drop, Ib 1 sq in.
FIGURE 39 Drop diameter vs. mixer Ap or power for various Kenics mixer combinations.
Figure 40 shows this effect more directly as drop size distribution curves for the
dispersion produced, respectively, by two 1/2 in. 21-stage mixers directly connected
in series, and by the same two mixers connected in series by a low shear relaxation
zone.
While for low viscosity ratio systems of high viscosity, the dispersion drop size
produced in static mixers is generally inversely proportional to the nominal shear
wall rate of the mixer section (or to throughput), with increasing throughput a point
is reached at which this is no longer true and the curve of drop size versus throughput
passes through a minimum, with the produced dispersion drop size actually begin-
ning to increase at higher throughputs.
Figure 41 shows data taken with a 1/2-in. 21-stage Kenics mixer in the region of
this critical throughput for a continuous phase viscosity of 600-800 poise, viscosity
ratio of 2 X and interfacial tension of 6 dynes/cm. Dispersion performance is
shown here in terms of the 50% median diameter, the 10% oversize diameter, and the
H.P.GRACE
%byVolume Larger than Size
99.99 99.9 99 9080 6 0 4 0 a 1 0 1 . . 0.1 0.01
100 I t 1 1 1 1 1 I I
One 1/2-21 + One Y4-21 Std. Mixer

Dispersion Drop
- 0
o One U2-21 + 96 cc holdup.

Diam., D
P'
- + One Y4-21 Std. Mixer
microns -
10 -
rD - 6 dynes 1 cm
-
- pc = 600 poise
- 5% injection at
15 Ib I hr throughput
% by Volume Smaller than Size
F I G U R E 40 Effect on drop size distribution of intermediate low-shear holdup between stages in Kenics
mixer.
maximum drop diameter microscopically observed. In all cases a minimum drop size
occurs in the region of 15-20 Ib/hr throughput.
Figure 42 shows the same drop size data expressed as drop size distributions result-
ing at the various throughput rates, illustrating the shift in drop size distribution
which occurs near this critical point.
Figure 43 shows similar data for 1/4-in. 21-stage Kenics mixer, which indicates
that the critical throughput rate may be on the order of 9-10 Ib/hr for this particu-
lar system, with drop size increasing beyond this throughput.
Again Figure 44 shows the shift in drop size distribution with increasing through-
put for this 1/4-in. mixer.
Figure 45 shows that by staging two 21-stage, 1/2-in. Kenics mixers followed by
two more 21-stage, 1/2-in. Kenics mixers, the critical throughput at which this min-
imum in drop size occurs can be extended to higher throughputs. In fact, the min-
imum has not yet been reached in this case a t throughputs on the order of 30 Ib/hr,
although the curves have flattened indicating a rather broad minimum.
Figure 46 shows the drop size distributions over the range of throughputs indicated
in the previous figure for the particular system employed and the two 1/2-in., 21-
stage mixers followed by a second pair of 1/2-in., 21-stage mixers with low shear
holdup section between.
Analysis of these dispersion performance versus throughput data. in terms of feed
drop size, effective mixer shear rate and required shear time t o break the feed drop
size as determined from our basic correlation of reduced shear rate and time required
MIXER TUBE WALL SHEAR RATE ~ v / D , s ~ c - '

5 10 15 20 25 30
-
I-
I I 1 I I
-
-
-
-
-
OVERSIZE DIAM.
5 0 VOL O/o MEDIAN

--
-
-
-
- S I N G L E 112- 21 K E N l C S
A V ~ . ~ =, 840 P ,+,I~= 1.2 ( l o r 3

-
= 6 dyne/cm 5.9% DISP PHASE
1.0 I
I I I I I I I I I I
4 8 12 16 20 24 28
MIXER THROUGHPUT, I b / h r
FIGURE 41 Single 1/2-21 Kenics-dispersion performance vs. Ihroughput.
260 H.P. GRACE
% BY VOLUME LARGER THAN SIZE
IOC 99.9
I
99
I
95 80 60 40 20 10 5 2 1 0.1
Av. pc * 8 4 0 p
7, 6 dynes/cm. 5.9% DISP PHASE
In
Z
0
fK A 11.0 Ih/hr. THRUPUT
-
0
a o 14.6
3
a
20.7 "
o 25.4
g IC
a
0
Z
0
In
a
W
P
E!
n
I I 1 I I I I I I 1 . I I
1.C
0.1 1 2 5 10 2 0 4 0 60 80 95 99 99.9
% BY VOLUME SMALLER THAN SlZE
FIGURE 4 2 Single 1/2-21 Kenics--drop size distribution vs. throughput.
for breakup of single drops, indicates the minimum in the dispersion drop size versus
throuput curves, and the continuous increase in drop sizes beyond some limiting
throughput, can be accounted for by inadequate shear time to extend and break the
effective feed drop size when a limiting throughput is exceeded. Thus, a minimum
dispersion drop size occurs as the ratio of maximum shear time required to average
residence time approaches 0.5, with dispersion drop size increasing continuously as
this ratio exceeds 0.5.
Figure 47 shows this for the 1/2-in. 21-stage Kenics mixer, in terms of a plot of
drop size versus ratio of required time to break the feed size t o the average residence
time. Well defined minimums are seen in the curves: the 50% median diameter, 10%
oversize diameter, and the microscopically observed maximum drop diameter, as
ratio of required breakup time to average residence time approaches 0.5.
Figure 48 shows similar data for the 1/4-in., 21-stage mixer, and indicates that in
this case all operating data points shown are for values of this ratio greater than 0.5.
Figure 49 shows the effect of dual point versus single point disperse phase feed (i.e.
a smaller disperse phase feed size) o n the dispersion performance of a 1/2-in., 21-
stage static mixer. Dashed curves are foi drop size versus throughput for single point
disperse phase feed already discussed, while solid curves result from a 30% reduction
in disperse phase feed size obtained by doubling of injection points. Produced drop
MIXER TUBE WALL SHEAR RATE, 8v/o,sec:'
obs. MAXIMUM DIA.
V)
Z
10% VOL. OVERSIZE DIA
0 -
5
W
-
N 50% VOL. MEDIA'N DIA.
V)
0
LT
-
-
-
-
- Av. pc = 760 POISE, )LC= l . 3 ( 1 0 ) - ~
Yo = 6 Dy ne/cm., 5.9% DISF! PHASE
1.01 I I I I I I
4 8 12 16 24 20 28 1
MIXER THRUPUT, Ib./hr.
FIGURE 43 Single 1/4--21 Kenics-dispersion performance vs. throughput
H.P. GRACE
Yo BY VOLUME LARGER'THAN SlZE ,
99.9 99 95 80 60 40 20 10 5 2 1 0.1
I I I I I I I I
yo = 6dyne/cm 5.9% DISP: PHASE
o 11.0 Ib./hr. THRUPUT

o 14.6
* 20.7 "
.
e ..
I 1 1 I 1 I I I 8 I I ,
0.1 1 2 5 10 20 40 6 0 80 95 99 99.9
FIGURE 44 Single 1/4--21 Kenics-drop size distribution vs. throughput.
size is decreased, while critical throughput rate above which performance degrades is
extended to higher values. This is directly related to the reduced time under shear re-
quired for breakup with a smaller feed size.
Model system dispersion runs employing variable dispersed phase fractions over
the 0 to 20% range, both with comparable combinations of static mixers and a rota-
tional mixer, have shown that at the average shear rates existing in the mixers, the ob-
served linear decrease in mixer torque and/or pressure drop with increasing dispersed
phase injection can be approximated by an Einstein-type relationship for dispersion
viscosity, in which the constant has an observed value of 2-1/2 to 3 compared to a
theoretical value of 2-1/2.
Figure 50 shows that under these variable injection conditions, both the 50% vol-
ume medium diameter and the 10% oversize volume diameter of the produced dis-
persion drops increase nearly linearly with increase in fraction disperse phase. These
increases in dispersion drop size are a direct result of both the linearly decreasing dis-
persion viscosity and of the increase in effective feed drop size caused by increasing
injection ratio at a single injection point.
Figure 51 shows that increasing dispersed phase fraction generally shifts the entire
drop size distribution curve to a larger size without greatly affecting the slope of the
drop size distribution curves for dispersions produced in either static o r a comparable
rotational mixer.
NOMINAL MIXER TUBE WALL SHEAR RATE, 8 VID, sec-1
AVG. MAX I MUM DIAM. -

-
-
.-l i
10VOL O/O OVERSIZE DIAM. Y
- \
u
ul
5 0 VOL % MEDIAN DIAM. --

-
T W O 1 1 2 - 2 1 K E N I C S + 103 CC HU
+ TWO 112- 2 1 K E N I C S
-
= 6dynelcm 5.9O/0 DISF! PHASE

I I I I 1 I 1 I I I I I I lo2
8 12 16 20 24 28 32
MIXER THROUGHPUT, Ib/hr
FIGURE 45 Two 1/2-21 + LS holdup + two 1/2 + 21 Kenics-dispersion performance.
H.P. GRACE
999 9
:
% BY VOLUME LARGER THAN SIZE
951 8: 6: 40 2: I? )
/
, 0;
I I I I I I I I I I I I
0.1 1 2 5 10 2 0 4 0 60 80 95 99 99.9
% BY VOLUME SMALLER THAN SIZE
FIGURE 46 T w o 1/2-21 + LS holdup + two 1/2-21 Kenics-drop size distribution
So much for the performance of static mixers on high viscosity systems of low vis-
cosity ratio. We shall discuss now the performance of static mixers used for high vis-
cosity systems of high viscosity ratios when the irrotational shear becomes a require-
ment for drop extension and breakup.
Figure 52 shows that for a high viscosity ratio requiring irrotational shear for drop
breakup and dispersion, the volume average surface diameter of dispersed drops
produced in a Kenics static mixer again decreases with increasing superficial wall
shear rate, but with a slope of -1/2 rather than -1. Also, whether the injected
dispersed phase jet stream introduced upstream on the pipeline centerline remains
unbroken o r breaks up by natural o r artificially promoted growth of Rayleigh waves
before entering the mixer appears to have little effect on the dispersion performance
expressed in terms of average drop diameter.
Figure 53 shows the same general inverse half power relationship between the max-
imum fragment diameter and the superficial wall shear rate. However, in these terms,
whether the dispersed phase jet stream is broken or unbroken a t the entrance t o the
mixer seems to have some slight effect.
Figure 54 shows that the volume average drop size produced for a given superficial
wall shear rate is changed little by either extending the mixer length beyond standard,
o r by direct combination of mixers to yield staged shear. However, the combinations
of mixers to yield shear staging with a n intermediate low shear holdup zone, of suffi-
IOC
/
10VOL %OVERSIZE DIAM.
E
a
I'
5
0
50 VOL ' l o MEDIAN
a 10 DROP DIAM.
0
a
0
z
-
0
m
a
W
a
m
-
n /
/ '
/
112- 21 KENlCS M I X E R
/'
A V ~ = .8 4~0 ~p pdpC
= I. 2 ( lor3
~=6dyne/cm 5.9% D l S P P H A S E
1. o 1 1 , 1 , , , , , 1 , , 1
0.1 0.2 0.3 0.4 0.5 0.6 I
R A T I O t / l ~FOR MIXER CONDITIONS

FIGURE 47 1/2-21 Kenics dispersion drop diameter vs. ratio of reduced burst t i m e h i x e r average res-
idence time.
cient duration for completion of breakup of the deformed drops between shear
stages, results in a significant reduction in fragment size of the produced dispersion.
Figure 55 shows this same effect in terms of the maximum fragment diameters in
the produced dispersion, and indicates again that the maximum effect occurs upon
staging mixers with a holdup zone between the staged mixers.
The basic correlation of critical shear required for drop breakup at high viscosity
ratios in irrotational shear shown in Figure 18 has been used, combined with the ob-
served average drop sizes produced a t high viscosity ratios in the Kenics static mixer,
to calculate the effective irrotational shear within the Kenics mixer as a function of
superficial wall shear rate, or mixer throughput.
266 H.P. GRACE
100.
'. 7-
u'
w
V)
MICROSCOPICALLY Obs.
A v . O F M A X I M U M DIA. w
ti-
-
X
H
V) 10% OVERSIZE
-
z
Z Z
V O L . DIA.
0
a
0
V)
-
U a
W
5
- n
V)
3 5O0/o VOL. 0
10-- MEDIUM DIA.
n
0
a
P
Z
-- \
-
\
----- -
-
-
0
LL
W
+
a
2 - a
- pc = 760 POISE
V)
a p,,/pc = 1.3 ( 10)-3 a
w a
0 W
-cl
V)
- 7, = 6 dyne /cm. I
V)
W
>
-
- +
0 U
/ W
/ LL
/ LL
/ W
/
1.o I I I I I I -102
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
RATIO t/tR FOR M I X E R CONDITIONS
FIGURE 48 1/4--21 Kenics dispersion drop diameter vs. ratio reduce burst time/mixer average resi-
dence time.
A V
~ .
. 6 ~0 0 p 1 . 2 1 0 ' y d . 6 dYn*/cm
5.9% DISPERSE PHASE INJECTED THROUGH:
--- SINGLE C 2 0 HYPODERMIC NEEDLE lPREVlOUS 04TA I
ON MIXER CENTERLINE
-DUAL C 2 2 UYPODERMIC NEEDLES i 1/32'
EACH SIDE OF MIXER CENTERLINE
I I I I I I I I I I I I I I 1 lo
6 8 10 12 14 16 I8 20 2 2 24 26 28 3 0 32
M I X E R THROUGHPUT. I b l hr
FIGURE 49 Effect of dual vs. single point injection on dispersion performance single 1/2-21 Kenics.
Figure 56 shows the resulting correlation which is a power function with a slope of
1/2.
Figure 57 shows this effective irrotational shear for the Kenics mixer, compared
with the previously determined effective rotational shear in the Kenics mixer.
Thus, the effective irrotational shear rate is pioportional to the square root of the
superficial wall shear rate whereas the effective rotational shear rate for systems of
low viscosity ratio is directly proportional to the superficial wall shear rate. In terms
of relative magnitude the effective irrotational shear rate is only 7.0 to 0.7 percent of
the effective rotational shear rate over the practical operating range of wall shear rates
from 1 to 100 reciprocal seconds. This quantifies the irrotational shear component in
the Kenics mixer as being small compared to the rotational component at realistic
wall shear rates. However, this irrotational component is sufficient to enable the
mixer to be used for dispersion at viscosity ratios above 3.5 if dispersion fineness is
not too critical, or if relatively high pressure drop operation can be tolerated. Physi-
268 H.P. GRACE
I I
Ydu6 DYNE/cm. 15.5 IWhr. THRUPUT V12

IMB ~ K E N I C S + H O +KENIC
~
MIXER ( 4 2 STAGES TOTAL 1
50% VOL. DIA. 0
IO%VOLDIA. A A
VOLUME % ADDITIVE IN JECTlON

FIGURE 50 Effect of additive injection ratio on dispersion drop size produced by IBM mixer and by
1/2-21 + +
H U 1/2-21 Kenics.
% BY VOLUME LARGER THAN SlZE
99.9 99 95 80 60 40 20 10 5 2 1 0.1
I I I I I I I I I I / I
%ADDITIVE INJECTION
0 5.9%
PC' 9 0 0 ~ Po/pc = I
= 6 d y n e /cm.
MIXER THRUPUT 15.5
I I , , I I I I I I t
0.1 1 2 5 10 20 4 0 60 80 95 99 99.9
FIGURE 51 Effect of additive injection ratio on drop size distribution produced by 1/2-
21 +
H U f 1/2-21 Kenics mixer.
3% injection. ~0 I $-
6.66. 7- -
4.9 dynes 1 cm
8 Unbroken jet stream feed
0 Broken jet stream feed
U4-215 Kenics mixer
* Disperse-to-continuous phase viscosity ratio

Oo Interfacial tension
Superficial Wall Shear Rate. 8V I D, sec
FIGURE 52 I/4-21s Kenics-av. drop diameter produced for high vsicosity ratio system.
270 H.P. GRACE
-3% injection, p~ I 6.66. y .

4.9 dynes Icm
8 Unbroken jet stream feed
0 Broken jet stream feed
U4-215 Kenics mixer
0.01 1 I I
I
0. 1 1.0 10
Superficial Wall Shear Rate. 8V I D, sec"
FIGURE 53 1/4--21s K e n i c s m a x . drop diameter produced for high viscosity ratio system.
-
3% injection. pD Ipc 6.66. y = 4.9 dynes I cm
-- Y4-21s mixer
0 Y4-40 "
0 Y4-20- Y8"-17 mixer comb.
Q Y4-40- Holdup zone- Y8"-17
mixer comb.
1.0 10
Superficial Wall Shear Rate. 8V I0, sec
FIGURE 54 Effect of Kenics mixer staging on av. drop diameter produced at high viscosity ratios.
3% injection, rD I kc - 6.66. r - 4.9 dynes I cm
---
0 114-20- U8-17 Kenics mixer comb.
E
u
-
v -
0 U4-40 Kenics mixer
1\4-40 Holdup zone - 1/8-11Kenics comb.
.-
0
Superficial Wall Shear Rate. 8V I D, sec-I
FIGURE 55 Effect of Kenics mixer staging on max. drop diameter produced at high viscosity ratios.
-I
- 3% injection, pD I pC = 6.66, y - 4.9 dynes I cm
h: 8 Unbroken jet stream feed
- -
V)
d o Broken jet stream feed

m
PI 114-215 Kenics mixer
L
m
a
3
c
-
v, 1.0-
m
C
.
.-
0
@
m
d
-
-El
L
-
--
a
.->
U
w
5
0.1 1 I 1
0.1 1.0 10
Superficial Wall Shear Rate, 8V I D, sec-'
FIGURE 56 Effective irrotation shear of Kenics mixer vs. nominal pipeline wall shear.
272 H.P.GRACE
Calculated
Effective Shear -
Rate for Drop cosity ratios < 1.0
Dispersion. (Principally rotational shear component)
-1
set
Slope = 1.0
/
/ For viscosity ratios > 4.0

(Component of irrotational shear only)
- / Slope = 0.5
/
0. 1 - I I J
0. 1 1 10 100
8V
Superficial Wall Shear Rate, T . sr-'
FIGURE 57 Comparison of effective shear of Kenics mixer for drop dispersion at low and at high vis-
cosity ratios.
cal origin of the irrotational shear component within the static mixer channels seems
to be related to the interlaminar transfer occurring because of the radial circulation
within the twisting semi-circular sections.
Figure 58 shows average drop size produced versus mixer wall shear rate data
taken with several 1/4-in. and 1/2-in. Kenics-type static mixers, for a system with a
dispersed to continuous phase viscosity ratio of 110, as compared to a curve of pre-
vious data, shown dashed, taken for a Kenics mixer at a viscosity ratio of 6.7. Over
this range of viscosity ratios our basic correlation of shear required to break drops in
irrotational shear fields indicates very little change in dispersion performance would
be expected for a constant continuous phase viscosity and interfacial tension. This
agrees with the observation shown here for the 1/2-in., 21-stage and 1/4 40-stage
mixers. Better than predicted performance with the 1/4 2-stage mixer may reflect
experimental error and/or some unaccounted for internal design tolerance in specific
mixers used.
Comparison of average dispersion drop size versus hydraulic power expended in
static mixers shown in Figure 59 for both the Kenics-type design and the 2 X 2 multi-
flex design show that at high viscosity ratio, the Kenics design excels in a manner sim-
ilar to that found previously for systems of low viscosity ratio.
Figure 60 presents data for viscosity ratios from 8 to 110 and continuous phase vis-
cosities of45 to 625 poise which show that for constant interfacial tension, both aver-
age and maximum dispersion drop size can be correlated with mixer pressure drop
alone for a variety of static mixer sizes or combinations, by a simple power function
E
a
U1
>
9
1
9
a
a
0
a
0
\
3
LU
lo=.--
-- p, = 45.5 POISE
\
a 3 . 0 % DISPERSE PHASE
- -- PREVIOUSLY
3
V)
\ - REPORTED DATA (2) pD/pC = 6.7
i
0
- 0 I - 21 MIXER yd = 4.9 dyne/cm
> -
t
2
!V) - a --1 4 0 MIXER pD/pC=I10 y d = 4.7 dyne/cm
a I
w A F-21 MIXER
a
: 'O0.1
, , , , ,
1.o
, , , , , , ,,,
10
, , , , , ,
MlXER TUBE WALL SHEAR RATE, WID, sec-I

FIGURE 58 Effect o f changing viscosity-ratio on high viscosity-ratio dispersion performance-Kenics
mixers.
a
n -
9
>
-
1
9
-
a
a
0
a
n
Y
a
a
"
3 7, = 4.7 dyne/cm 3.0% DISPERSE PHASE
0 $ -21 KENICS MlXER
2 A + -21 KENICS MlXER

-
0
w 0'-26 MULTIFLUX MIXER

16
a
10' I 1 1 1 1 1 I 1 1 I I 1 1 1 1 1 1 1 I I , I I I I I
10" 10-6 10-5

MlXER HYDRAULIC POWER ( Ap) (a). Hp
FIGURE 59 Comparison of dispersion diameter produced vs. mixer hydraulic power for Kenics and
multiflux mixers at high viscosity-ratios.
274 H.P. GRACE
1 A '(2-21 MIXER
V Ik2-21+ 1/2-21 MIXER 0 It2-21 t 1/4-21
MIXER A p , psig
FIGURE 60 Kenics mixers at high viscosity-ratio--av. drop diameter vs. Ap.
I
B
E
0
i 103:
9
0
a.
0
2 lo2, V TWO &-21 MIXERS

H
Z o h-21 + k 2 1 MIXERS
0 o TWO 4 - 2 1 + .$-21
V)
a
W
P I I I I I
,
v, 10
10" 100 10 102 10
MIXER Ap, psig
FIGURE 61 Kenics mixers at high viscosity-ratio-max. drop diameter vs. Ap.

with an exponent in the range of -0.49 to -0.66 o r very close to 0.5. This figure shows
data in terms of dispersion volume surface average diameter.
Figure 61 shows similar data in terms of dispersion maximum drop diameter. In
both cases the data span a 10-'-fold range of mixer pressure drop and are qualified
only in that mixer residence time must exceed the time required for feed drop size de-
formation and breakup to occur, as predicted from the previously presented basic
correlations for breakup under irrotational shear. Although not included as a varia-
ble here, absolute level of dispersion drop size produced can be expected to be in-
versely proportional to interfacial tension as indicated by the basic correlation for
drop breakup in either irrotational o r rotational shear fields.
Figure 62 shows, for a constant 15 Ib/hr throughput, the dispersion performance
of single and multiple static mixers in terms of hydraulic power expended for a high
viscosity ratio system. Again, this is shown in terms of volume medium and 10%
oversize drop diameters. These decrease in proportion to 0.5 power of horsepower
-
E
p c = 62spoiso p D l r C8
IS lb/hr THROUGHPUT
Y d m 4 . 7 dyno/cm
5 . 9 % DISP. PHASE
3
a
i VOL 10%
-
MED. OVERSIZE
10 VOL% OVERSIZE
P DROP DIAM: DROP DROP
0 Dl AM. Dl AM.
z SINGLE V2-21 KENICS 0 0

--
a
1/2-21 KENICS t IOeccHU
+ 1/2-21 KENICS
+
1/2-21 KENICS 108ccHU
0 rn --
W
P
V1
+TWO 1/2-21 KENlCS
+
1/2-21 KENICS lOBcc HU
A A
-
0
+ 112- 21 KENICS
+ 1 / 4 - 2 1 KENICS V V --
DROP DIAM.
1.0
lo4 10-2 lo-'
TOTAL HYDRAULIC H P EXPENDED I N MIXER
I I
100 500 1.000
MIXER A p , psi9
FIGURE 62 Staged Kenics mixers at high viscosity-ratio-drop diameter vs. mixer Ap + total hydrau-
lic power.
276 H.P. GRACE
lo3
--- 1 ' ' ""'1 I 1 1 1 1 1 1 4 1 I 1 I
--
I I I I -
-- --
- -
- pC = 625 poise pD/ p C = 8 % 4.7 dyne /cm
-
- 15.5 I b / h THROUGHPUT 5.9% DISPERSE PHASE -
E
lo2= ---
f --- -
--
9
a -
0 - -
0
a
a - -
z
-0In - 0 VOL % OVERSIZE DROP DIAM. -
a
W
0 10
--= VOL- MEDIAN DIAM. ---
9
a
-- --
-
- -
- -
- -
1 .o I 1 1 1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 I I
10-3 10-2 lo-' 1.0
MIXER ROTATIONAL POWER, hp
3 7 I8 36 72
M I X E R ROTATIONAL SPEED, r p m
FIGURE 63 Drop size vs. IMB mixer speed and power-high viscosity-ratio system
expended, with roughly 10-2 hp being expended t o achieve a volume mean diameter
of 4.0 microns and a 10% oversize diameter of 20-microns.
Figure 63 shows dispersion performance of a proprietary driven rotary mixer, on
the same high viscosity system and at the same throughput, as a function of mixer ro-
tational speed and power input. Drop diameters decrease as only the 0.13 power of
input horsepower, with inputs 100-fold greater than with static mixers being required
to yield the same 4-microns average and 20-microns oversize drop diameters. Thus,
at high ratios as well as at low viscosity ratios, the static mixer appears to be a particu-
larly effective dispersion device which should minimize problems of temperature rise
during dispersion in high viscosity immiscible fluid systems.
While the correlations already presented provide a basis for predicting dispersion
performance of static mixers on specific immiscible fluid systems if their properties
including dynamic interfacial tensions are known, the counter problem of scaling up
laboratory-scale static mixer dispersion performance for an immiscible fluid system
of unknown but constant properties is of frequent interest. The following four points
summarize a recommended procedure for handling such problems.
1. Constant wall shear (8V/D of mixer tube).
2. Constant average residence time
i.e. constant number of stages and constant ratio stage length t o stage diameter.
3. This results in
-
a. mixer length mixer diameter
b. constant mixer Ap
-
c. mixer diam. cube root of throughput
-
d. number of disperse phase feed nozzles to 0.66 power of throughput.
4. If impractical to increase number of disperse phase feed nozzles as indicated
a. then for constant number of disperse phase feed nozzles, mustfurther increase
mixer length in proportion to cube root of throughput.
In summary, I have presented for high viscosity immiscible systems, basic data and
correlations of single drop deformation and breakup in both rotational and irrota-
tional shear yields, and have indicated their application to interpretation, prediction,
and scaling of dispersion performance of static mixers on such systems.
REFERENCES
I. Taylor, G.I., Proc. Roy. Soc. (London), Ser. A146, $01 (1934); see also Ser. A138, 41 (1932)
2. Rumscheidt. F.D.. and Mason. S.G.. J. Coll. Sci.. 16. 238 (1961).
3. Bartok. W.,and Mason, S.G.. JI. Coll. Sci.. 14, 13(i959) and 16,354 (1961); Rumschcidt. F.D..and
Mason. S.G.. JI. Coll. Sci.. 16, 210 (1961).
4. Karam, H.J., and Bcllingcr, J.C., Ind. Eng. Chem. Fundam.. 7, 576 (1968).
5. Ballman, R.L., Rheologica Acta. 4, 137 (1965).
6. Lord Rayleigh. Proc. London Math. Soc., 10.4 (1878).
7. Tomotika, S., Proc. Roy. Soc. Ser., AlSO, 322 (1935); see also Ser. A153, 302 (1936).
8. Meister, B.J., and Scheele, G.F., AIChE Jl., 15, 689 (1969); 13, 682 (1967); 14. 9 (1968).

Grace 1982

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Grace 1982

Încărcat de

Drepturi de autor:

Formate disponibile

This article was downloaded by: [University of Ottawa]

On: 13 March 2013, At: 09:01

Chemical Engineering Communications

DISPERSION PHENOMENA IN HIGH VISCOSITY IMMISCIBLE

To link to this article: http://dx.doi.org/10.1080/00986448208911047

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

DISPERSION PHENOMENA IN HIGH VISCOSITY

FIGURE 1 Simple shear

droplets of antistatic o r antisoiling agents of much lower viscosity, the dispersion of

Viscosity ratio, p = p o / p , f@) =

At very high viscosity ratios, ~ a ~ l opredicted

Uniform shear field 2

fluid of low viscosity Lamp

Velocity Grodient Across Drop,Grd,cm/sec

FIGURE 3 Drop deformation.

O Value from Small Deformation Measurements in Couette Shear

0 Values from du Nuoy Ring Tensiometer Measuremants

FIGURE 4 Effect of quiescent contact time on measured interfacial tension.

00 pC= 45.5 poise

FIGURE 5 Effect of viscosity ratio.

FIGURE 6 Critical shear gradient.

,001 .01 0.1 1.o 40

FIGURE 7 Critical shear gradient. Karam and Eellinger.

rium value between mutually saturated components which could be expected to be

Viscosity Rotio, pD/pc,

FIGURE 8 Effect of viscosity ratio.

Breok at E, A2,815 " I, I# I

11 I U ~ I rn 1 ' ' 1 ' 1

Viscosity Ratio, p,/pc ,dimensionless

FIGURE 9 Effect of viscosity ratio.

Viscosity Rotio, ~ O / , L L ~dimensionless

FIGURE 10 Effect of viscosity ratio.

FIGURE 11 Effect of exceeding En on drop draw ratio at burst

VORITE 125 CONTINUOUS PHASE

FIGURE 12 Effect of exceeding Ea o n number of drop fragments at burst.

t VORITE 125 CONTINUOUS PHASE

Dibutyl phthalate at EB ,r.1, Dibutyl phthalate at 247 E Dlbutyl phthalate at 3.85 Eg

Drop Fragment Size. 5 of original drop diameter

FIGURE 14 Effect of exceeding Es on drop fragment size distribution.

-VORITE 125 / Si 30,000 SYSTEM

FIGURE 15 Effect of exceeding Es on fragment-size-to-original-size ratio

VORITE 125 / S i 5 0 0 SYSTEM

FIGURE 16 Effect of exceeding EB on fragment-size-to-original-size ratio.

Near Point 0 G' = (Apparatus Constant) (Roll Speed)

FIGURE 18 Drop break-up in hyperbolic shear field.

Dispersed lo Continuous Phase Viscosity Ratio, pO 1 uC, dimensionless

FIGURE 20 Maximum stable drop deformation-irrotational shear

FIGURE 21 Drop deformation for burst at critical shear-irrotational shear.

Viscosity Ratio, pD I vc, dimensionless

Viscosity Ratio, I P , dimensionless

Vlscositj Ratio. )ID I vc, dimensionless

Time from Start of Shear, sec

FIGURE 25 Four-roll run-shear removal and no break.

c = to exp(at + ikz) (6)

DJ/DNz JVN/2VToV WHERE DN/DT<cl 8

0.01 0.1 1.0 10.0

FIGURE 27 Observed injected jet diameter-centerline injection.

For forced jet breakup

--- ~EOICTEORAYLEIGH IlUMOTIKA THEORYI f D 3

line jet, and is entirely empirical.

FIGURE 29 Observed injected jet diameter-wall injection

EQUILIBRIUM JET DIAM. Dj CM

0 NOZZLE CENTER OF' TUBE