for coalescence
TRANSCRIPT
-
8/19/2019 For coalescence
1/8
Chemfcol Engineering Scknce, Vol. 41, No. I, pp. 65-72, 1986.
ocQ9-2509/86 s3.00 + 0.00
Printed in Great Britain.
0 1986. Pergamon Press Ltd.
BREAKAGE OF VISCOUS AND NON NEWTONIAN DROPS
IN STIRRED DISPERSIONS
J. S. LAGISET TY, P. K. DAS and R. KUMAR
Department of Chem ical En gineering, Indian Institute of Science, Bangalore 560 012, India
and
K. S. GANDHI
Department of Chem ical Engineering, Indian Institute of Technology. Kanpur 20801 6, India
(Received 15 February 1985)
Abstract-A model of breakage of drops in a stirred vessel has been proposed to account for the effect of
rheology of the dispersed phase. The deformation of the drop is represented by a Voigt element. A realistic
description of the role of interracial tension is incorporated by treating it as a restoring force w hich passes
through a maxim um as the drop deforms and eventually reaching a zero value at the break point. It is
considered that the drop will break when the strain of the drop has reached a value equal
to
its diameter. An
expression for maxim um stable drop diameter, d,.
is derived from th e model and found to be applicable
over a wide range of variables, as well as to data already existing in literature. The model could be naturally
extended to predict observed values of d,, when the dispersed ph ase is a power law fluid or a Bingham
plastic.
INTRODUCTION
The analysis of rate processes in liquid-liquid disper-
sions requires knowledge of the sizes of the drops
existing in the vessel To estimate the drop sizes, it is
essential to know the break-up mecha nism of a drop in
a turbulent dispersion. Hinze (1955) suggested a model
for predicting maximum stable drop diameter by
comparing the restoring elastic stress in
the
drop due
to interfacial tension with the inertial stress across the
drop diame ter. He proposed that break-up of a drop
occurs w hen the ratio of the inertial stress to the elastic
stress, i.e.
PCu* (4 40,
exceeds a critical value.
Assum ing that turbulence is isotropic a nd that the
diameter
d
1 (Kolmogoroff length), the mean square
velocity fluctuation across distance, d, is given by
UZ(4) cc (& ‘3.
(1)
If the Reynolds number is sufficiently large, then for a
fully baffled vessel, the power dissipation per unit
mass, E, can be expre ssed a s
E a N3D2 .
(2)
Substituting eqs (1) and (2) in Hinze’s proposed
criterion, Shinnar (1961) derived the following equa-
tion for the maxim um stable drop diameter,
d ,,,:
d
-225 = constant (We)-o.6
D
(3)
where We is the Weber number.
Equation (3) has been used by a numbe r of investi-
gators. Coulaloglou and Tavlarides (1976) have dis-
cussed exhaustively all the correlations available to
predict
d,
in a turbulent dispersion.
Equation (3) is based on the assump tion that there is
hardly any difference in densities and viscosities of the
two phases. The viscosity of the dispersed phase has,
however, been found to have a significant influence on
the drop size (Arai et al. , 1977; Ko nno et al ., 1982). The
viscous stress resists the flow inside the drop, leading
to
its breakage, and som etimes this stress has the same
order of magnitude as the inertial stress due to pressure
fluctuation. There exists very limited information (Arai
et al . , 1977; Konno et al. 1982) on the prediction of
maxim um stable drop size taking into account the
effect of dispersed phase viscosity. Konno et a l . (1977)
were the first to propose a model for predicting the
maxim um stable drop diameter incorporating the
effect of dispersed phase viscosity. They considered
that the deformation of a drop under external stress
can be described by the Voigt model which simul-
taneously takes both interfacial tension (restoring
force) and viscous dissipation (due to resistance to flow
inside the drop) into account. In deriving the model
equations, they assumed that the pressure fluctuation
across a distance due to turbulent flow is periodic and
the break-up of a drop occurs when the deformation
strain (0) reaches a critical value, 13~~. They have
obtained a semi-empirical correlation for the maxi-
mum stable drop diameter in terms of two dimension-
less groups, Weber numbe r and viscous number.
Although their semi-empirical expression predicts
their experimental re sults reasonably well, the model is
open to criticism on several grounds. The Voigt mode l
has a maxim um equilibrium deformation and it is
reversible. As such, it is hard to v isualize the breaking
of a drop through a classical Voigt model. Arai ef al.
(1977) overcame this problem through the artifice of
associating drop breakage with a maxim um deforma-
tion, Qr,
which is left as an arbitrary parameter.
Further, the regular and periodic pattern assigned by
them to the turbulent flow around the drop does no t
CES 41:1-E
65
-
8/19/2019 For coalescence
2/8
66
J. S. LAGISEITY et al.
appear realistic, especially on the length scale of drop
size being dealt with. Apart from this, their mode l does
not give rise to the low viscosity limit of the maximum
stable drop size naturally and had to be introduced in
an ad hoc mann er. Finally, their model does not yield a
final expression which can be use d directly for evaluat-
ing the two constants contained in the model. Instead,
the final expression is assumed and the two constants
involved in it are evaluated from their own experimen-
tal data. Such a procedure does not allow extension of
their model to other rheologically complex fluids.
The present work aims at proposing a new phenom-
enological model which accounts for the effect of
viscosity of the dispersed phase in a more ra tional way.
It also tests the validity of the model against exper-
imental results.
EXPERIMENTAL
The experimental apparatus consisted of a glass
vessel of 14.5 cm i-d. and 20 cm height. The impeller
used was a six-bladed disk turbine, placed centrally in
the mixing vessel through a stainless-steel shaft, which
in turn was connected to a motor, the speed of which
could be regulated. A set of four, equally spaced
stainless-steel balIles, arranged vertically at the wa ll of
the vessel, was used. The width o f each baffle was equal
to one-tenth of the diameter of the vessel. A schem atic
diagram of the equipment is shown in Fig. 1.
The continuous phase was taken in the stirred vessel
and the stirrer speed ra ised to the desired value. The
dispersed phase w as then added to the vessel. The
dispersed phase volume fraction was kept at less than
0.02 to minimize coalescence. The equipment was run
for 1 h to achieve steady state. Samples were then
D.14.5 cm
Fig. 1. Diagram of the stirredvessel.
drawn from the vessel and the particle sizes measu red
with an optical microscope. At least 150 drop di-
ameters were measu red to obtain the value of d,. To
avoid coalescence of drops during sampling and
subsequent size measu rements, polyvinyl alcohol
(PVA) was added immediately before drawing the
sample. This was done when the aqueous phase was
continuous_ When the aqueous phase was dispersed,
PVA was added to it at the beginning itself.
The liquids used together with their properties and
experimental conditions are listed in Tables 1 and 2,
respectively. Interfacial tensions were determined by
the pendant drop m ethod. Viscosities w ere measu red
with a coaxial cylinder viscometer.
DEVELOPMENT OF THE MODEL
When a viscous droplet becomes deformed in a
turbulent flow field, the viscous stress due to the
internal flow will act simultaneously with the interfa-
cial tension force to resist the deformation of the drop
against the external inertial stress arising from the
turbulent p ressure fluctuations. If the inertial stress is
sufficiently large and sustained over a long enough
time interval, the drop w ill deform and a thin column
of liquid will appear so mew here in its bulk. The
column will break due to instability producing two
daughter droplets. This has been assumed to occur
when the magnitude of deformation is of the order of a
drop diameter. As the interfacial tension restoring
force is absent both at zero deformation and at the
breakage point, this restoring force should pass
through a maxim um as deformation proceeds. In this
connection, it is interesting to note that based on
calculations for low Reynolds numbers, Rallison
(1984) states that for globular drops, surface tension
acts as a restoring force but that surface tension may
even promote breakage once the drop becom es
elongated. As the interfacial tension and viscous stress
simultaneously oppose the inertial stress, the Voigt
model offers a suitable description of drop deforma-
tion. How ever, the elastic stress due to interfacial
tension needs to be represented in a more realistic
fashion. The Voigt m odel corresponding to the process
of deformation is presented in Fig. 2.
In accordance with the Voigt. model, the applied
stress (T) must be equal to the opposing elastic stress
due to interfacial tension (rJ plus the viscous stress 7,).
The governing equation therefore has the following
form
7 = 7, + 7”.
(4)
Table 1. Experimental conditions
Impeller speed
Reynolds number of the
continuous phase
Dispersed phase volume
fraction
Temperature
3.33-iOrev/s
1.5 x l@-5 x 104
0.02
26°C
-
8/19/2019 For coalescence
3/8
Breakage of drops in a stirred suspension
Table 2. Properties of the $ontinuous and dispersed phases for the systems studied
67
Continuous phase
Dispersed phase
Description
c1
P
Description
K
”
T
(Poise)
(Mn3)
13/(cn=2-“)
(dynes/cm) (g/&3)
Water
0.01
1.0
Polystyrene
0.43-
1 20
0X8-0.92
in styrene
37.50
lO-30% by
wt
Kerosene 0.021 0.78 lOOm1 of 14.5 2/3 50 1
CMC in water
(2.5 %I +
60ml of 2%
PVA
CaCO, aqueous
0.137 1 45.2 1.47
suspension
(59.5 % Caco3 +
2.00 y0 polyvinyl
alcohol)
-
i
Fig. 2. Voigt model for the deformation of a drop.
Constitutive equation for T,
When the drop suffers a small but finite strain, the
interfacial tension force generates a restoring stress
proportional to the strain. However, as the drop
undergoes further deformation and approaches the
breaking point, its structure will be more like that of a
dumb bell: two “yet to be born” daughter droplets
connected by a thin filament of liquid. The filament will
break because of its inherent instability. Thus near the
break point,- surface tension has no restoring effect.
Hence a more realistic description of the role of surface
tension must be to assign to it a retractive force that
increases for small deformations of the drop but
decreases afterwards and e ventually reaches zero at the
break point. Assuming that the total deformation
undergone by the drop up to the point of breakage is of
the order of ma gnitude of
d,
the following simple
functional relationship between 5, and 0, has been
assumed.
TS= $ es (1 -e,>
e, c 1
=
0
e 3 I.
(5)
The condition that 7, is zero for 0, > 1 explicates the
essential feature of the model, i.e. the drop has reached
its break point when the dimensionless strain has
reached the value of unity and surface tension can no
longer bring it back to its original state. Thus the Voigt
model proposed in the present work cannot retract to
its original state, once 0, & 1, even when the applied
stress is withdrawn.
Constitu tive equation for TV
For the fluids considered in this work, the constitut-
ive equation for viscous stress can be written in a
generalized form as
7v = TV + K (de,/dtY.
(6)
Equation (6) simplifies to the case of Newtonian,
Bingham plastic and power law models when ap-
propriate simplifications are made.
Model equation
For a Voigt element, the total strain 8 = 8, = 0,.
Noting this and substituting eqs (5) and (6), respect-
ively, in eq. (4), we obtain
r-=,--ae l-e8)+K
d
c
1.
(7)
The flow inside the drop during breakage will be quite
complex. In the present model, however, it has been
assumed to be a simple shear flow.
Description of z
In eq. (4), T represents the dynamic pressure dif-
ference across the drop diam eter. In a turbulent flow
field the pressure difference normally associated with
the arrival of eddies is a quantity which fluctuates with
time and distance over which it operates. However,
exact knowledge of the fluctuations is not yet available.
Therefore, it is assum ed that the deformation takes
place under the influence of a mean stress which
remains constant over the average life time of an eddy.
-
8/19/2019 For coalescence
4/8
68
J. S. Llsorszrrv et al.
Expr ess ions for aver age va lu es of 7 and T
Let the average values of 7 be + and the average life
time of an eddy beF The expressions for these average
values given by Hinze (1955) a nd Coulaloglou and
Tavlarides (1977) have been used in the present
investigation. They are
and
7 ~PcU2W (8)
=&*
(9)
It is necessary to express u2 (d) in terms of the
parameters associated with the stirred vessel. The
expressions for the rate of energy dissipation per unit
mass, E, and u2 (d) given by Coulaloglou and
Tavlarides (1977) are
E = 0.407 IV= D 2
(10)
and
u2 (d ) = 1.88 s2’3 d2’3.
(11)
Therefore + can be expressed as:
+ -_ C pc N2 D4/= d2j3 (12)
where C is a constant dependent only on the geometry
of the tank and agitator. The average life time is given
by
(13)
The drop experiences no stress prior to its coming
under the influence of an eddy a nd from then onwards,
a constant stress equal to + for a tim e interval of T
Thus
r = O t t o
=p
o< t gT
Substituting these into eq. (7), we obtain
C PcN2 D413 ‘ 213 _
de n
Tg +1 -e)e+K -&
.
(14)
(>
Equation (14) is more conveniently represented in
dimensionless form as
(dlJ/dt = C We (d/D)‘ / ’
- (q,d/o) - (0 - 02) (15)
where rl is the nondimen sional time given by
q = (o /dK)“” t .
(16)
Max i mu m s tab l e d r op s i ze
The drop is exposed to a stress of i only for a time
period T After this time, the eddy wou ld have dis-
sipated its energy and the external stress on the drop
becomes zero once again. If at the end of the time
interval T the value of B -Z 1, then the surface tension
spring w ould still have a finite retractive stress and the
drop would return to its original state. Thus for the
drop to break it is not on ly ne cessary that 8 = 1 is
reached but also that it should be attained at some
point during the time intervalT. The time required to
reach 0 = 1 can be obtained by integrating eq. (15); an
initial condition is needed to do so. The dashpo t
cannot show any instantaneous deformation on the
application of the stress. Hence
8=0 at q=O_
(17)
By separating the variables, eq. (15) can be written as
18)
where
a = C We (d/D )s’3 - (T,, d/u) - l/4.
(19)
The solution of eq. (18) w ith the initial condition
depends on the constant a in eq. (19). There are two
possible solutions corresponding to a < 0 or a > 0. If
a -Z 0, it can be shown that w hen t -B co, 8 --r [l/2
- fi]_ Thus, Bean never reach a value of unity, and
breakage is not possible in the finite time interval ofT_
But tI can reach unity in a finite time period when a
> 0, and therefore to predict the maxim um drop
diameter, only values of a greater than zero need to be
considered.
Our main concern is to know the max imum size of a
drop that can ex ist in the stirred vessel. All drops above
such a size will break. As stated earliei, the drop has
been assum ed to de form under the influence of T for a
period of?? T he max imum stable drop size therefore is
the max imum threshold drop size which reaches a
deformation of unity when exposed to a constant stress
F during the time interval F. Such a drop size can be
calculated as follows.
Equation (18) is integrated to find rl at 8 = 1. If rl at
8 = 1, the dimensionless time required to reach defor-
mation correspon ding to breakage, is more than the
nondimen sional life time of the eddy, breakage wo uld
not occur. Thus for breakage to occur, the following
condition must be satisfied:
q (0 = 1) <
(u /d IQ1 ’“Z
(20)
A s
the applied stress increases with diameter, q (0 = 1)
decreases with increasing diameter. The maximum
stable drop size is that for which
q (0 = 1) = (a/d_, K)““T.
(21)
Thus eq. (18) can be integrated up to rl(0 = 1)
= (a /d_K)““T to find the maxim um stable drop
size. Using the expression for T in eq. (9), ~(0 = 1) can
be written as
r.t(e = 1) =
(Re/We)“” (d/D)2’3-“n
where
Re =
D” NoI2 - n
,
2”-‘ (3+l /n r K
(23)
Forr,=Oandn= 1,1/3,1/2and2/3,eq.(18)canbe
integrated analytically; the solutions are listed in
Table 3. It should be noted that C = 8.0 has been used
in these results for both Newtonian and non-
Newtonian fluids. This will be discussed further in the
Results and Discussion.
It should be noted that the mode l developed in the
-
8/19/2019 For coalescence
5/8
Breakage of drops in a stirred suspension
Table 3. Solution of eq. (18) for non-Newtonian fluids
n
Solution of the model equation
[32 We (d_/D)s’3 - I]“* tanm
=
(RejWe) (d-/D)- I i3
>I
Re/We)3/2 (d-/D)- 5’ 6
u2
1.6
51.2
We d_/D)“ ’ (32 We d_/D)513 -
1) +
[32 We d_/D)5’3 - l]3/2
x
tan-’
c32 We(d_;D)“” _ 1,~,2 = cRe/~~’ axP-4’3
1’3
2.66(44We(d_/D)5’ 3 -1
[(32We(d_/D)“ ” -l)* Wez(d_/D)‘ o’ “] + (32 We(d,,,f/ ” - 1)5f*
x tan-’
[32 We d_;D)= - l]“*
= [Re/
We]
’
(d- /D) - ‘ j3
present work involves a single parameter whereas
the model proposed by Arai et al. (1977) requires the
knowledge of two parameters to be evaluated from
experimental data. A positive
test that can be
applied to
the present mode l is that the constant C involved in eq.
(12), which relates the turbulent pressure fluctuations
to d, N and
D,
should not depend on the rheological
properties of the dispersed phase. If this is satisfied,
then clearly the present mod el offers a distinct ad-
vantage over the model of Arai et al. (1977) which
requires the evaluation of two parameters.
RESULTS
AND DISCUSSION
Newtonian fluids and inviscid limit
One of the interesting aspects of the model is that as
the dispersed phase viscosity becomes very small, i.e.
,ud+ 0, the solution of the model equation redu ces to
the well-known and widely tested equation
given
by
Shinnar (1961). For example, looking at the equation
for a Newtonian fluid (n = 1) presented in Table 3, the
only solution that is possible as pd - 0 or
Re -+ co
is
that We (d_/D)5’ 3
-+ constant. This result is identical
to eq. (3) reported by Shinnar (1961).
Using the data of drop diam eters against dispersed
phase viscosity for different speeds, the value of
constant C in eq. (12) has been determined. The value
of the constant worked out to be 8.0. Figures 3 and 4
show the experimental drop diameter, d,, against
dispersed phase viscosity, pd. and impeller speed. The
results predicted by the model are shown as lines
whereas the points are experimental results. The
theoretical predictions are in good agreement with the
observations. From this, it can be seen that C is
dependent on ly on the geometry of
the
vessel and
agitator, and not on cl,, . Further, it is worth noting that
the effect of dispersed phase viscosity on d, becomes
significant only after about 20 cP.
The predictions of the present model and the
numerical value of constant C can be further tested
using the experimental observations of other workers.
69
Fig. 3. Com parisonof experimental esultswith the present
model (effectof Jo)_
When the dispersed phase viscosity becomes small, i.e.
p, -V 0, the model predicts that
(d-/D) = 0.125 (We)-“* 6.
Sprow (1967) ha s given the following empirical re-
lationship when the dispersed ph ase viscosity is small:
(d-/D) = a1 (We)-“.6.
The value of a 1 ies between 0.126 and 0.15, w hich is in
agreement with the value found in the present work.
The present value of 0.125 is somew hat better because
it fits a large amo unt of other experimental data. For
instance, the experimental results of Arai et al. (1977)
also agree with the present model. This is shown in
Fig. 5.
The present model can be easily modified to in-
corporate the effect of the dispersed ph ase hold-up.
The value of the mean square velocity, u* (d), changes
with the change in the dispersed ph ase hold-up. The
-
8/19/2019 For coalescence
6/8
J. S. LAGISETN et al.0
0
0.
E 0.
1
0.
0
I
I
L
I 300
Loo m 6
Impcllsr speed N (r.p.m 1
pd c ‘3 CP
o- z 20 dyneslcm
D = 4.5 cm
Fig. 4. Effect of impeller speed on the maximu m stable drop
diameter.
I
1.5
d
I
B
/
i L-0
d
I J
/ /
/’
d’
0.5
/‘)/’
_ A;./’
___--b-
_--,_A=
Experimental
Theoretical
l3.P.M
D
xl0
0
400
0 -.- 600
a- = 20 dyneslcm j D= 6. 5 c m
Fig. 5. Comparison of experimental results (Arai et al. with
the present model.
value of m
changes with 4 as (Coulaloglou and
Tavlarides, 1977)
m+=+ = (1
+a2+)-2.0[uz(d) ]+o.
(24)
Using the modified expression for u2 (d) , the equation
for maximum drop size for Newtonian fluids in the
limit of small viscosity becomes
d
z = 0.125 (1 + r~~+)“ ~ (We)-“.6.
D
c-w
The data available in the literature is in terms of dx2.
Sprow (1967) and Coulaloglou and Tavlarides (1976)
indicate that da2 is linearly related to d,. For
continuous system s, Coulaloglou and Tavlarides
(1976) indicate that d, = 1.5 dS2. Therefore
d
2 = 0.083 (1 + a2 )‘.” (We)-o.6.
D
(26)
It is found that for a2 = 4.0, the model predictions
given by eq. (26) fit the experimental data of
Coulaloglou and Tavlarides (1976) on continuous
systems reasonably well. This is shown in Fig. 6.
It can be concluded that the value of 8.0 assigned to
constant C is related only to the geometry of the vessel
and agitator, and arises from the nature of the
turbulent fluctuations. It is not affected by the prop-
erties of the dispersed phase, which have been varied
over a wide range. The m odel itself can then be seen to
account for the variation of the viscosity of the
dispersed phase. This can be tested further by examin-
ing non-Newtonian fluids.
Non -New ton i a n i d s
Figure 7 presents the experimental results of maxi-
mum stable drop diameter with change in impeller
speed for a power law fluid (2.5 0A CMC in water
+ PVA, n = 2/3). The grade of CM C used was the
same as that employed by Kumar and Saradhy (1972).
These authors u sed a test attributed to Philipoff to
ensure the absence of viscoelasticity in CMC solutions
up to a concentration of 4% by weight. The other
compon ent of the solution use d in the present work
was PVA, with a molecular weight of about 70,000. At
this molecular w eight and low concentrations, PVA
will also not impart an elastic character to the sol-
utions. Therefore the solution em ployed in this work
can be safely considered to be inelastic. The drop
diameter decreases with impeller speed in a nonlinear
fashion. As m entioned earlier, according to the model,
+
0 0.15
*
0.Y)
. 0 05
0 0.025
a2 z 4.0
Fig. 6. Comparison of
Coulaloglou and Tavlarides’ exper-
imental values (effect of 4) with those predicted by the present
model.
-
8/19/2019 For coalescence
7/8
Breakage of drops in a stirred suspension
71
Impeller speed ti (r.p.m 1
Fig. 7. Comparison of experimental results with those pre-
dicted by the present model for a power law fluid.
constant C must be independent of the rheology of the
dispersed phase. The solid line represents the results
calculated from the model using constant C evaluated
with the experimental results of Newtonian fluids, i.e.
calculated from the second equation of Table 3. It was
found that the model predicts the results reasonably
well. This lends further credence to the model.
Gene ra l i zed rep resen t a t i on
The model predictions can be represented in gener-
alized graphs b y plotting
(d- /D)
against the Reynolds
number,
Re
defined on the basis of the viscosity of the
dispersed phase) with the Weber numb er, We, as a
parameter. Figure 8 shows plots of
d- /D
against
Re,
with We as a parameter for the values of n equal to I,
2j3 and 1 3. The plots show that at low Reynolds
numb ers, i.e. at high va lues of p,, all the curves merge,
indicating that interfacial tension does not play a
significant role in the breakage of drops. This result is
expected, since when the drop is sufficiently viscous,
the applied force essentially overcomes the mean
viscous resistance, compared to which the restoring
surface tension force is insignificant. On the other
hand, at high Reynolds numb ers, the restoring force
due to interfacial tension is predominan t compared to
the viscous forces and the applied force essentially has
to overcome the restoring interfacial tension forces.
That is why at high Reynolds numb ers different lines
are obtained for different W eber numb ers.
B i n am p l a s t i c s
For Bingh am plastic fluids the theoretical equations
can be easily developed by replacing
7
in the differential
equations for purely viscous fluids with 7 - 70)_ I t is
easily shown that this corresponds to replacing
[32 We (d _j D )s’3 - l]
by
[32 We (d _jD )5/3 - 4 7,, d-/a) -
l]
in equations of Table 3. This implies the usage of the
same value of C determined from New tonian fluids.
Figure 9 presents the experimental results of the
maximum stable drop diam eter with change in the
impeller speeds for a Bingham plastic fluid aqueous
CaCOS suspension with PVA, c = 45.2 dynes/cm T,,
= 75 dynes/cm’, n = 1, p,, = 13.7 cP). It is found that
the drop diameter decreases with increase in the
impeller speed in a nonlinear fashion. The solid line
represents the theoretical results calculated from the
first equation of Table 3 with the modification de-
scribed above. The dashed line represents the drop
diameters that would have been observed if the fluid
had been Newtonian, i.e.
7. =
0. The experimental
results are higher than those for a Newtonian fluid
since the effective applied stress is now lower. The
model reflects this and predicts the d , values quite
well, thus accruing more evidence in its favour.
Fig. 8. Effect of the Reynolds numb er on d /D with the Weber number as a parameter.
-
8/19/2019 For coalescence
8/8