Two Phase Flow and Heat Transfer

Dr. Arup Kumar Das

Department of Mechanical and Industrial Engineering
Indian Institute of Technology, Roorkee

Lecture No: 09
Droplet Annular Flow and Stratified Flow Model
Hello, welcome in the ninth lecture of Two Phase Flow and Heat Transfer. In this lecture
we will be learning about droplet annular flow and we will be later on discussing about
stratified flow models. So let us see the outline of the lecture. At the end of this lecture
we will understand the following points.

We will be understanding the methodologies for generation of droplets from annular flow
and how that transforms into droplet flow. We will be correcting the Two Phase friction
multiplier for calculation of friction factors whatever we have given in your last
presentation.
(Refer Slide Time: 00:47)

That we will be correcting based on the presence of droplet in the core. In case of droplet
annular flow, you will be finding out that the annuli is shedding some droplet and which
is being carried in the core right. So that will be actually changing the friction factor at
the wall and accordingly the frictional force will be modified. So those issues we will be
tackling in this lecture.
We will be also evaluating flooding limit for droplet annular flow based on drift flux
model. So we know that there will be flooding by gaseous phase whenever there is very
high velocity of the gas okay.

So in the extreme upside direction whenever droplets are flowing inside the pipe you will
find out flooding can occur. So we will find out that what is the limit beyond which
flooding can occur from the droplet flow. So for that we will be using drift flux model
our earlier knowledge of drift flux model and at the end section of this lecture we will be
discussing about stratified flow.

We will be showing that how wave structure in stratified flow can be clubbed in
horizontal orientation and then we will be extracting the limits of large and small
wavelengths. So both the extremes we will be see the large and small wavelengths okay.
To begin with, first we will see how droplet annular flow actually generates. So we will
start with annular flow what we have discussed in the last lecture.

We know in case of annular flow around the in the wall of the tube you will be having a
liquid film. Now here I have shown one side of the tube wall with which the liquid film is
adhered okay. So I have explained the gaseous flow in the core as ug over here.
(Refer Slide Time: 2:56)
Now if you see that in magnitude the liquid velocity uf is higher than ug then you will be
finding out that due to the liquid, higher liquid velocity there will be bulge formation in
the film in the film which is adhered with the wall. And as it continues further, you will
be finding out that this film is forming a bag kind of thing over here for the air okay. So
air can accumulate over here inside that extrusion of the of the liquid which will be acting
as a bag right.

This continues to grow further and you can find out that there is lengthy protrusion of the
liquid film which further can pinch off at some point over here and form a droplet. Now
here we are having continuous flow of gas which will be carrying forward those droplets
which are being formed along with it in the core okay. So in this fashion, inside the tube
you will be finding out lots of droplets are being created of different size depending on
the flow velocities and you will find out in the core we are having lots of droplets right.

So that is characterized by droplet annular flow. Next another mechanism I will be

showing you for which droplet is also created. Here you see we have started from similar
configuration; we are having film liquid film adhered with wall. So this is this vertical
line is actually the tube wall let us consider one half of the tube wall we have shown over
here. So here we are having the core gas flow characterized by its velocity ug.

Now if we consider the magnitudes of the flow velocities are such that ug is greater than
uf then you will be finding out due to this flow velocities okay. A lamella, fluid lamella is
actually being sheared off by the fast moving gaseous stream. So you can find out this
lamella can pickup in length and you will be finding out that at some point of time
whenever lamella length is becoming very high.

There will be chances that surface tension will increase in magnitude and you will find
out that a small droplet can be shed off from that lamella okay. Now this fast moving gas
over here, we will be actually picking up this droplet and take it in the core okay. So this
two phenomenon these two phenomena are actually important for generation of droplet
from your annular flow.
The first one we name as undercutting okay and the second one is called rolling okay.
These two phenomena are actually very, very important for generation of droplet annular
flow. Now to characterize that what amount of droplet you are having in the core we
define a parameter called e. So this e is actually the ratio of droplet flow rate in the core.
So whatever volume of droplets you are having in the core or mass of droplet you are
having in the core because density is not going to change.

So that we find out over here droplet flow rate divided by total liquid flow rate. So total
liquid flow rate will be actually your droplet flow rate plus what is whatever film we are
having over here with the wall adhered with. So that flow rate also will be coming into
picture. So this total liquid flow rate will be film flow rate plus the droplet flow rate. So
the ratio between these two is actually called e. So e can be 0 whenever you are having
perfect annular flow okay. That means there is no droplet in the core right.

And in case of your droplet annular flow, you will be finding out that e is lying in
between 0 and 1. Usually typically we will find out that this e value will be definitely less
than 0.15 okay. Next let us see that how this ratio of the volumes, droplet volume and the
total liquid volume can be found out okay. So there are number of correlations available
for this but best suited correlation has been given by Ishii and Mishima in 1982. What
they said that this e can be characterized as tanh parabolic of [7.25 into 10 to the power
minus 7 jg 2 plus.
(Refer Slide Time: 07:33)
So this not jg this is modified jg actually. I will be telling you next. So, jg plus to the
power 2.5 into D plus to the power 1.25. So this is also modified diameter or non-
dimensionalized diameter we can say multiplied by Re f, so this is liquid Reynolds
number, modified version of liquid Reynolds number, no this liquid Reynolds number
actually. So Ref to the power 0.25 okay. So let me quickly explain that what are these
terms jg, D and Ref. So jg plus, you can write down this is actually a non-
dimensionalized version of jg.

So it is jg multiplied by [Rho g to the power 4 by 3 divided by sigma into g (Rho f minus

Rho g) to the power one third] whole to the power 1 by 4th. So this is actually giving you
the unit of meter per seconds sorry meter per second and this is also meter per seconds.
So you will be finding they are canceling out and this will be non-dimensional number
okay. In a similar fashion D plus can be written as D multiplied by [g into (Rho f minus
Rho g) divided by sigma. So this terminology this numerator and denominator they are
coming from the balance of your buoyancy force and your surface tension force.

So we are non-dimensionalizing the diameter of the tube by this non-dimensional length

square half, to the power half of that will be coming into picture for non-
dimensionalizing the tube diameter. So D plus is equal to D into this one okay. And all of
we know that film Reynolds number will be nothing but Rho f into jf into D divided by
Mu f okay. So all the fluid properties will be taking into picture. So once we know, you
see in this in this expressions all the values mostly are from fluid properties.

Only D is actually from your system whatever pipe diameter you are having and jg will
be dependent on your ug, so you can find out once you know the flow parameters and the
velocities you can find out what is the value of e okay. So the ratio between the droplet
mass and the total liquid mass you can find out in case of your droplet annular flow. So
that’s why this correlation is very, very important okay.

Next in the last lecture we have talked about that how to calculate the friction factor for
annular flow and there we have seen that Two Phase multiplier phi will be very, very
important, Two Phase multiplier phi g square will be very, very important. So here I will
be showing you whenever we are having droplets in the core of the annular flow how the
values of phi g square will be changing okay.

And once you calculate the phi g properly, so your calculation of the frictional force will
not be difficult and the approach for finding out the friction factor using phi g square
already I have shown you in case of your homogenous or separated flow models okay. So
let us try to see how this phi g square is changing with the respect to the presence of
droplets.

So here we are considering that we are having entrainment by the way this liquid
whatever it is coming in the gaseous phase liquid droplets those are called entrained
droplets so let us see that what is the effect of liquid entrainment okay, on the pressure
drop that means on the frictional pressure drop. So first, we will be trying to explain what
is the interfacial shear stress Taui. So Taui, we have earlier shown that this will be half f
into your Rho g into (ug minus ui) whole square.
(Refer Slide Time: 11:24)
Earlier all those i’s were replaced by ug’s. So this was the f g, this was Rho g and this
was uf okay. So now here in the core what we will be finding out it is not only the
gaseous phase we are having liquid droplets also. So the core density if you try to find out
this will be not essentially the gaseous phase density, this will be modified by the liquid
entrainment okay.

So that we need to take care of. So here what we have written that what will be the core
density now Rho co, so Rho co in ideal case, annular flow case that will be definitely
equivalent to Rho g but as we are having droplet entrained in that. So what we are doing
over here we are increasing the density by some factor which is nothing but (wg plus e
into rho f) by rho g.

Now it is very good factor to add why because if e is equals to 0 then you will be finding
out that Rho co is becoming Rho g that means if there is no entrainment then you will be
finding out it is becoming Rho g okay. Now once we have found out this Rho co, next
task to assess what is the value of interfacial velocity that means wi okay. Now for this
derivation I am considering that ui is known to us and that is equivalent to 2 into uf. uf is
essentially the liquid velocity so this is the assumption.
So if you are having some another value of ui accordingly you need to change. I have
taken this one as 2 into uf just to simplify our calculations okay. Now you see if we add
all these things in the expression of Taui, the Taui will be changing to half into fi Rho co
into (ug minus 2 uf) whole square. Though I have not placed this Rho co expression from
here to this one but one can do that one. So now I am interested in this (ug minus 2 into
uf) whole square part. So I will be explaining that one separately.

So please remember that whatever I will be deriving now from now onwards this half fi
Rho co will be coming as multiplier to that, okay. So here you see Rho g minus 2 uf ug
minus 2 uf whole square what we can do we can take common ug square. So it will be 1
minus 2 into [uf by ug] whole square. Now all of we know and we have also shown you
in the first lecture that uf by ug can be written as (1 minus x) by x into Rho g by Rho f
into alpha minus alpha by (1 minus alpha) okay. In the first lecture, introduction lecture I
have shown the proof of this one.

So we can replace uf by ug by this expression, right. Now next as we are having over
here one minus mass quality by mass quality. So what we can do depending on the
whatever masses we have in the core. And the wall we can replace this mass quality by
the liquid mass flow rate and the gaseous mass flow rate. So we will be finding out over
here liquid mass flow rate adhered with the wall will be wf into (1 minus e).

Because wf into e portion is always with the core. So that’s why we are writing wf into (1
minus e) okay and obviously wg will be remaining same. So here we find out that (ug
minus 2uf) whole square becomes this full expression. So what we can do finding out
Taui, we can hook this expression with the multiplier half fi Rho co okay. So let us see
that.
(Refer Slide Time: 15:28)
So if you proceed in this fashion. Already in the last lecture if you remember, I have
shown you based on Wallis correlation that phi g square can be written as 1 plus 75 into
(1 minus alpha) by alpha to the power 5 by 2 okay. Where you know fi is not equals to
fTP, Wallis has given some correlation and based on that we have explained this one.

So what will be finding out that with this gas entrainment in the in the core, you will be
finding out that along with this term there will be one multiplier. That multiplier is
nothing but essentially this multiplier whatever you have, so we will be finding out over
here [1 minus 2 into wf into (1 minus e) by wg into (Rho g by Rho f) into (alpha by 1
minus alpha). Whatever I have shown you in the previous slide along with this, what is
this; this is nothing but your Rho co value.

You see Rho co; I have already shown you over here. Rho co is wg plus e into wf by
Rho g. So that multiplier will be coming also over here because Rho is changing Rho g is
changing to Rho co and your and your ug square which was earlier in the Wallis criteria.
That is changing to (ug minus 2 uf) whole square that means this whole expression. So
these terminologies will be coming as multiplier over here with the phig square.

So in this way of phig square can be modified based on the knowledge of e okay. e, once
again how to calculate I have already shown you. So based on the knowledge of e, you
can modify this phig square value and calculate the pressure drop accordingly okay. Next
let us move to the flooding criteria. So as we know that we are having, we are we are
moving towards the droplet flow.

So what we need to do, we need to find out that what is the flooding condition or limit
given for the tube. So what we will be following over here. We will be following one
correlation given by Wallis once again. So what he has mentioned, he has mention jg* to
the power half plus m into jf to the jf* to the power half is equals to c.

Once again if you remember this is actually coming from our drift flux model, where we
have shown that some multiplier into jg plus another multiplier with jf equals to constant
which is a straight line manner and we have shown some examples also of pipe flow,
adiabatic pipe flow where we have found out the operating points using jg and jf values.

So similar type of equation Wallis has also proposed over here but what he has done over
here in place of jg and jf. He has given some non-dimensionliazed jg and jf to make
generalized. So let us see jg*, jg* is actually involving jg and then he has actually
multiplied Rho g to the power half divided by gd multiplied into Rho f minus Rho g to
the power half. So as he has included this diameter and the density, fluid density.

So it will be valid for any fluid pairs and as well as your pipe diameters okay. Similarly jf
also defines, so jf* is nothing but jf into Rho f to the power half gD (Rho f minus Rho g)
to the power half okay. So this is a very important criteria. So using this equation you can
find out for a given value of jf what will be the value of jg. So that you can avoid
flooding okay.
Now very important thing over here in this correlation, to find out the value of m and c.
So Wallis has suggested that c will be lying in between .725 to 1.

So this will be depending on the pipe material whatever material you are having as well
as it also depends on the pipe diameter. So you have to pick up. So he has given a chart
actually for picking up the values of c in his research paper. So one can pick up the
correct value of c by knowing the condition of your experimental setup and evaluate the
value of c. Now the rest part whatever is remaining actually m. So let us see how m can
be found out.
(Refer Slide Time: 19:39)

So what Wallis has mentioned. He has actually given two curves like this okay, for
finding out the value of m and c once again. So what he has related he said this both this
m and c they are actually dependant on your non-dimensionalized inverse viscosity. What
we have defined in case of your dispersed flow model okay.

So this, once you know the fluid properties, definitely will be finding out the non-
dimensionalized inverse viscosity and once you know the viscosity so using these two
curves you can find out the value of m as well as you can find out the value of c. It is
interesting to note that upto certain limit of non-dimensionalized inverse viscosity, the
value of c remains constant okay. After that it takes some variation.

So you have to take the values of m and c accordingly right. Wallis has also proposed that
if you are having a short pipeline that means the droplet annular flow is happening in a
short tube. So in that case you can go for simpler value because finding out m and c in
this manner is little bit tedious. So you can go for very fast and calculation, short
calculation by taking m and c is equal to 1.
So that means essentially it becomes jg* to the power half plus jf* to the power half is
equals to one so, which is an equation of straight line having 45 degree inclination. So if
you see, I have shown here that curve that this is nothing but your jg* to the power half
and this is your jf* to the power half. So if you plot this is the 45 degree line okay. So
below this you will be finding out that droplet flow can exist over there okay.

And if you go beyond this one that means keeping a liquid velocity, same if you go for
higher gas velocity will be leading towards flooding. So this this type of curves are very,
very helpful for finding out the flooding limits okay. Next let us move our to our next
part where we will be mentioning about stratified flow. So stratified flow all of we know
that is a part of your horizontal channel and you know that in case of horizontal channel
we will be having wave stratification between the liquid phase and the gaseous phase.
(Refer Slide Time: 21:51)

So, here I have shown a figure where you can find out liquid phase is flowing at the
bottom and gaseous phase is flowing at the top right. So to show the velocity vectors over
here we have shown that incase of gaseous phase. There will be higher moving velocity
compared to liquid part over here. So that means in case of Two Phase your parabolic
velocity profile exactly will not be valid. So we can find out the maxima of the parabola
has shifted towards upward side okay but there will be no discontinuity of velocity at the
interfacial point okay.
So if you consider that one in some cases, people consider the slip but here we are not
considering that slip. Anyhow so here Rho g and ug are the corresponding density and
velocity for the gaseous phase and Rho f and uf are the corresponding velocities for the
velocities and densities for the liquid phase. Now what we will be doing over here. In this
stratified flow we will be imposing the waves okay. So all of we know that let us say this
is an undulated stratified phase interface okay somewhere lying over here okay.
At a height of hg for the gas and hf for the liquid whereas total height or the channel
length is actually channel height is actually h. Now if we impose the wave over here okay
let us take the wave’s equation is etta xt is equals to etta max into e to the power ik (x
minus ct) where we know that in this type of waves k is actually the wave number
presented by 2 pi by lambda. Where, lambda is the wavelength okay and c is the wave
velocity okay, and etta max can be the maximum wave amplitude okay. So, if we impose
this wave, so this will be looking like this okay.
(Refer Slide Time : 23:59)

So let us try to assess what happens if we impose this wave on this non-undulated
structure of the stratified flow. So first if we want to calculate the wave speed, we will be
taking one important work done by Weda. He has given in 1968 that his wave velocity
can be approximated as Rho f uf If plus Rho g ug Ig divided by rho f uf plus Rho g ug
plus minus root over of you are having sigma k which is the wave number plus Rho f
minus Rho g into g by k once again divided by Rho f If plus Rho g Ig minus Rho f If
Rho f Rho g Ig into ug minus uf whole square divided by Rho f If plus Rho g Ig.
So this is actually you know your equation of a quadratic nature. So from their the roots
he has found out. Anyhow so will not be going detail of that one but using this one will
be trying to prove the extreme limits. So here by the way have not mentioned the I f is
nothing but your cot hyperbolic of wave k into hf where hf, is height, I have shown in the
previous figure and Ig is equal to cot hyperbolic of k into hg. Now if you first try to find
out what is the unstable flow condition?
Unstable flow condition means as the wave, we are imposing on the stratified flow. So if
the wave is very unstable, so then there will be chaotic situation inside the pipeline. So to
check that we have found out what is the unstable flow condition so for unstable flow
condition, definitely the wave velocity will become imaginary. So for making it
imaginary we know if we are having something under root negative than that will be
imaginary. So the limit will be finding out this one is equals to 0. So if we make this one
is equals to 0 from there, we can get what is the relative velocity (ug minus uf).
So that (ug minus uf) by making this one is equals to 0 will be coming as Rho f If plus
Rho g Ig by Rho f If Rho g Ig multiplied by sigma k plus Rho f minus Rho g into g by k
to the power half, okay. So this is very, very important equation okay. Next let us see the
extreme limits. So first one what will be showing you hf by lamda or and hg by lamda
that is higher than 0.25 that means the wave is very steep okay. So the amplitude is small
but the wave is very steep okay. So the height is actually very big compared to this
lambda okay. So let us see now what is the consequences?
(Refer Slide Time : 26:02)

So if you see already I have shown you that If and Ig is cot hyperbolic of khf and khg. So
if hf and hg they are very big values then you will be finding out that this If and Ig will
become nearly equals to 1 right. So, if I put this values of If and Ig once again back to the
limiting (ug minus uf) then you will be finding out that (ug minus uf) becomes Rho f plus
Rho g divided by Rho f Rho g multiplied by sigma into k plus Rho f minus Rho g into g
by k to the power half. It is interesting to note that in this equation; still we are having the
wave number value k okay.
So let us try to find out that for what amplitude this relative velocity becomes your
minimum okay. So let us try to differentiate this whole equation with respect to your
lambda. So if you do so then we will be finding out d/ d lambda of this part is equals to 0
for the minimum velocity because these are constant terms. So these will not be coming
into differentiation. So eventually this differentiation of this will be giving as the
minimum conditions okay.
So, once we do the differentiation, we will be getting that minimum lambda comes out to
be 2 pi root over of sigma by Rho f minus Rho g into g okay. So this the minimum
lambda for which we will be getting the stable solution okay at minimum velocity okay.
Now if we put this value of lambda min once again back to (ug minus uf) then you will
be getting expression like this. That (ug minus uf) limiting is equals to root over of 2 into
Rho f plus Rho g by Rho f into Rho g into sigma Rho f minus Rho g into g to the power
1 by 4.
This is very easy to calculate just you put the value of lambda min over there in place of
2 pi by lambda and here. 2 pi by lambda; you will be getting this equation okay. Now
here if you keep negligible surface tension okay. So surface tension is very negligible
then you will be finding out that this (ug minus uf) will be turning out to be this one okay.
So here you will be finding out that in this. We are not having surface tension, so let us
take this equation.
So over here if you nullify this surface tension value so let us say it is not surface tension
is not there, remember this is not for the minimum velocity this is only for the no surface
tension case. Then you will be finding out (ug minus uf) is equals to this factor Rho f
plus Rho g by Rho f Rho g multiplied this one okay. So that same thing we have found
over here okay. Let us see the other extreme case. So if we are having Rho f by lambda hf
by lambda and hg by lambda is less than 0.1 okay.
(Refer Slide Time 29:07)

That means you are having a very long wave length okay. So, in that case you will be
finding out that If and Ig those are turning out to be 1by k hf and 1 by k hg. So essentially
that is lambda by 2 pi hf and lamba by 2 pi hg okay. So let us do one thing put this values
of If and Ig back in the (ug minus uf). So if you put this value (ug minus uf) over here, so
I will be finding out that. Once I put these values this will be simplifying okay.
So and it is very interesting to note over here that here we have having k to the power
half and here we are having 1 by k to the power half and eventually (ug minus uf) will
become one expression involving sigma Rho f Rho g g but interestingly there is no there
is no lambda value over there or wave number value over there. So this is irrespective of
any wave number okay. Now, if we go for negligible surface tension so make this surface
tension 0, so you will be leading towards hg by Rho g plus hf by Rho f into Rho f minus
Rho g into g to the power half. If you remember your knowledge of fluid mechanics this
is typically the expression we see in case of Kelvin Helmholtz instability. So basically for
long wavelength that will be dominated by Kelvin Helmholtz instability right.
(Refer Slide Time : 30:25)

So let us summarize this lecture. In this lecture we have considered droplet flow rate and
shown the modifications of frictional pressure drop. We have correlated flooding limit
okay and highlighted its empirical constants and we have also shown you how empirical
constants can be found out. We have found out the wave formation in stratified flow
okay.
And discussed what are the wave number value relationships okay. And physical
properties involved in that and at the end we have seen the extreme limits for the wave
amplitudes that means maximum and minimum extreme amplitudes okay. So those things
we have seen and found out that whether it is dependent on the surface tension or not
okay. So with this let us test your understanding after the end of this lecture.
(Refer Slide Time : 31:09)

So first question goes like this. Flooding limit in droplet annular flow depends on, 4
answers we are having. Surface tension, viscosity, tube material and tube diameter. So I
think all of you have guess the answer, it is viscosity and tube diameter because in the
expression if you see, we are having mu as well as the capital D okay. Next question
undercutting is significant in following conditions. In the first slide I have shown you, so
probably it is very easy to get that answer.
So we are having 4 answers ug greater than uf, ug less than 0, uf greater than ug and ug
equals to 0. So correct answer is the answer c. Next, in stratified wavy flow relative
velocity will not depend on wavelength. Mention correct assumption for approaching the
statement. We are having 4 answers. Negligible surface tension, same viscosity,
negligible shear stress and fourth one is negligible buoyancy. So the correct answer will
be negligible surface tension. So with this I will be ending this lecture, thank you.