Microfluidics Lecture Index

INDEX
S. No Topic Page No.

Week 1
1 An Introduction 1
2 Interface and Surface Tension 20
3 Flow Regime 1 40
Week 2
4 Flow Regimes 2 65
5 Taylor Flow 1 79
6 Taylor Flow 2 105
Week 3
7 Computational Techniques 123
8 Bubble and Droplet Generation 144
9 Interface and Surface tension 2 159
Week 4
10 Void Fraction and Pressure Drop 185
11 Liquid-Liquid Flow: Flow Regimes 217
Week 5
12 Ideal annular Flow 243
13 Taylor Flow: Heat transfer 1 257
14 Taylor Flow: Heat transfer 2 282
Week 6
15 Taylor Flow: Meat Transfer 1 296
16 Taylor Flow: Meat Transfer 2 323
17 Flow boiling in microchannels 335
Week 7
18 Flow boiling in microchannels contd. 354
19 Flow Measurement Techniques 376
20 Particle image Velocimetry 396
Week 8
21 Inertial Microfluidics 420
22 Microfluidic applications 447
23 Microfluidic applications contd. 471
24 Concluding Remarks 489
Multiphase Microfluidics
Dr. Raghvendra Gupta
Department of Chemical Engineering
Indian Institute of Technology, Guwahati
Lecture – 01
An Introduction
So, let us start with the understanding the term multiphase microfluidics. So, this course
the title of this course has 2 terms multiphase and microfluidics, and first we will try to
understand what does the term multiphase mean, and what does the term microfluidics
mean. And then try to understand each term individually and then look at the different
aspects of the course. So, the term multiphase has 2 syllables into it multi and phase. Multi
means many, and phase; phase generally refers in a general sense it refers to the
thermodynamic state of the matter which is generally gas, liquid and solid.
So, in a general sense multiphase refers to flow of gas-liquid, gas-solid, liquid-solid or gas-
liquid-solid. And if; however, it can so happen that 2 immiscible phases can coexist, or
multiple phases can coexist in the same state of matter. As in this course we are concerned
with the fluids or the flow of fluids. So, such a case can arise when we have flow of 2
immiscible liquids. For example, water and hexadecane flowing in a channel, and the
interaction between them in terms of the flow pattern, flow structure, flow behavior the
size of the droplets or the different flow structures for example, parallel flow angular flow
and so on and so forth.
So, that is about what the term multiphase means, now coming to microfluidics. So,
microfluidics as the term suggests micro and fluidics. So, micro is the flow at micro scale,
it can be flow in a channel of micron size or flow in a constriction of micro size, sometimes
it can also refer to the flow of micron size droplets, or bubbles or particles. So, it is at the
micro scale when I say micro it can be from few microns to few 100 or sometimes few
100 microns or sometimes it can be even few millimeters.
And fluidics is flow or manipulation of fluids. So, basically what we are going to study in
this course multiphase microfluidics is, what happens when we have flow of more than
one phase in microchannels.
1
(Refer Slide Time: 03:46)
So, in last 2 decades, there has been lot of emphasis in micro of fluidics and nanofluidics
or in general if I talk about micro technology and nanotechnology. If you look at the trend
in the technology has generally been towards miniature miniaturization looking at the
small and small.
So, if we look at what Feynman envisaged in his now very popular or very celebrated
lecture titled as there is plenty of room at the bottom; where he looked at the possibility of
manipulating matter on an atomic scale. That might be probably the smallest scale possible
that can be manipulated. And he described interesting ramifications of such phenomena.
So, I would suggest you to look at the lecture and understand the ramifications that he
described about 60, or about 60 years ago without any real development in the field of
micro technology and nanotechnology ok.
2
So, in microfluidics what we are looking at is flow at small scales. So now, the question
comes that what how small is small. At what size can the channel be considered in
microchannel? You will see that in the literature, the flow happening from few microns or
say tens of microns up to few millimeters; the term micro has been used frequently. So,
the question arises what is my question. So, in my opinion, the first question we should
ask what is so different about microfluidics, or flow in microchannels that we need to study
it separately.
What are the effects, or what are what is the physics that was not present in small diameter
channel or in conventional channels, but has become dominant in the micron size channel,
or small diameter channels when then we look at that effect, then probably it will be easier
for us to define or cut off if there is one such thing for classifying the channel to be a macro
channel or channel to be a microchannel?
3
So, let us look at some of these size effects. So, if we look at the different size effects, can
be classified into these 3 different classifications. And when I look at or when I talk about
the classifications, I would suggest that are not only microfluidic, but nanofluidic of going
to a small scale, one need to consider, one need to take into account, one need to check are
these effects changing; are these becoming important are these losing the significance. So,
let us look at these effects non-continuum effect, surface-dominated effect, low Reynolds
number or a low diffusion length effects one by one.
4
So, let us look at continuum hypothesis. Many of you might not be aware of what the
continuum hypothesis is. So, let us look at. We all are know that of fluid or a matter a gas,
a liquid, a solid, they are all constituted of small molecules. And these molecules are
randomly in motion, depending on what state of matter the distance or the mean free path
will depend on that. So, basically, they have in a room or in the space, there are number of
such molecules millions or trillions of those.
Now, when we look at the description for the engineering purposes. For example, for the
fluid flow, we describe the properties of such systems or properties of such materials these
fluids liquids and gases as a continuum; that means, we do not consider that at one
particular point of space, there is a molecule or there is not a molecule. We take or the as
has been written here, that the molecular effects are lumped together to space average
micro macroscopic or continuum-based quantities such as density temperature, and
velocity. we can do so because the number density of molecules. It is so large that we can
treat the molecules to be continuum, and analyze the fluid or analyze the system.
So, for reference a volume which have 10,000 molecules, there is one percent statistical
fluctuation in these average quantities. So, probably that should be good enough. So, how
we describe? Because then the question will come, that when you go to smaller and smaller
scale probably you might not have enough molecules to assume it to be continuum. So,
every time we go to a smaller scale, we need to check the continuum hypothesis ok. So,
for air at standard conditions this volume is about 3.7 into 10 to the power minus 22-meter
cube in which one can expect to have about 10,000 molecules ok.
5
So, if we look at the continuum hypothesis. It is often described or determined by a non-

dimensional number, or it can be described by a non-dimensional number which is known
as Knudsen number. And this Knudsen number is the ratio of 2 length scale, one is the
length scale for the molecular motion which is mean free path of the molecules, and
another one is the character length scale of the problem or of the system or of the channel
that we are looking at.
So, when it is obvious that, when the mean free path is very small or the sufficiently small
then the character skill characteristic length scale of the problem, then we can assume or
that or then we can use the continuum hypothesis, then the continuum hypothesis will be
valid. So, for a low value of Knudsen number where means; that means, a low value
Knudsen number means mean free path is very is very small, then the characteristic length
scale of the problem the continuum hypothesis is valid.
So, for those who do not remember what mean free path is, the mean free path is the
distance traveled by the molecules between collisions. So, the molecules of a fluid they
are moving continuously in the fluid; they have a thermal fluctuations, and because of that
they will collide with each other very frequently. So, the distance that the molecules on in
an average traveled between the 2 collisions, that is called mean free path. And for air and
other gases; this value is as you would expect that, this will be the largest out of the 3
6
phases. It will be largest for the gaseous phase. So, that is about the order of 10-8 meters
for air and other gases.
So, it can be considered, the literature suggests that when the Knudsen number is less than
10-2, then fluid can be considered continuum. And when it is greater than 1 then we have
to consider it as free molecular flow and in between there are 2 different classifications
slip flow, and transition flow with. So, based on this, we can see that in a channel, the
characteristic length scale will be few microns.
Say let us say 10 micron. So, in such a case the Knudsen number will be above 10 -2. So,
the continuum hypothesis will often be valid in microfluidics. Having said that one of
another effect of continuum hypothesis is slip boundary condition at the solid surfaces.
And it has often been a contentious issue that is there a no slip condition, at the
microchannels or is there a slip condition at the in microchannels.
So, let us look at what does this no slip and slip business is above. As we all know if you
have done a course an undergraduate course in fluid mechanics and you have studied about
the viscous fluid flow, then you will know that at a surface, the no slip conditions refers
that on a surface, the velocity of the fluid will be equal to the velocity of the surface. So,
that means, there is no slip between the solid surface, and the liquid layer adjacent to it.
This is the only this from the empirical from the experimental observations. So, now the,
some experiments in microchannels have suggested, that there is some amount of slip at
7
the walls. However, another group of researchers, another school of thought is that the slip
behavior that has been observed in microchannels is because of the roughness effects or
the air trap in the channel, and not because that there is that the no slip condition is not
valid.
So, I would suggest you to be aware about the fact that in microchannel there can be a no
slip condition and due to analyzing data, you should look at these effects we will talk about
this a bit in detail in later lectures. I would also like to point out here that, when there is
slip condition, it can often be modeled using navier slip condition; which suggests that if
this is your wall, and the liquid flowing adjacent to it if say from bottom to top, then in
case of slip, the there is some non-negligible velocity at the wall. And if you consider a
linear variation of velocity inside the wall, then the length at which the velocity will be 0
is known as Navier slip condition.
So, the parameter for such a model for the slip will be the slip length, which is λ slip, and
this has to be either found out from experiment or some other means ok. So, the next size
effects we are going to discuss is about surface dominance.
So, In microchannels surface area to volume ratio. So, let us write this surface area as A
and volume as V surface area is proportional to length squared. And volume which
proportional to L3. So, we will have this as
8
1
.
𝐿
The ratio of surface area to volume varies with respect inversely proportional to the length.
So, for one-meter channel, this value will be about 1. For 1 mm it will be about 103. And
for 10 micron is value is going to be 10-5. And that is what makes surface effects so
important in microchannels. If we look at number of physical effects, some of them are
proportional to the volume, and some of them are proportional to the surface area.
𝐴
So, in general this ratio in a conventional channel or in a macro system 𝑉
is small surface effects may not be so dominant in large channels. But as we can see as
the size decreases, this ratio becomes important, or this ratio becomes larger and larger.
So, surface force which become important in large diameter channels. Now we need to
look at a bit in detail, what these surface effects can be. So, as I have listed here these can
be friction in liquids, this can be viscous effects can be electrostatic effects.
So, and at the same for the same reason, you might hear that in microchannels, the heat
and mass transfer rates are very high. So, if you write the heat transfer rate for convective
heat transfer, you will write
𝑄 = ℎ𝐴∆𝑇
9
where A is the area of the surface. So, this area of the surface or the Q is proportional to
the area of the surface. Similarly, in the mass transfer the rate of mass transfer is
proportional to the area of the surface.
So, higher the area per unit volume; which is interface or a which is area density. Higher
this area higher the rates of heat and mass transfer going to be. So, that is the reason that
in microchannels, there are number of heat transfer applications of microchannels. For
example, in electronics cooling. In 2 phase flows or in multiphase flows, say for example,
gas-liquid flow the interfacial area that one can achieve in microchannel is 2 or 3 order of
magnitude larger, than one can achieve in smaller channels in general.
So, the interfacial area density is the interfacial area, we represent this by Ai per unit
1
volume is again proportional to 𝐿. And whereas, the interfacial area will be order of 10 or
100 in conventional channels it can go up to 103 or more in microchannels. So, it will be

beneficial, or it will be advantageous to conduct fast gas-liquid reactions. For example,
which are limited by the rate of mass transfer. So, if one have high interfacial area density;
that means, if one have high interfacial area available for the mass transfer, then the rate
of mass transfer will be high, and the yield of the fast reactions can be bettered in
microchannels.
Similarly, in the electronics cooling, or where one need to remove high heat fluxes any
such applications, one can remove very high heat fluxes by the use of microchannels ok.
Now coming at the third effect which is you can say another surface effect, that in
microchannels the viscous forces are dominant.
10
So, viscous force they can be represented by a Reynolds number; which is the ratio of
inertial forces and viscous forces or in terms of this the stresses.
So, Reynolds number is generally defined as
𝜌𝑑𝑈
𝜇
where U is the velocity of the fluid. And we will also try to see here, how we can represent
𝜇𝑈
this in terms of the 2 stresses. So, 𝑑
, which is the length scale of the problem. And inertial
stress is generally represented as 𝜌𝑈 2. So, the ratio of ρ U square and μ U by d is the

Reynolds number. And at low length scale, the d is low say microchannels the d is low,
the velocity will also often be low. So, we will have low Reynolds number. That in turn
mean that viscous effects are dominant. So, viscous effects are dominant. And the other
thing is that the transverse length scales are small.
So, when the viscous effects are dominant; that means, the flow is laminar. Now in general
in conventional channel flow is often turbulent. So, the turbulent fluid means, that there
will be a number of eddies, and the mixing of the fluid occurs because of the turbulent
flow, in a number of applications especially in chemical engineering one need to have the
mixing of the fluid. So, where as one can achieve the mixing by the turbulent nature of the
flow it becomes a problem in microchannels.
11
So, if you look at the literature in microfluidics in last 2 decades, there have been a number
of efforts or to devise different strategies to achieve mixing in microchannels, because this
is not the flow is not turbulent there, and one need to devise new ways of mixing this can
be by the means of electrokinetic effects utilizing the electro kinetic effects or utilizing a
meandering channel, or or many other passive or active scales or active means of mixing.
So, this can be a disadvantage. But at the same time when one have the flow as laminar,
then one can predict it very accurately. One can control it in a better manner. One can
know what is going to happen and knowing that one can try to manipulate and control the
flow, and do number of things precisely the way one wants to do. So, that is the advantage
of having low Reynolds number, but at the same time it is also has a disadvantage that no
inherent mixing is present in the flow. The small length scales or the transverse length
scales are small, so that means, the velocity gradient in a channel.
So, if you look at diffusion that is happening, the length scale over which the diffusion is
required is smaller and smaller. So, any diffusion process it might be momentum transfer
which is discussed way where the diffusivity is or momentum diffusivity is kinematic
viscosity, and it can be heat diffusion, or it can be a mass diffusion in all cases. The
diffusion lengths are small. So, the gradients velocity gradient or the temperature gradients
or the concentration gradient will be large, and the rate of that is transfer will be larger in
the fluid flow because the viscous force is dominating. So, the pressure drop per unit length
or will be larger ok. And we have already discussed that mixing is an issue, because the
flow does not become unstable and turbulent. So, viscous damping dominates the flow in
most of the cases ok.
So, one of the effects surface tension one of the surface effects that we have looked at in
microchannels is that what can the viscous effects do now, another physical phenomenon
which becomes very important microchannels is surface tension.
12
So, what is surface tension? Surface tension of course, occurs at an interface. So, an
interface is a boundary between 2 fluids. So, for surface tension to occur of surface tension
requires that there is an interface. So, it is an multiphase phenomenon. If there is an
interface, then it will have multiphase associated with it.
Now, what is surface tension? So, if we have a bulk fluid, let us talk about having a liquid,
and a gas above it. If there is a fluid molecule in the center, it will be attracted by different
molecules from all the sides. So, it will have an equilibrium. Whereas, a molecule which
is at the surface, it will be interacting with the molecules in the bottom half with the liquid
molecules. Whereas, the interactions with the gaseous molecules will be relatively weaker.
So, there will be a net tendency, or there will be a net pull on the molecules towards the
inside. And that tension and that will be that is what in a microscopic sense or in a
macroscopic sense is referred to as surface tension, and because of this effects the fluid
tends to minimize its surface area, so that the surface energy can be minimized.
And again this is a surface phenomenon, and it happens only at the interface between the
2 fluids. So, as we have discussed previously, that when the surface area to volume ratio
becomes higher surface tension becomes important. So, there are number of applications
in microfluidics; where the surface tension effect is taken into account or it is being
explored or it is utilized to manipulate the droplets on liquid bubbles and so on.
13
So, there is non-dimensional number, which is called Bond number, which compares the
buoyancy force and surface tension force. And this is represented by
𝜎
𝑑
and buoyancy force is
∆ ρ g d.
So, ρ so, that will be
∆𝜌𝑔𝑑
𝜎
So, this often this number is also known as eotvos number. And this is often used in gas-
liquid flow to demarcate a boundary between the minichannel or between the
microchannel and the large channels.
So, this can be used to find out for a particular gas-liquid system, at what size one will
have the surface tension effect dominating the gravity force. So, for example, in large
channels one often observe stratified flow; where the liquid is at the bottom the gravity is
acting in this direction. And the gas is at the top and one call is stratified flow; however,
in large diameter channels. So, one observes the stratified flow; however, in small diameter
channels, because of surface tension the interface does not remain like this it one has
annular flow and bubbles forming either one will have bubbles forming there or one can
have annular flow where liquid in between and the gas.
So, the bond number is as you can see it proportional to d2 . So, when the length size
becomes smaller. The effect of gravity becomes smaller, which is a body force and the
surface tension start dominating the flow. And we will see number of effects.
14
So, another non-dimensional number which compares surface tension, with another source
which is important in microchannels, which you can remember we have discussed just
now is a viscous force and surface tension. We often define the ratio of these non-
dimensional numbers in terms of forces. But in my opinion it is better to define this in
terms of stresses, because that is what we generally write while deriving these. So, for
example, viscous a stress is given by
𝜇𝑈
𝑑
And surface stress or surface tension generated a stress is
𝜎
𝑑
So, this will be
𝜇𝑈
𝜎
ok. Another non-dimensional number which is called weber number, and it is a ratio of
inertial force and surface tension force or the stress it because of the 2. So, the inertial
𝜎
stress is given by ρ U square, and the surface stress surface tension it is 𝑑. So, this is
15
𝜌𝑈 2 𝑑
𝜎
ok. So, that is weber number and one can see that weber number can be defined as a product
of Reynolds number, and capillary number Reynolds number is inertial effects divided by
viscous effects and capillary number is the ratio of viscous effects and surface tension
effects.
So, this becomes the ratio of inertial and surface tension effect. So, by the way when we
talk about liquid-liquid, there also we have our similar effects present. But this will not be
so dominant as in gas-liquid, and this is often known as interfacial tension in the case of
liquid-liquid. In case of miscible liquids, the interfacial stress is 0. For example, between
a salt water solution and fresh water solution and the interfacial stress is 0. Another effect
that has been explored or utilized very well for several microfluidic applications another
surface effect is called electrokinetic effects.
So, how does this combine? When we have a charged surface, or not charge surface, when
we have a solid surface, and it is brought into contact with a aqueous medium, which is
polar which has ions present in the liquid, then there is a charge; that is established on the
surface this charge, can be positive or negative depending on the conditions depending on
the aqueous solution. What is there in the aqueous solution? Depending on what surface.
16
So, the surface acquires a charge, and now the aqueous solution has ions which are present
in it in the aqueous solution and the surface has the charge.
So, as a result the the ions will have thermal motion as well as or the ions near the surface
will have a electrostatic force because of the attraction or repulsion from the surface, or
the from the charge present and the surface. So, a equilibrium will be established, and a
certain charge distribution or a certain distribution of the ions will happen. Just adjacent
to the surface and this charge distribution is known as electrical double layer. So, this
electrical double layer does not have significant effect in large channels because it of any
such effect can be neglected; however, this can be utilized for number of applications in
electric in microchannels.
So, the 2 effects which are often used in microchannels are electro osmosis and
electrophoresis.
So, electro osmosis is that the stationary charged surface, which is not necessarily it is
because of the electric double layer. In general, if the surface is charged and when a electric
field is applied, the movement of liquid so, when an electric field is applied on a stationary
charged surface and the liquid move, then it is known as electro osmosis. On the other
hand, electrophoresis is that a charged surface itself moves related to a stationary fluid,
because of an externally applied field.
17
So, you can see the difference between the 2 effects electro osmosis is that when an electric
field is applied the liquid moves over a charge surface it is called electro osmosis. But
when the solid or the charged surface itself moved related to the stationary fluid it is called
electrophoresis.
Now, 2 other effects which happens because of these electrokinetics 2 different potentials
are induced, one is called streaming potential and another one is called sedimentation
potential. So, electric potential reaches induced when a ion containing liquid, when a liquid
which has ion and it moves over a surface which is charged, then the electric field or the
electric potential that is generated, because of this is called streaming potential. So,
because of the stream of the fluid which contains ions, and it moves over a charge surface
the potential generated or potential induced, because of this motion is called the streaming
potential.
On the other hands sedimental potential is the potential that is induced when a charged
particle moves related to a stationary liquid. So, for example, a charged particle when it
sediments or when it drops from top to bottom say because of buoyancy, then it will induce
a potential when it is moving in a stationary liquid, then this potential is known as
sedimentation potential.
So, in summary in this lecture what we have looked at what are the size effects which are
important in microchannels. Initially we looked at 3 phenomena or we discussed 3
18
phenomena which might be important. One is the continuum hypothesis, we looked at
what continuum is, and is it valid in microchannels, and it appears that most of the cases
that we consider in microfluidics the continuum hypothesis is going to be valid.
But it is a good practice or one need to look at the validity, especially then was slip or slip
boundary condition. Then comes the surface effects we looked at that because of the high
surface area to volume ratio small scale, the surface effects becomes dominant; which
causes high rates of heat and mass transfer, the diffusion paths are small, again which
causes the high transport rates.
We looked at 3 specific surface phenomena viscous effects; which gives rise to low
Reynolds number and the flow is laminar. Then the surface tension effects; which have
number of interesting applications and the third one is electrokinetic effect. So, we will
explore more and more the surface tension effects in this course. We will look at gas-liquid
flow more and we will look at both diabatic as well as adiabatic flow in microchannels.
We will look at droplet microfluidics, bubble generation techniques etcetera, in the
microchannels.
Thank you.
19
Lecture – 02
Interfaces, Surface Tension and Wetting
Hello, in this lecture we are going to cover 3 important topic; which are fundamental to
any multiphase flows and more so for multiphase microfluidics. The first one is interfaces,
then the related is surface tension and when we have 3 phase contact lines then we talk
about wetting. So, interface is in any multiphase system interface is basically a the
boundary line between 2 phases.
Now, if we look at in the framework of continuum mechanics, or under the assumption of

continuum, where we assume that each fluid is continuous, all the molecules we can the
Knudsen number is so small. That we can assume the fluid to be in continuum. So, one
fluid and another fluid the boundary between them is called an interface.
So, in that sense this can be represented by a geometric surface that can differentiate 2
different fluids for example, gas and liquid. And the boundary between them is what we
call interface. Now, from our day to day life, we can conclude that the interface between
2 phases is generally a curve, curve if it is in 2 D and planar surface of 0 or negligible
20
thickness in 3 dimensions and these interfaces, and the interface is considered to be
smooth.
So, there might be certain instances where you have the interface number of a small
amplitude waves are present, but if you look at microscopically in under the continuum
mechanism the interface is considered to be smooth. But at the molecular level, the
separation between 2 fluids at the molecular level it depends on the interactions between
the molecules and the brine Brownian diffusion.
So, if you look at microscopically you can have a picture of an interface. So, let us say this
is a droplet on a solid surface. And so, this is liquid phase and this is gas phase, and this
wall is solid. So, what we have the interface between gas and liquid. This is the interface
between gas and liquid.
Now, if we have a molecular view of this, then what we will see is that let us say one fluid
is representing by molecules; which are filled and another fluid represented by the
molecules which are not filled.
So, that is a molecular or microscopic view of the interface.
21
So, when we have a interface separating 2 fluids, 2 different fluids. Generally, what
happens that the molecules in the bulk, they are attracted by other molecules. So, this
molecule for example, at the center is attracted by molecules on all sides of it. And in the
bulk they have a certain equilibrium. Now these interactions can be Van der Waal
interactions for organic liquids or for polar liquids like water these interactions can be in
terms of hydrogen bonds.
Now, when this molecule is at the interface, then it experiences the forces from the
neighboring molecule, but on the other side it does not experience any or even if it does,
then this the force is not symmetric. So, there is a net surface energy. So, at the interface
it has interactions with the same molecules and the interaction of let us say this is liquid
and this is gas. So, the interface the molecules which are at the interface, they have
interaction with the liquid molecules at the bottom and the gas molecules at the top.
So, there is an no a symmetry or dissymmetry that is caused at the interphase, and to take
into account, this asymmetry and the surface energy caused by this we define what is called
surface tension. So, this take into account this molecular effect. Now this phenomenon can
be for any type of interface, it can be the interface between liquid and gas, it can be
interface between liquid-liquid to immiscible liquids it can be interface between solid and
gas or it can be interface between solid and liquid. But more commonly the surface tension
22
concept or the interface and surface tension concept comes into picture for gas liquid and
liquid-liquid flows ok.
So, the surface tension is it has the units of energy per unit surface area. So, it can have
unit of joule per meter square, or it can have the unit of Newton per meter of force per unit
length. Or generally it may be represented by milli Newton per meter. It can be represented
in the literature; you might see that it will represented by the symbol 𝜎 and another place
as it is represented by the symbol γ.
So, surface tension basically takes into account the surface energy or the force imbalance
that is caused because of that interaction of the molecules at the interface is not symmetric.
So, that is taken to account in surface tension. Now if we want to make an estimate of
surface tension; let us say, if in the fluid the total cohesive energy per molecule is U, then
at the interface because it is molecule is able to interact only with half the molecules.
Then the interest the energy loss is
𝑈
.
2
And if the molecular dimension is delta, and the corresponding molecular surface area will
be proportional to 𝛿 2 .
23
So, surface tension is this energy loss, because the fluid or the molecule is at the interface
the energy loss per unit area. So, we can say that surface tension is, about
𝑈
2𝛿 2
where delta is the molecular dimension. The surface tension from this one can say that if
U is high than surface tension is going to be high, so the molecules or the fluids or the
liquids which has large cohesive energy and small molecular dimensions they are supposed
to have higher surface tensions.
For example, the highest surface tension is for liquid metals. So, for example, one of the
examples of liquid metal is mercury. And mercury has surface tension of 425.4 or about
500 millinewton per meter. And oil and organic liquid, they have a smaller values of
surface tension, because So, for example, ethanol has a surface tension value of 22.1
millinewton per meter at room temperature. For water the surface tension is 72.8 milli
newton per meter.
Now, it might be taken into account here; that when we talk about the surface tension. The
values that we represent, is generally the surface tension value with the liquid and air. the
value is generally not different very different for any other gas. So, liquid and gas surface
tension value will be like this.
(Refer Slide Time: 10:19).
24
And now, the similar arguments what we have discussed for a gas and liquid system, same
thing holds true for a liquid-liquid system; where the 2 liquids are immiscible. However,
the surface tension in the liquid-liquid system is generally known as or interpreted as or
called interfacial tension. When the fluids are miscible; that means, the molecules can
diffuse from one fluid to another, and in such case interfacial tension is 0. One example is
that you have fresh and saltwater than the salt molecules can diffuse from one fluid to
another and the interfacial tension in this case is 0 for solid-liquid case also the interface
is the solid surface that is in contact with a liquid that is the interface.
Generally, it is not easy to measure the surface tension for the solids, but many theoretical
analysis for example, we will see Young’s law later in this lecture. They use the solid
surface tension for analysis ok.
So, the surface tension it depends on or it changes with mainly 2 properties the first one is
temperature. So, if the temperature keeps increasing as we know from thermodynamics
that at critical point there is no distinction between the 2 phases gas and liquid. So, at the
critical point the surface tension becomes 0. And Eotvos found a relationship for surface
tension or temperature dependence of surface tension. So, where
2
1 3
𝜎 = ( ) (𝑇 − 𝑇𝐶 )
𝜈𝐿
25
where Tc is the critical temperature and 𝜈𝐿 is the molecular volume.
Later on, Katayama and Guggenheim, the improve the relationship that was given by
Eotvos and they given a relationship in the form
𝑇 𝑛
𝜎 = 𝜎 ∗ (1 − )
𝑇𝐶
where σ star is a constant for a liquid and n is an empirical factor. now this n has a value
of about 11 by 9 for organic liquids. Now it has been suggested that it can be the n can be
approximately close to 1.
So, the surface tension temperature dependence of surface tension can be represented by a
linear expression; where
𝜎 = 𝜎 0 (1 + 𝛽(𝑇 − 𝑇𝐶 )
where 𝜎 0 each tem the surface tension a temperature T naught. So, this is a major difference
value and from that one can calculate the surface tension and different temperatures ok.
So, the surface tension it decreases always with temperature. So, you might see this 𝛽; that
means, is negative. And when surface tension let us say at an interface there is heat transfer.
So, because of heat transfer there is a temperature gradient. And the temperature gradient
will cause a gradient in surface tension. So, imbalance of these tangential forces, because
26
surface tension causes a normal stress, but then if the surface tension is different at 2
different places at the interface, then it results in an interface a tangential force at the
interface. And the fluid from warm region for low σ to high σ a motion is established at
the interface and this motion is known as Marangoni convection.
So, Marangoni convection is basically the convection of the fluid, near the interface
because of the gradient in surface tension. And in this case, this gradient and surface
tension is caused by gradient in temperature. The fluid moves from low surface tension
value to high surface tension value, and low surface tension means at high temperature.
So, fluid from the warm regions, it will move to the colder regions ok and this effect is
propagated this effect is at the interface, but because of the viscous effects, the motion or
Marangoni convection is propagated to the bulk of the fluid ok.
Now, another parameter or another thing that affects the value of surface tension for the
fluids is the presence of surfactants. So, what is surfactant? Surfactant is it is basically
what is called surface active agent. So, these surface-active agents are generally long
amphiphilic molecules; which have a hydrophilic head and hydrophobic tail. So, hydro
means water or in general terms it can be liquid or so, hydrophilic head; that means, the
head is attracted towards the liquid or water and hydrophobic it is not attracted or it is
repelled by the water
27
So, these surfactants are generally added to biological sample to prevent aggregate
formation, or to prevent the addition of target molecules from nearing to the surface, and
it has a number of applications where one want to change the value of surface tension or
reduce the surface tension. So, the surfactants when it is introduced in the liquid they
diffuse in the liquid and then reach at the interface and because of their amphiphilic nature
they do not move at the interface and they are captured at the interface. So, they lower the
value of the surface tension of the liquid.
So, if we plot an interface and the interface, and there will be the surfactant. So, will
something like this where the head is, hydrophilic head. And these are hydrophobic tails.
So, with the increase in surface concentration when the surfactant concentration in the bulk
is increased, the surfactant concentration at the interface or at the surface will also increase
until it reaches a critical value; which is called CMC or critical micelle concentration.
So, the surfactant molecules at above this concentration the surfactant molecule start
making groups together in the bulk and formations. At low concentration, the if we plot
the surface tension with concentration of surfactant, then it is something like this. So, as
we can see, initially it decreases linearly and slowly at initially means at low surfactant
concentration the surface tension value decreases slowly. Near CMC or near the critical
concentration when the entire interface is occupied by the surfactant molecules, the surface
28
tension value decrease significantly and after that, because no more surfactant molecules
can occupy the interface because the interface is already saturated with it.
So, the surface tension value remains constant. So, like when we have temperature gradient
a Marangoni convection is observed. Similarly, a gradient in surfactant concentration also
gives rise to a gradient in surface tension. So, which again gives rise to Marangoni
convection and the motion is from low surface tension value to high surface tension values;
so, low surface tension values means high concentration of surfactant, and high σ is low
concentration of surfactant.
So, what we have learned is about surface tension and the effect of temperature and
concentration of surfactant on the value of surface tension. Now another important point
is that what is the pressure jump, or what do we call Laplace law that relates the curvature
of the surface or curvature of the interface with the pressure drop across the interphase.
So, to understand that first we need to understand what is curvature and as the curvature,
is it can be defined as the inverse of the radius of curvature.
So, curvature can be for a curved for a 2-dimensional case and it can be for a plane or
planar surface for a 3-dimensional case. So, looking at the curvature for a curve, it can be
generally defined the by the radius of the osculating circle. So, for example, for this curve,
this the curve that I have drawn by a solid line, at this particular point, the oscillating circle
the radius of the osculating circle is the radius of curvature. And the curvature which we
29
generally represent by κ is 1 over radius of curvature. So, the curvature is inverse of radius
of curvature if this radius is R then
1
𝜅=
𝑅
Now, osculating circle oscillating circle means the circle that can fit closes to the curve at
the contact points. So, the radius of curvature will be different at this point and at this point
because the radius of the osculating circle should be different at these point ok. And the
radius of curvature it can be positive it will be positive for convex surfaces, and it will be
negative for concave surface.
Now, coming to the curves the curvature in mathematical form it can be defined as in this
manner. So, for a curve which is represented by in terms of 2 variables, function of 2
variables say x and y two coordinates in 2D x and y. So,
f (x, y) =0
if it represents a curve. Then the curvature is defined as
∇𝑓
κ = ∇. (|∇𝑓|)
So, this is the divergence of the gradient of f.
30
It is direction because this is the gradient of f, and it is divided by the norm. So, this is
basically the direction vector for the gradient of f. its divergence is what gives us the
curvature. For an explicit function here we can write
𝑑2𝑌
𝑌 = 𝑓(𝑥) = 𝑑𝑥 2
𝑑𝑦 3
(1 + ( )2 )2
𝑑𝑥
And in many cases it can be approximated by d 2 y by dx 2 ok.
when so, this will be possible when dy by dx is significantly less than 1 than we can do
ok.
Now so, we have discussed the curvature for a curve. Now for a plane it becomes a more
complicated concept. So, if you have a plane. And a vector normal to the plane, any plane
that will have this normal vector will cut the plane. It will intersect this surface along a
curve. And this can be there can be number of curves, or there can be number of planes
that can contain this vector.
So, each such curve will have its own vector or its own curvature. So, the mean curvature
of a plane or a for of a surface is defined by the principle curvatures of this where the
principle curvatures are maximum and minimum curvatures; that means, the curves which
are formed by the intersection of a plane that contains this vector. The curves which had
31
maximum and minimum curvature containing this vector the sum of those is defined as
the mean curvature. So, the principal curvature, and it so happens and these the principle
curvatures are located in 2 perpendicular planes.
1 1 2
𝜅= + =
𝑅 𝑅 𝑅
1 1 1
𝜅= + =
∞ 𝑅 𝑅
So, if for example, for a sphere, there are 2 mean curvatures, and both of them are 1 over
R, and another is also 1 over R. because all mean curvatures or all the curvatures are same.
So, that is equal to 2 over R is the curvature for a sphere on the other hand for a cylindrical
surface, the maximum radius of curvature and the minimum radius of curvature. So, in one
case it will be infinite, and another case it will be R where R is the radius of the cylinder
κ is equal to 1 over infinity plus 1 over R. So, that is equal to 1 over R.
So, what we have learned, but about the curvature that how we can calculate the curvature
for a curve and the curvature for a surface. So, Laplace law, it relates the pressure inside a
bubble and or a droplet to the curvature. if you have done the excess pressure inside a
bubble or droplet is generally given by
32
2𝜎
∆𝑃 =
𝑅
for a spherical bubble or droplet.
So, let us see how we can find out this expression. now consider a sphere or a spherical
droplet, which has a radius R. and the pressure inside this droplet is p 0 outside pressure
is p 1, and now we want to extend this spherical droplet by an infinitesimal amount so that
the radius become R+ dR. and what we need to do is to calculate the work that is required
for this expansion from R to R + d R.
So, in such a process, there are 2 different type of works that are being done. The one that
will be equal to; let us say
𝜕𝑊𝑒𝑥𝑝 = (𝑝1 − 𝑝0 )𝑑𝑉
𝜕𝑊𝑠 = 𝜎𝑑𝐴
del W for expansion that will be equal to p dV work because there is a change in volume.
So, dV and the pressure is p 1 - p 0 because it is working against this pressure. So, p 1
minus p 0 dV that is the work for this expansion, and then the work due to surface change
in surface area. So, that will be equal to σ where σ is surface tension into dA.
33
(𝑝1 − 𝑝0 )𝑑𝑉 + 𝜎𝑑𝐴 = 0
𝑑𝐴
(𝑝1 − 𝑝0 ) = 𝜎
𝑑𝑉
8𝜋𝑅𝑑𝑅
(𝑝1 − 𝑝0 ) = 𝜎
4𝜋𝑅 2 𝑑𝑅
2𝜎 𝜎
∆𝑃 = 𝑝1 − 𝑝0 = = 𝜎𝜅 =
𝑅 𝑅𝑐𝑢𝑟𝑣
So, at the thermodynamic equilibrium, this total work is equal to 0; that means p 1 minus
p 0 dV plus σ dA is equal to 0. So, from this we get this expression that, p 0 minus p 1 is
equal to σ dA by dV.
Even though we have done this derivation with respect to a spherical bubble, but this can
also be done for any arbitrary shape of the interface. and dA for a spherical bubble will be
for because A is 4 π R square. So, dA is 8 π R dR. and V is 4 by 3 π R cube, and dV is
equal to 4 by 3 π into 3 R square dR. So, 3 3 will cancel out that dV is equal to 4 π R square
34
dR. So that means, that excess pressure is equal to σ dA is 8 π R d R divided by 4 π R
square dR.
So, p 0 minus p 1 is equal to 2 σ over R for spherical droplet. and in general for a surface
we can give σ as equal to dA over dV. This you can say excess pressure, or the pressure
difference or the normal stress jump, that is defined and the interface that delta p is equal
to σ in general it is given by σ κ and. So, the κ is the curvature or σ over radius of curvature.
So, this law is known as Laplace law. And this tells us the normal stress jump or pressure
jump across the interface for a bubble or a droplet. So, it depends on the curvature κ.
So, until now we have been talking about the interface between 2 fluids. Now if we look
at the 3 phase, there will be 3 phase contact between the fluid. So, you have a solid surface,
and at the solid surface there can be a droplet sitting over it. Now this droplet may be
sitting like this, or it may spread like a thin liquid film.
So, this is liquid film, now it depends that when these 2 different phenomena will occur.
Sometimes the droplet might be sitting as a semi spherical or part of spherical part of a
sphere, or portion of a sphere. And other case it might be a planar liquid film. So, these 2
different situations when it wets it completely and makes a film then it is called complete
wetting, wetting and when it is forming a droplet part of a sphere then it is called partial
wetting. And this depends on that if it is a completely wetted surface or partially wetted
35
surface, it will depend on the properties of the solid surface liquid as well as the gas
surrounding it.
So, a liquid spreads if the energy of the system is lowered by the presence of the film. So,
that means, that if a the surface energy per unit surface area, let us say for this is σ this is
solid surface and there is a gas surrounding it. So, this is 𝜎𝑆𝐺 .
Now if we have a liquid film, and it wets it then we have 2 different surface energies here.
This is liquid, solid and gas at the top. So, here we will have σ liquid and solid, plus σ
liquid and gas. So, if the presence of a liquid film lowers the surface energy then complete
wetting will occur. So, let us write this. If the presence of a liquid film on the solid surface
lowers the total surface energy of the system, complete or total wetting would occur. If
that is not the case then only partial wetting will occur.
So, in case of partial wetting 𝜎𝑆𝐺 is less than 𝜎𝑆𝐿 +𝜎𝐿𝐺 .
36
So, in case of partial wetting, the contact may be hydrophilic or hydrophobic. So, it is
called hydrophilic, let us define contact angle first. So if we have a droplet and the angle
between the solid surface and this and interface is called θ. And it can take another kind
of shape where the angle is more than 90 degree. So, in this case θ is less than 90 degree,
and in this case, θ is greater than 90 degree.
So, θ is less than 90 degree or θ is called contact angle. And if the contact angle is less
than 90 degree, then it is considered to be a hydrophilic surface, and if it is more than 90
degree, and it is called hydrophobic for water the term hydro is used for any general liquid
one can say lyophilic and lyophobic contact ok.
37
Now, Young’s law, which gives us the contact angle for the surface turns, and for different
surfaces, let us consider a droplet which is placed on a solid surface and the contact angle
at this surface is 𝜃 and if we defined the surface tension for different interfaces. So, for σ
liquid and gas this is σ solid and liquid, this is σ solid and gas. So, as we have seen that
surface tension can be considered as force per unit length.
So, using the force balance if this is our, we can have a coordinate system, in the x direction
along the surface, and y direction normal to it. So, force balance at in x direction if we
balance the forces. then we can see that 𝜎𝑆𝐺 force per unit length 𝜎𝑆𝐺 will be equal to 𝜎𝑆𝐿 ,
plus 𝜎𝐿𝐺 cos θ, where θ is contact angle. So, if we rewrite it, we will get cos θ is equal to
𝜎𝑆𝐺 minus 𝜎𝑆𝐿 over 𝜎𝐿𝐺 .
So, this law is known as Young’s law. And this gives us a relationship between the contact
angle, and the 3 different surface tension values. So, in summary in today’s lecture what
we have covered is basics of surface tension. First we talk about what interface is, and in
or we can consider the interface to be surfaces of 0 thickness, and smooth surfaces which
delimits or which separates the 2 fluids.
Now, we also discussed that because of the different or asymmetry or dissymmetry in the
molecular attraction set the interface because at one side the fluid, experiences the
attraction from one kind of molecules, and on the other side of the interface it experiences
attraction from different kind of molecules. So, there is asymmetry or there is surface
38
energy generated. And this is accounted for in surface tension. And this surface tension,
there are 2 different 2 2 laws that we have discussed one is Laplace law, or it is also called
Young Laplace law. So, this gives us delta p is equal to σ into κ in general.
So, delta p is the excess pressure excess pressure means the pressure inside the liquid
droplet or gas bubble, minus the pressure surrounding it. That pressure difference or the
normal a stress difference is σ κ; where κ is the curvature and σ is surface tension. The
surface tension depends on temperature, and surfactant concentration.
So, with increase in temperature surface tension decreases and with increase in surfactant
concentration also surface tension decreases and if there is a gradient in surface tension at
the interface then we see a convection setting up at the interface, and this convection is
known has Marangoni convection. Another law that we have discussed today is Young’s
law; which relates the surface tension between 3 different phases or surface tension for 3
different phases with the contact angle.
39
Lecture – 03
Gas-Liquid Flow: Flow Regimes
In this lecture we will talk about different flow regimes that may occur in gas liquid flow.
So, to understand gas liquid flow regimes in micro channels or small diameter channels,
first we need to look at what are the other flow regimes, or what are the flow regimes that
had been studied that have been found in conventional channels. So, as to appreciate what
is different in macrochannels and microchannels, so let us look at the flow regimes.
So, first this word flow regimes comes to our mind, and if we have had no exposure to
multiphase flows. But we have studied single phased flow in a typical fluid mechanics
course at an undergraduate level.
Then you would have been introduced to the terms such as laminar flow, turbulent flow
and in between transition flow. So, our understanding of the flow regime will be with the
when we say flow regime, then you will immediately think about laminar flow regime and
turbulent flow regime. So, let us look at the why do we have laminar flow regime and
turbulent flow regime. So, if you remember, we classify that a flow regime is laminar or
40
turbulent by a Reynolds number. And this Reynolds number is a ratio of inertial as well as
viscous effects inertial and viscous effects
So, the Reynolds number is a ratio of inertial effects, and viscous effects. And when see
when we say that the flow is laminar. So, we say that it is at low Re and; that means,
viscous forces are dominating, let us say viscous effects dominate. Whereas, in the
turbulent flow, it happens at high Reynolds number and we have inertial effects, or inertia
dominate the flow. And this is in between at. So, in a single-phase flow because of the
dominance of one particular physical force, or one particular physical effect we have
different flow regime.
So, that is that a one thing that we can see from this that the flow regimes occur because
of the dominance of one particular physical effect one particular physical force at certain
flow conditions ok. Then if we look at the flow structure, we call it laminar, because if this
is a channel and these are the channel walls the flow happens as laminar. These laminar
might be moving with different velocities. So, the flow is happening as different layers of
fluid, and these different layers of fluid do not have much mixing.
So, this is what we call because the flow happens in terms of laminar, and we call it laminar
flow. Whereas, in turbulent flow we have number of eddies, and then these eddies can
have different flow structures. So, they can be of different size and so on. So, in terms of
the flow behavior or flow pattern, what we have is we get different flow structure; which
is caused by the dominance of different physical effects in one particular flow regime.
41
So now, moving on if we think about 2 phase flow, what else do we need to study in 2
phase flow? We already have inertial effects, but now we have 2 phases. So, depending on
the phase flow rates, we will have the inertia of a gas, and inertia of liquid. So, we can
have a inertial effects of a gas and inertial effects of liquid. Then we will have viscous
effects, because viscous effects in general or at least in whatever we will be discussing in
of flow in channels; viscous effects will be generally important for those of liquids.
So, we consider the viscous effects, one might consider because the viscosity of gas is
sufficiently small or at least 1 or 2 order of magnitude smaller than that of liquid. Say, for
example, if we are looking at water and air, the viscosity ratio is about 40 or 50. Now one
might argue that the inertial effects might not be important, or the gas inertial effects might
not be important from the same argument. But in some cases, for example, in annular flow
the gas velocity is very large and the inertial effect is proportional to the velocity squared.
So, in such cases, where the gas velocity is very large the inertia of the gas and liquid can
actually be of the same order of magnitude ok. The 2 additional effects which were not
present in single phase flow are I have listed here, are gravity and surface tension So, the
gravity one might say that the gravity is present for single phase flow. But because the
flow is homogeneous a single-phase flow the properties are same everywhere. So, there is
no buoyancy effect but when we have a body force, and there are 2 fluids, then because
body force is a function of the density of the fluids. So, if the 2 fluids have different
42
densities, then the 2 fluids will respond differently to this body force. For example, we
have taken gravity here. But if there is a centrifugal force acting on the fluid, then the 2-
fluid will have different response, because of this centrifugal force ok. Similarly, the other
force is surface tension force. So, that acts at the interface between the 2 fluids, and that
will be the stress by surface tension is σ over l, and for buoyancy ∆𝜌𝑔𝐿 and that will be
the stress.
So, these are the different effects that we need to consider, when we are looking at 2 phase
flow. Now the things were simpler when we were looking at single phase flow, because
we had only 2 effects. But now here even without considering any other effect for example,
electrostatic force or magnetic forces, or centrifugal forces. We have our 2 more variables,
or 2 more forces coming here. So, we have to consider all these forces. And this gives rise
to different morphological patterns. So, that different morphological pattern means, the
different arrangements of the 2 fluids in the channel of course, when these fluids are
arranged in different manner they will affect the flow field as well.
So, we will have the flow to be laminar, and flow to be turbulent, we might also have that
one phase is laminar, and another phase is turbulent, and they will interact and affect the
other forces. So, it becomes very complicated for 2 phase flows and in 2 phase flow
because the flow structure which is that how the 2 phases.
43
Because in this lecture we are talking about gas liquid flows. So, how the gas and liquid
they arrange themselves will be a important criteria at the first hand. But after that when
we look at the modeling, we also need to understand the laminar and turbulent nature of
the 2 fluids.
So, all the properties that are of engineering interest for us for example, pressure drop or
energy loss or volume fraction after the fluids have been settled or the heat transfer. So, in
a set number or mass transfer mass transfer coefficient and heat transfer coefficient, they
all will depend on how the 2 phases have arranged themselves, while they are flowing
through a channel. So, all these parameters depend on the flow regime. So, when we start
looking at the 2-phased flow or gas liquid flow in a channel the first thing we need to look
at that for a given set of conditions for the given set of a geometrical the channel geometry,
what are the flow regimes that we are going to get.
Then we can study different flow regimes in detail, and try to understand; what are the
different parameters of engineering interest, that we want to study or that we want to
understand.
So, the flow regimes will depend on different physical factors, the first one I have listed
here is density difference. So, as I said because of the density difference, the 2 fluids will
respond differently to the body forces. So, the flow regimes will change, surface tension
will again play an important role. And it will tend to make one phase dispersed.
44
Generally, for a gas liquid combination at a particular temperature one will have a constant
surface tension. But if there are impurities is which are generally unavoidable in the in any
industrial scenario. Then we will have surface tension variations also; which might affect
the size of the bubbles for example, or the size of the droplets. And the temperature effects
are there for example, if one has boiling or condensation then the temperature gradients in
the fluid will also called surface tension gradient. So, that may have an effect on the flow
regimes. Then one of the common things one will have is say phase flow rates.
So, depending on the gas and liquid flow rates, and the relative flow rates one will have
different flow regimes. Generally, the flow regime map is plotted between the flow rates
or superficial velocities of the 2 phases. It will also be a factor or it will also be a function
of the velocity of the liquid because the gas velocity or the gas viscose effects are generally
negligible. So, it will depend predominantly only on the liquid velocity again, the system
configuration. So, what kind of channel one has, str straight channel or if it is a serpentine
channel for example, or if it a channel of circular cross section, or if it is a helical channel,
the flow regime will vary.
The orientation will also play a role. The orientation is connected with the effect of gravity.
So, if you are doing these experiments on earth a vertical channel, and a horizontal channel
will have a different responses to gravity. So, you will have generally the flow to be
axisymmetric if you have a vertical channel. But that will not be. So, in a horizontal
channel, and if you have inclined channel then you will have depending on what is the
angle of inclination. You will have different flow regimes.
So, extending over that, if you are doing those experiments on moon say; where the gravity
is 1 by 6 that is on the earth, then you will have different effect of gravity. So, the flow
regime map will change. So, people have done a lot of experiments for microgravity, where
the gravity is almost negligible, and that is generally achieved by parabolic flights. So,
people have done experiments in parabolic flight to understand the 2-phased flow behavior
for space applications for example.
Then channel size will also have an effect and that is the reason exactly why we want to
study here it for micro channels. And where if there is a phase change, so that means, if
during the flow the phase is changing. So, the volume of that 2 fluids are changing, then
we will have a different flow regime or we can even have a flow regime transition. So,
45
considering all these effects we will not consider phase change right now, we will consider
this during when we look at the boiling or evaporation. So, let us look at the different flow
regimes.
In conventional channels. So, if we have vertical co current upward flow. So, that means,
the gravity is acting opposite to the flow, flow is happening in the upward direction, and
the gravity will of course, act towards the downward direction. And depending on the gas
and liquid flow rates, one will have a different flow regimes. So, in this typical flow regime
map at a constant liquid velocity, if we have a small amount of gas introduced in it, then
the bubbles will be generated and then these bubbles at low volume fraction of gas, these
bubbles will start rising.
When you increase, further increase the velocity of the gas, then one will get slug flow;
these bubbles will start increasing the velocity. And they the number of bubbles will
increase and the number, when the number of bubbles will increase they will coalesce.
And then they will become larger bubbles and then these bubbles, what we call generally
it is important to know the name of these bubbles, they are generally known as Taylor
bubbles. And they are very important in microchannels. So, it is better to understand and
recognize them early on.
So, these bubbles Taylor bubbles are of the size of the channel, and there is a; very small
thin liquid film, that separate them from the channel wall ok. Further increase in the gas
46
velocity, we might have a churn flow, and once the velocity of the gas has increased to
very high values. what we will have is annular or dispersed flows. So, where gas is in the
core, and there is liquid film on the wall. We will also have this gas will have droplets into
it. At high liquid velocities, especially when the liquid flow is turbulent, these bubbles are
broken into a smaller bubbles, because of the turbulent effects of the liquid. And now one
will have a dispersed bubbly flow where the bubble size is small.
So, that is finely dispersed bubbles in bubbly dispersed bubbly flow. So, these are the 5-
major flow regimes that one can classify in a vertical co current of upward flow. There is
this thing with the flow regimes in macrochannels, in microchannels that different people
have seen different flow regimes. And at the transition, one will have a different flow
structure coming into picture. So, sometimes of flow regime a same flow regime might be
given a same name a one flow regime is given one name, by one set of research and another
name by another set of researchers.
Similarly, the 2 different flow regimes way which are morphologically different, might be
given the same name by 2 different researchers. So, one need to understand and look into
the flow regimes what flow regime is being talked about. But the research on say, a flow
regime in macrochannels have matured and people have looked at into these flow regimes
for last 60 years or so now, and they have agreed upon some of the broad flow regimes
that are there.
47
So, in the horizontal co current annular flow as one will think intuitively, that at low liquid
flow rates one gets a stratified smooth flow. So, for example, we have the flow in the
drains, where the liquid flows at the bottom and the air if it is flowing with a small velocity
it will flow, and then the interface which is smooth.
When the gas flow rate is increased further, there will be some waves that appear on the
on the liquid surface. And if the gas flow rate is increased further. So, all these flow
regimes have been plotted with increasing gas flow rate. And if that gas flow rate is
increased further then the amplitude of these waves will increase. And then one will once
the amplitudes will increase, they will at one point they will start touching the wall. So,
one will get slug flow. So, if you remember in the previous case we had slug flow; where
the flow was axisymmetric, and the bubbles are not touching the wall.
Whereas in this case, we have the bubbles touching the wall, and they are not
axisymmetric, but the same flow regime is termed as slug flow ok. At further high gas
flow rates, we will get annular flow. Again, it will not be axisymmetric the film thickness
at the bottom, and the film thickness at the top you can see from this schematic itself, that
it is non-axisymmetric. The film will be thicker at the bottom, and actually there will be
some drainage that might that will be happening.
So, when they have dispersed annular flow in which the gas core you will have droplets
disperse, that is what have been shown here. Whereas, in the when the liquid flow rate is
high. So, if it occupy the entire channel, then at low flow rates one will get bubbly flow,
but because of the density difference and because of buoyancy, these bubbles will be first
appearing at the top of the channel. If the gas volume fraction increases, and the liquid
flow rate is very high, then these bubbles will be finely dispersed. So, one get dispersed
bubbly flow.
Further on when the bubble volume or the gas volume fraction increases further, and then
these bubbles will have to coalesce, because there is not much space for them to move on.
So, they start coalescing and a slug flow, or what has been given as name as plug or
elongated bubbles. So, one can have this flow structure, or one can have these bubbles
separated from the wall but located towards the upper side of the channel. So, again Taylor
bubble kind of a structure one will observe, and then again at high liquid flow rates one
will have a high gas flow rates, and high liquid flow rates one will have annular flow ok.
48
So, you might notice here, that in the vertical flow we did not have stratified flows. In the
vertically upward flow what we studied there was no stratified flow in the channel for the
simple region, because the gravity is not acting in such a manner that it will allow or the
buoyancy is not acting in such a manner that the phase separation can happen ok. So, this
is one of the differences between the large channels was a vertical flow and horizontal
flow.
So, based on this now the question comes for example, one is looking at a particular flow,
then the first thing in terms of the models or in terms of the prediction one would like to
have that can I have a tool, where I can give my input parameters of the flows for example,
gas flow rate, liquid flow rate, the properties of the fluids, viscosity, densities, and the
surface tension, and the temperature, and a pressure conditions, and the channel size etc.
And one can is it possible to come up that for these set of conditions one is going to get
these flow regimes. So, this is the first thing than one would like to have. So, based on
those experiments that people have done over last 30, 40 years or maybe more, some flow
regime maps have been prepared and one of the common and frequently used for upward
co current flow. So, why we are studying only upward co current flow or why we will look
at only upward co current flow, because there are some similarities in the flow in micro
channels and upward co current flow.
49
So, as we have seen just now, that in the vertically upward channel, there was no stratified
flow, and the flow was axisymmetric. Similarly, in the micro channel, the effect of gravity
will be less ok. So, we will discuss that maybe in detail when we look at the flow in
microchannels, but let us look at this ∆ ρ g d square over 𝜎. And that is ratio of buoyancy
and surface tension, because this is proportional to d. So, as the d becomes smaller, the
effect of buoyancy will become smaller. So, at lower bond number; that means, when the
buoyancy force the importance of buoyancy force start decreasing at lower bond number,
and the effect of gravity will be reduced, and finally, become negligible as the channel size
decreases.
So, some of the effect that comes because of that buoyancy effect for example, stratified
flow asymmetric flow, that will be not observed in microchannels, which is similar to at
least visually what we have observed in vertical channels. So, we will look at the upward
co current flow and try to see if we can understand something from it which might be
useful for flow in microchannels. So, this is a flow regime map. This has been plotted
between 𝑗𝐿 and 𝑗𝐺 . So, where 𝑗𝐿 and 𝑗𝐺 are superficial velocities, L and G refers to the
phase. So, 𝑗𝐿 is liquid superficial velocity, 𝑗𝐺 is the gas superficial velocity.
So, for those who do not know what a superficial velocity is, the superficial velocity
sometimes it is represented at 𝑗𝐿 , or it has also been reported as ULS. So, U liquid
superficial, that is Q liquid divided by channel cross sectional area ok. So, if it a circular
channel, then one can have QL over pi R square for a cylinder. Or maybe or this will be
saying it as circulars channel. Similarly, the gas superficial velocity, or 𝑗𝐺 is equal to QG
over area of cross section. So, this flow regime map has been plotted between liquid
superficial velocities and the gas superficial velocities.
And based on the occurrence of the flow regimes, they have plotted that at low liquid and
low gas superficial velocity, actually a high range or a very large range of a liquid
superficial velocities what one observe is bubbly flow at very high gas velocities what one
observe is annular flow and in between one observes slug. And after slug one can have
slug or churn. At very high liquid velocities one observe finely dispersed bubbles. And
this flow regime map or representative flow regime map is based on after publication sorry
Taitel’s publication in 1980.
50
So, they have identified some of the transition regimes. Specifically, what we will be
talking about here is bubbly to slug transition and slug to annular transition. So, from
bubble flow to slug flow transition, which is this and slug to annular flow transition which
is this. So, bubbly to slug transition, what happens that initially the bubbles are generated,
and then they move. As the volume of the gas is increased, the number of bubbles will
increase. And when the volume increases, so much that it is difficult to avoid the
coalescence of some bubbles, then these bubbles eventually form and become larger
Taylor bubbles. And Taitel has suggested that happens at volume fraction or homogeneous
volume fraction of about 0.25.
Others have suggested that it happens about 0.3. So, that is about the range 0.25 to 0.3
when this transition happens. Now because the gas liquid slip; so, slip means that if there
is no slip between the gas and liquid, then the gas will be moving with the same velocity
as that of the liquid, and that will be true for when the bubbles are finely dispersed because
these bubbles in the flow around these bubbles between the stokes flow regime. And they
will respond quickly to the fluid and you will have that they will move with the same
velocity as that is of the liquid.
But for larger size of bubbles, there will be a certain amount of slip and mostly in general
the bubbles will rise especially in the upward flow the bubbles will rise faster than the
liquids. So, this slip is defined here, that UG minus UL is equal to the bubble velocity. So,
51
the bubble rise velocity is defined by UG minus UL or slip between the 2 phases is given
by the bubble rise velocity. And this bubble rise velocity from the literature, it can be found
as a ratio of ∆ ρ G σ, over ρ L square1/4. So, when this is the bubble rise velocity.
And one substitute the bubble rise velocity, and the fact that UL which is liquid velocity
is equal to jL a superficial velocity over α and UG is equal to j G over α which is α is the
volume fraction of gas, and this is volume fraction of liquid.
So, let us write this is 1 minus α G or you know we can just write 𝛼 as αG is equal to α. So,
remove the subsscript. So, when we substitute this here, and then take α is equal to 0.25
one will get this expression. So, from this if the liquid superficial velocity, one can come,
one can compare the 2 terms; which is liquid superficial velocity on the left, and this term
on the right. if the liquid superficial velocities is greater than the right-hand side, then one
gets a bubbly flow, and if the liquid superficial velocity is less than the right-hand side,
then one gets slug flow.
Another observation that has been made and it is relevant for us that this bubbly flow
cannot be sustained, when the rise velocity of a Taylor bubble is less than that of the
smaller bubbles. So, if somehow there is a bigger bubble a channel size bubble has been
formed. So, that means, the volume fraction is already significant. And when this larger
bubble has been formed, and it is moving slower than the smaller bubbles then what will
52
happened these bubbles will come and entrain into the larger bubble, and when we will
get a larger table Taylor bubble forming.
So, that simply means that when that has happened that, the bubble velocity is larger than
the Taylor bubble velocity, then the bubbles will or will not be sustained and they will
merge into the tailor bubble ok. So, they have suggested this or they have pointed out
Taitel et al. that it is difficult to sustain bubbly flow in smaller tubes.
And when the rise velocity the same fact than the when rise velocity of a Taylor bubble is
less than that a smaller bubble is sporadic, that it sometimes once upon once in a while
occurrence of a tailor bubble. That will cause the bubbles to coalesce and the tailor bubble
flow will develop.
So, the velocity of a freely rising Taylor bubble because of this in a quiescent liquid just
because of buoyancy the velocity of a rising Taylor bubble is given by 0.35 √𝑔𝐷 ok. So,
you will need to remember, that this is for freely rising bubble that is why it is only the
function of gravity and the channel diameter. And so, that means, we have an estimate for
the bubble rise velocity in a channel, which is a function of the channel diameter. And the
bubble velocity that we have seen just before that this bubble velocity can be given by this
expression.
53
So, if the Taylor bubble velocity is less than the bubble rising bubble velocity, then we
will not have a bubbly flow. So, the next question will come to mind ok, let us find out
what is this D. And it turns out if you substitute this the D is 5.1 centimeter for a freely
rising bubbles. So, but remember that this condition depends on that one should have a
sporadic occurrence of Taylor bubble. So, once you have a significantly large bubble, and
which can be Taylor bubble, which can have this velocity, then one will have this thing to
be valid ok.
Now, looking at finely dispersed bubble, which is the boundary between the 2 bubbles, or
the finely dispersed bubble; and the bubbly flow and slug flow, then this when can this
happen, that these bubbles are nearly spherical bubbles, and then they remain discrete
because of a strong turbulence effect. So, there are bubbles when the liquid is a strong the
turbulence is strong in the liquid the inertial effects are high the fluctuations they break up
the larger bubbles.
So, the larger bubbles are broken because of the turbulent effects, and at a beyond a critical
size these bubbles are break up because of the hydrodynamic forces or the inertial forces
or the turbulent fluctuations. So, this critical size of the bubble will depend on the level of
turbulent energy dissipation or the inertial factors that we have. So, in non-dimensional
numbers that can be a Weber number.
54
Then another important transition, for us is annular to disperse all or from slug to annular
from slug flow, to annular flow or annular disperse flow regime transition, that will
happen, because when we have Taylor flow then we have the gas bubbles which are of the
size of the channel and they are arranged in a periodic manner and between 2 bubbles are
the liquid slug.
So, as the gas volume will increase what will happen? The size of this bubble will become
larger and larger, and the liquid that is there between the 2 bubbles or 2 conjugative
bubbles, the size of that liquid or liquid slug will become a smaller. So, eventually because
the gas velocity is becoming higher, when we move from slug flow to annular flow, so the
gas inertia will be larger. And this the shape of the bubble, what we have is say, the bubbles
shape is like this, and there is a bubble at the front which is like this. And this shape is
because of surface tension and the inertia is acting, ρ u2 .
So, at a particular when we increase the gas inertia. At a particular Weber number, the gas
inertia will be able to overcome this the diameter the that will overcome the liquid bridge
that is there between 2 bubbles, because when before coalescence or just before transition,
when we have something like this.
So, this liquid will be broken. Now the other thing is the other factor that comes into factor
the other effect that comes into factor, that gas velocity is sufficiently high to shatter the
liquid slug between 2 bubbles into smaller droplets. So, what will happen when this liquid
55
slug when it is thicker let us say is sufficiently thicker. Then if the gas inertia is sufficient
and it can break it up a smaller droplets, and then these droplets can be sustained.
So, remember we are talking about vertically upward flows if we are talking about
vertically upward flow, then the gravity is acting in this direction, and the flow is
happening in this direction. So, for a vertically upward flow, this is my bubble and this is
the tail of another bubble. And this liquid is breaking up, when it will break up it will have
a smaller droplets. So, because of buoyancy because of the gravity these drops will start
dropping down, because when that happens one will have a annular flow. And in the
annular flow one will have gas phase.
So, they will because of the gravity these droplets will try to come down. So, they took at
a droplet it will have a gravity acting on it and it will have a drag because the flow is
happening in this direction. So, the flow overall flow is happening in this direction but the
droplets will try to come down. So, this is droplets flow direction. So, the drag on the
droplet will be acting in the opposite direction. So, when the drag is sufficient, that these
droplets can be held there. Then or can overcome the gravity, only then this flow regime
can be sustained.
So, there is a critical Weber numbers what he gave that at a Weber number of 30, this can
happen the droplet weight can be balanced by the drag force. So, that transition to slug
annular slug to annular flow regime happens ok.
56
Now so, based on this what we have learned until now is the flow regimes in large channels
and the different flow regime transition criteria more specifically we have looked at, the
transition criteria from bubbly to Taylor flow, or bubbly to slug flow and from slug to
annular flow.
Now, in this the slug and slug annular have been merged together, or the slug and
associated flows what Taitel map has a slug or channel flow they have been put together
ok. So now, what we just briefly discussed before that, when the transition is from large
channels to small channel, the relative importance of the gravity force and surface tension
force will change. So, when d decrease bond number will decrease, gravity will gravity
effect will be lesser dominant and surface tension effect which is proportional to 1 over d
and this is proportional to d.
So, they will be dominant. So, based on this there are 2 transition criteria that have been
discussed in the literature. So, for example, Bretherton in his classical paper he said that
the Eotvos number; which is same as bond number, and if the Eotvos number is less than
0.84, then a Taylor bubble would no longer rise under the effect of gravity in a water filled
capillary. And this, the for a air water flow the channel diameter corresponding to this
comes as 2.96 mm. So, he has given this as the, say transition from where one can start
looking at that what forces start dominating. So, a Taylor bubble would no longer rise
under the effect of gravity in a water filled capillary. Then the another criteria has been
given by Suo and Griffith in a 1964, where they said that when the Eotvos number is less
than 0.88, very close to what Bretherton has predicted.
The buoyancy on a Taylor bubble moving through a horizontal tube can be neglected. So,
they were looking at a horizontal tube. Whereas, they were looking at the a vertical channel
and the criteria that they have suggested is quite there.
57
Then another work that was done by Brauner and Moalem and Maron and they suggested
I mean what they were looking at the stability and well-posedness of unidirectional
continuity momentum equation in stratified flow in pipes. And they have suggested, that
the stratified flow will be observed, or this the it will be stable only when Eotvos number
is less than 2 pi square which is 17.1.
So, they have not suggested than the stratified flow will not occur, only on the limit of the
stabilities are in this number, that the transition happens. So, one can see from there that
at about few millimeters. specially the one that the limit has been given by Suo and Griffith
and by Bretherton, that when the when these effects or the microchannel effects or the size
effect to small diameter size effects, have started coming into picture is at about few mm.
And they have they are both are saying that about between 2 to 3 mm one can see such
effect for air water flow.
I would like to talk about another effect, or another type of channels, which are not micro
channels. But it is important or they are relevant the flow regime in these channels, or
under these conditions is important, to understand the micro channel. Because there have
been lot of studies on micro gravity effects. So, people have gone into parabolic flight done
the experiments and try to analyze the data.
58
So, one of these studies has been done by Jayawardhane and professor Balakotaiah in the
microgravity conditions where the effect of gravity are not present. And what they observe
generally they are the effect of gravity is not present.
So, the relevant non-dimensional numbers depending on the physical effects, that are
present we have gas liquid Reynolds number and the so, gas inertia and liquid inertia. And
the Weber number or by combining this Reynold number one can have capillary numbers.
So, either of those, and the gas liquid ratio and viscosity and density ratios but these 2
things do not seem to have much effect, because the gas liquid ratio and gas liquid density
and viscosity ratio they do not seem to have much effect on as a variable on the flow
patterns.
So, they correlated the flow pattern in terms of these 2 dimensionless groups. So, if you
go back to our few slides back, and the at the start of this lecture when we talked about
that there are a few effects present the inertial force is viscous forces and the gravity and
surface tension forces. So, in microgravity there is no gravity present. So, one is looking
at the inertial forces, inertia might be of the gas inertia might be of the liquid. So, those
Reynolds numbers, and the Weber number which is the inertia and a surface tension
So, it is taking upon taking into account the viscous forces, inertial forces as well as surface
tension forces. So, based on these dimensionless groups they have tried to correlate things.
59
Now another non-dimensional number using this they have plotted the flow regime map
is called or what is called as Suratman, Suratman number; which is the ratio of Reynolds
number and capillary number or Reynolds number by a Weber number. So, one can look
at say you have to see the relationship of these numbers.
So, Weber number is inertial forces, or inertial effects divided by capillary effects. We can
also write it as surface tension ok. Now if you divide by viscous force effects here. So, one
will have inertial divided by capillary sorry, inertial divided by viscous multiplied by
viscous, divided by capillary, then we will have the Weber number is equal to inertial over
viscous is Reynolds number into viscous or capillary is a capillary number. So, we have
this Weber number is equal to Re into Ca. So, this Sutratman number is Re over Ca if we
substitute that here, then this becomes equal to Re over Ca is equal to Weber number over
Re. So, that will become Re square over Weber number.
So, one can use either of this definition and this is also 1 over Ohnesorge number; which
is also frequently used. And let us see what we get this in terms of properties. So, if we
write Su is equal to Re over Ca 𝜇, sorry Re is ρLULS diameter over μLdivided by μLULSσ.
So, the velocities will be cancelled out. And what one will have is ρLd σ over μLsquare.
And one can see that ρLσ and 𝜇 are μLare properties from the fluids. And d is the channel
diameter. So, this is not a property of the fluid velocity, but the properties of the fluids and
the channel diameter ok.
60
So, they have plotted a flow regime map based on; remember, just to remind you that the
earlier flow regime maps that we have seen is based on gas and liquid superficial velocities.
So, jL and jG that we had for Taitel et al. and then there we had different flow regimes say
bubbles slug annular and there finely dispersed bubble ok. So, what they observed that in
this of course, there is no gravity. So, they did not observe that is stratified flow. They had
bubbly flow, slug flow, annular flow and they also had a what is the transition regimes.
So, sometimes you call it wavy annular or slug annular. So, they had this.
But in this the in this flow regime they have not shown those transition data points or they
have not taken into account those transition data points. So, they have plotted this on the
x axis the ratio of 2 Reynolds number. So, the ratio of a gas Reynolds number and and
liquid Reynolds number, with the Sutratman number, the Sutratman number for the
conditions they have done the experiments, it comes from 10 to the power 4 to 10 to the
power 7. And look at this line which is the transition line between the bubble and slug.
So, this line when they have identified in 2 different regimes. So, you can see here, that
this transition occurs at a particular value of ReGs by ReLs. So, this the slope on this line
that they have found is ReGs by ReLSis equal to 464 point. So, this is some constant and
you can call that or one they have called it as k 1. And then by looking at the experimental
data from this value to be 464 and this is Re over Ca. So, this is Re over ca power minus
61
2 by 3 Ca power 2 by 3 is an important factor when we look at Taylor bubble. So, I would
like to remember it that that this power is 2 by 3.
So, this and the slope of this line is given that. For, all the range of Sutratman number
whatever they have studied from 104 and 107. The transition between bubble and slug
happen. Or it can be given by this line for slug annular transition.
So, if I go back and look at this slug annular transition there are 2 different behaviors at
about 106 , you have a similar line which is parallel to this and you can keep this transition
by a similar line.
However, for the other case, which is 106 when the Sutratman number is more than 106,
there is a different this does not if we extend this line. This does not really define the
transition between slug and annular flow regimes. So, they have looked at this more closely
and plotted this graph now between Sutratman number but on the x axis only the gas
superficial velocity. So, this dramatic change had happens at Sutratman number of about
106. And then again, they have been able to plot a line which gives that Reynolds number
is proportional to Su square ok.
So, one might see that this number is very significant, but then one have to look at sorry
this is 106. And one get this value I think this is 4641.6, 10 times of that; and this is ReGs
is equal to 2 into 10-9 Su squared. So now, this Su is 106. So, this numbers is going to be
62
significant 1012 or more. So, this will be 2 into 103 or more ok. So, this dramatic change
happens at about Sutratman number 10 to power 6 for slug and annular transition.
Now, change in the transition boundary line what happens that at 106 Sutratman number a
change in the transition line as well as the coordinate ok. So, this was the transition criteria
for a microgravity regime. So, what we have looked at in summary today.
Gas liquid flow regimes in macrochannels under 2 conditions, one is normal gravity, and
in the normal gravity we have looked at channel orientation is vertical and when the
channel orientation is horizontal. The other one is microgravity.
So, one can look at the flow regimes under these 2 conditions. Now we also looked at that
in micro channels the bond number start decreasing as channel or in not what we should
say that. the bond number starts decreasing d as d decreases; that means, gravity not
important. So, the lessons that we have learned for microgravity that might be also
important for gas liquid flow in microchannels.
We have looked at the flow regime, but we have also looked at the transition lines. So,
what we have looked at bubbly to slug transition and slug to annular transition. and I
believe that these 2 transitions are most important, because the slug flow is the most
important flow regime in which people want to work in a microchannels. So, in order to
maintain slug flow regime, one need to know the both the boundaries; when it can go to
63
bubbly flow, and when it can go to annular flow. So, in the next class, we will look at in
detail the flow regimes in microchannel ok.
Thank you.
64
Lecture – 01
Gas-Liquid Flow: Flow Regimes
So, in this lecture we will talk about flow regimes, what we were discussing in the previous
lecture. But now we will come to the flow regimes in small diameter channels.
So, briefly we looked at the Eötvös number or bond number criteria, that can be used to
define that what defines or what is the critical bond number at which one can start looking
at that ok these are the channels; where you can expect the micro size effects; starting to
show up in. There were three different criteria that we discussed in the previous class.
So, the experimental studies all they have shown that in a micro channels because the
gravity does not have much effect. So, for channel size less than 1 mm about less than 1
mm for air water flow. For different gas liquid systems, this number may change that about
1 mm or 2 mm, the channel orientation does not have any effect. So, if we look at the bond
number so, which gives us an idea about the things, which is
∆𝜌𝑔𝑑 2
𝜎
65
So, if we take the density of water to be 1000, and the density of air is almost negligible.
So, we keep 10 to the power 3 kg per meter cube as ∆ 𝜌. Let us take g for convenience 10
meter per second squared. And d if we take 1 mm so, that becomes 10 to the power minus
6-meter square, divided by sigma which is 0.072 Newton per meter. So, if we cancel this
out we will get 10 raised to the power minus 2 here. So, that will be 0.01 divided by 0.072
or about 1 by 7.2.
So, the buoyancy force is about one seventh of the capillary forces or surface tension force
for a channel of 1 mm. And as you reduce the size of the channel, this ratio is going to
decrease further and further. So, with that in mind let us look at the flow regimes. The
major flow regimes that have been observed in the micro channels are similar to what we
have observed or what is observed in vertical flow in conventional channel. So, one have
bubbly flow, slug flow, churn flow. annular flow and slug annular flow.
But if we look at the literature, which has been there for quite few decades, at least 2
decades there have been lot of publications on flow regimes in micro channels and few
reviews in the last decade. So, it seems that there is a lot of confusion in the nomenclature
of the flow regimes. Specially the transition flow regime, everybody will agree that there
is a bubbly flow regime, there is a slug flow regime, and there is a annular flow regime.
But what happens in between these the transition flow regimes might be given different
names. So, we will look at some of the representative literature and see.
66
So, let us define the flow regimes one by one. So, the bubbly flow regime, I have put 3
pictures from 3 different publications or 2 are from the same publication actually. So, this
is the bubbly flow in 1 mm channel, in 100-micron channel and 25-micron channel all for
air water flow.
So, bubbly flow is characterized by distinct bubbles. So, one can see there are bubbles,
and they are significantly smaller from the channel diameter at least in the first 2 images.
In the second image where we see the bubbles are of small size, and then these bubbles
are almost spherical. Because at this 100 micron, the bond number will be 1 by 7.2 into 10
to the power minus 2.
So, the gravity will not be important in surface tension will be dominating, and you see all
these bubbles to be spherical. In 1 mm channel one see that there are bubbles, but these
bubbles might be not necessarily of a spherical shape. Some of them which are small they
might be of spherical shape, but it is not clear from the image. But larger bubble clearly
are not of spherical shape that might be because of the higher velocity.
The third one, where we have a larger bubble which is almost of the size of the channel
and the smaller bubbles. So, this we can treat as the boundary between the Taylor flow
regime and bubbly flow regime. So, the bubbly flow regime can be the flow regime, where
the bubble size grows that it is spherical, but starts occupying almost the entire channel so,
this is where we can say the transition might happen. But the general picture of bubbly
flow is something like this or this where the bubbles are smaller. and they are sufficiently
smaller than the channel size. The shape might be spherical or non-spherical depending on
the flow conditions ok.
67
So, the next flow regime is slug flow regime that occurs at low gas and liquid velocities.
The bubbly flow regime might occur, at higher gas velocities and sorry, at higher liquid
velocity and low gas velocities. So, the slug flow regime as you might see here they are
characterized by the flow of bubbles of the size of the channel.
And these bubbles are separated by a thin film surrounding the channel wall, again this is
in 1 mm channel from Triplett et al. and this is from 100-micron channel and 25 micron.
and the flow direction in this case is in this direction in here you have the flow in this
probably happening in this direction and in this case again that is not very clear from here.
So, anyway so, this is 25-micron channel, and this is 100 micron channels the bubbles they
are also known as Taylor bubble, bubble train flow, slug flow and this is a often used flow
regime in microfluidic applications. So, we will look at it in detail later.
68
The next flow regime is what we call churn flow regime, and churn flow regime is a
transition regime. It occurs at relatively higher liquid velocities, and intermediate gas
velocity. So, these 2 pictures which are from Triplett et al. in a 1 mm channel. They have
defined the churn flow for two different processes. The first one you have elongated bubble
so, something like this. And these bubbles start becoming unstable, because when the gas
flow rate is increased and the tails of the bubbles are disrupted by the bubbles dispersed
into ok it.
So, in one case we have the bubbles long Taylor bubbles, and then the gas flow becomes
higher and the Taylor disrupted. The other cases we have churning waves that periodically
disrupted otherwise wavy-annular flows. So, you have waves or annular flow where we
have waves, and on this there are some more bubbles or droplets. So, the churning waves
so, this is called churning wave.
Now, I have also kept in this category, what is called frothy annular and transition flow
regime. So, this is a transition flow regime where we have wavy annular flow, and in
between there are some bubbles there. Similarly, in frothy annular there is annular flow,
but there are some bubbles in between what we see, and these 2 images are from Serizawa
et al in a 100-micron channel. And no such thing was observed in 25-micron channel.
69
Further on slug annular flow so, I will also keep slug annular flow in the category of
transition flow regimes between slug flow or Taylor flow and the annular flow regime. So,
it can be characterized by 2 processes; it occurs at low liquid velocities, and intermediate
gas velocities.
And one can have the slug flow is; because it is periodic slug flow is periodic, and when
it starts becoming non-periodic. For example, you have lot of bubbles coming together
there is almost no slug between them and this is; sometimes called as slug annular flow
people have given a different name for example, Serizawa et al. has called these flow
regimes as transition flow regime or a skewed flow regime.
So, I would suggest the slug flow regime, a periodic or non-periodic, slug flow or Taylor
bubbles in contact. And the other one is just after that which is wavy annular flow. So,
flow one have the interface between 2 bubbles have broken and there is a continuous core.
But there are waves on it for example, this picture, and this is termed as liquid ring flow
by Serizawa et al.
So, they have this liquid ring slow observed in a 100-micron channel as well as in a 25
mm, 25-micron size channel. They also have a liquid lump flow; which does not seem to
be axisymmetric and there are lump of liquid in between. So, I have also kept this in the
slug annular flow regime, but it might also well be put in the category or churn flow
regime.
70
Then finally, we have annular flow regimes of and the gas velocity is increased further.
One get the bubbles have merged the waves have subsided, and there is a continuous gas
code, and there is a thin liquid film on the wall. So, this occurs at low liquid velocity and
very high gas velocities.
So, one have a central gas core in when the liquid flow rate is relatively small and the gas
velocity is high, and the channel is hydrophobic, then one can also have a rivulet flow. So,
I have kept this in this category, and this these are the few images from Serizawa et al.
Triplett et al., and this is for rivulet flow regime any 100-micron channel by Serizawa et
al.
71
So, this picture gives an idea about the flow regime distribution based on the liquid, and
gas superficial velocities. And where what kind of force which are dominated so, Shao et
al. sao and gavalidis they have identified or specified 3 regimes. One at low gas velocities,
where the surface tension is dominated and one get bubbles. And the other one where the
inertia is dominated, and one gets annular flow regime. The dispersed flow regime even
though they have put put the suggest at the dispersed flow regime has rarely been observed
in micro channels and in between these flow regimes, there are churn and Taylor annular
flow regime.
So, basically what we can say that broadly speaking there are 3 well agreed flow regime
that are observed in micro channels bubbly, Taylor and annular. And there are transition
regime between bubbly and Taylor flow regime, so that transition will come by as we can
see here by it can be because of increase in gas velocity, or it can be because of increase
in the liquid velocity. And there is a transition between Taylor and annular flow regime,
which is basically with increase in gas velocities.
72
So now let us look at some effect of some of the parameters, and what are the studies that
have been done in the literature, and what are the general conclusions that we can make
about the effect of some of the flow conditions, or the geometric conditions on the flow
regime.
So, the first one is up their orientation. So, basically orientation will affect the flow regime
map if there is an effect of gravity. So, we said at the start of when we discussed the started
discussing the flow regimes in micro channel. That there does not seem to be any effect of
gravity.
And this study by Hassan et al, where they have plotted flow regime map, for channels
size ranging from 0.1 to 1 mm and they have plot the boundary, they have plotted the
boundaries for horizontal which are given by capital, and boundaries for the vertical
channel which are from the smaller, small letter.
Now, bubbly flow, if you see, the bubbly flow is extended to lower liquid velocities. You
can see the liquid velocities this is for the horizontal channel and this is for the vertical
channel. So, what we can see that the bubbly flow occurs at lower liquid velocities in
vertical channel. similarly, the annular flow it is shifted towards lower gas velocity in
vertical channels.
73
So, for this transition, what they have suggested that it might be because when the bubble
breakup happens, then in the vertical channel the bubble breakup process is also assisted
by the gravity force along with inertia. So, in vertical flow the bubble breakup might
happen earlier so, that is why, this is at lower liquid velocity.
Similarly, in the annular flow the annular flow regime the gas core it is assisted by the
inertia because of the buoyancy force or the velocity; that is because of buoyancy, that will
assist or that will increase the gas core velocity. So, the annular flow transition may come
in early.
Now, another effect the effect of channel size. So, as we have seen that the bond number
it’s value decreases as the channel size decrease. And there will be a significant difference
between the bond number at 1 mm and bond number at 10 mm to be precise it will be
about 10-4. So, the 4 order of magnitude difference at 2 different, 10 microns, 10-micron
size, and 1 mm or 1000-micron size, there will be very large difference in the bond number.
So, as the channel size decreased Taylor flow is observed for much higher gas and liquid
superficial velocities of the area occupied by the Taylor flow is large. This one of the
explanation for this might be that the bubbly flow regime shrinks. Because even if a
smaller size bubble comes in the channel, it occupies the channel cross sectional area and
then because the channel cross sectional area is small. So, it has to expand and become a
Taylor bubble.
74
So, the transition to Taylor bubble comes in early as the channel size decreases. the bubbles
of the smaller volume will yeah, and at the smaller size the surface tension will be high;
however, there is no change in transition to inertia so, there is no transition no change in
transition to annular flow, Taylor flow or annular Taylor flow or slug annular flow, to
annular flow there is not much change in the transition in that; now channel cross section.
So, microfluidics, if you look at the number of applications that microfluidics has been
used. The channel is not necessarily a circular channel or a cylindrical channel as a
generally the case in the conventional channels. The channel is often a rectangular or
semicircular or sometimes even triangular channel or a square size channel. So, in all these
non-circular channels, they have sharp corners. And the liquid near the corners that stays
there and the liquid goes there and stays there it remains at stagnant.
So, the liquid tends to be pushed to corners, in these non-circular channels, and the area
occupied by the gas. So now, the gas has less area available in the channel to occupy. So,
the gas velocity because the gas; the area for which is available for the gas for example, in
annular flow, the gas velocity becomes higher. So, this results in higher gas velocity in
non-circular channels.
So, this results in that the transition to annular flow it occurs at lower gas velocity. There
is not much change in the slug or slug annular or in the case of bubbly to slug transition.
But what has been observed that the annular flow occurs early probably because the
75
velocities for the same flow rate and the same channel size are larger the actual gas
velocities are larger in the noncircular channels.
Now another point that is very important in micro channels that the properties of the
channel walls. The channel walls it might be hydrophilic or hydrophobic. So, hydrophilic
means it likes the water and hydrophobic means it does not like the water. So, that means,
there is in the hydrophilicity and hydrophobicity that comes into picture only when there
is a 3-phase contact line. So, on one side you have liquid another side you have gas so,
because and that that will depend on the contact angle.
So, generally when it is hydrophilic, then contact angle 𝜃 is small and for hydrophobic
channels θ is large. Few studies have been done on the wall waiting or hydrophilicity and
hydrophobicity or wall contact angle properties on the flow regimes. Notable among them
is by Barajas and Panton, Lee and Lee and the one by cubaud.
So, in the hydrophobic channel as we can see that the liquid does not always lubricate the
wall well. So, there is a 3-phase contact line here whereas, in hydrophilic channel we see
there is bubbly flow, and the bubbles which are there they are smooth bubbles, and then
one has the liquid Taylor annular flow and slug flow what they have observed. And this is
done in a rectangular channel, which is 525-micron size. The channel is silicon channel
with a glass top the silicon have a contact angle of 9 degree and the glass top have and
then you know of contact angle of 25 degree.
76
The other one the other study for hydrophobic channels has been done in the same kind of
same channel, but with the Teflon coating which has a contact angle of 120 degree and
one can see the difference in the flow regimes that one get in the channels. So, one thing
is, that once is when the channel is hydrophobic then one see a lot of 3 phase contact lines
forming.
So, for example, one gets rivulet flow; rivulet flow means, that on the channel wall if we
are looking at the cross section. Annular flow will have a continuous liquid film which is
represented by a hatched area here; this is annular flow normal to screen in this case rivulet
flow, one can see that there are reverse small rivulets. But there is some dry area as well;
so, this is a rivulet flow.
A very common example of rivulet flow, if you are sitting on a bus and it is raining and
the bus is moving with a very high speed. You might see some of the liquid on the window
screen, you might see that the liquid is going in the moving in this direction, right? So, it
is not a continuous film, but they are small river kind of thing which are observed. So, that
that is similar to rivulet flow.
So, in general the hydrophobic or large contact angles they favors the distortion of the gas
liquid interface because the interface the liquid does not like to be near the wall and the
interface so that there are all the chances of formation of a 3-phased contact line, you have
gas liquid and solid touching each other. And this also promotes breakage of liquids and
gas into bubbles and droplets.
77
So, in summary what we have looked at is the flow regimes in micro channels, and we
have 3 main flow regimes; what we call bubbly, slug or Taylor and annular flow regime.
We also have a transition flow regime between slug and annular flow which we call slug
annular, it can have wavy annular or it is also known as liquid ring flow. Taylor bubbles
touching. Apart from that we also have churn flow; which is basically when you have
higher liquid velocity in the regime where you have intermediate gas velocities, and higher
liquid velocities, then one can have churn flow regimes. and this is again another transition
flow regime.
And we have also looked at the effect of some of the parameters; such as the channel cross
section, channel, hydrophobicity and hydrophilicity, the orientation of the channel. So, we
will stop here and in the next class this is bubbly and in the next lecture, we will look at
another flow regime map for the which has been come from the microgravity flow regime
map, and how it can be adopted or how it has been adopted for the micro channels. And
then we will look at the different flow regimes ok.
Thank you.
78
Lecture - 02
Tylor Flow 1
Hello in today’s lecture, we will be looking at Taylor or slug flow regime in particular. So,
first of all let us look at the characteristics of Taylor flow.
Taylor flow or slug flow regime is characterized by the regular and periodic flow as I have
written here regular and periodic flow of bubbles which are of the size of the channel. The
picture shown here is for a vertically apart flow, and the bubble is in the black color, and
you can see that there is a film on both sides of the bubble which separates it from the wall.
So, there is a thin film that separates it from the wall. And the flow of bubble is periodic.
So, you have a number of bubbles. And that is why it is also called a train bubbles or
bubble train flow. And in between 2 consecutive bubbles so, if we look at the estimate a
schematic for a bubble train flow we will have something like this a picture something like
this.
That one have bubbles moving with velocity, let us say UB is the bubble velocity, and this
bubble is moving with a velocity UB. And in between the bubble we have what we call
liquid slug ok because these slugs are trapped between 2 bubbles. So, they will also move
79
with a velocity which is equal to UB, because if they do not, then eventually what will
happen that the 2 consecutive bubbles they with come close and eventually may coalesce
or if the slug. or if or the distance between them keep increasing which does not happen
and if that that happened and if flow is not periodic. So, the slug velocity is equal to the
velocity of the bubbles.
In micro channels, where the gravity effect is not dominant; so, we the bubble shape is
generally axisymmetric ok.
So, before we go on to do anything else, let us look at some of the terminologies, that
might be required to understand Taylor flow. So, superficial velocity, it can be the
superficial velocity of the gas phase, superficial velocity of the liquid phase. And it is
basically US is equal to Q over area of cross section of the channel. now if we are looking
at gas then it is USG or superficial velocity of gas is equal to Q of gas, divided by channel
cross section area. Similarly, superficial velocity of liquid will be Q of liquid divided by
the channel cross sectional area ok.
𝑄𝐺 𝑄𝐿
𝑈𝑆𝐺 = ; 𝑈𝑆𝐿 =
𝐴𝐶𝑆 𝐴𝐶𝑆
80
Mixture velocity is, it is also called 2 phase velocity. And the term TP comes from 2 phase.
So, U TP is defined as Q L plus Q G. So, the total flow rate divided by the channel cross
sectional area. So, from this we can see this is also equal to U TP is equal to U, superficial
velocity of gas plus U superficial velocity of liquid.
𝑄𝐿 + 𝑄𝐿
𝑈𝑇𝑃 = = 𝑈𝑆𝐺 + 𝑈𝑆𝐿
𝐴𝐶𝑆
Now, bubble velocity, the bubble will generally be moving faster, then the average velocity
of liquid in the channel. So, the velocity of the bubble we generally term it as or represented
using UB. Then homogeneous volume fraction and void fraction, or gas traction or gas
hold up; there are 2 terms here, which represent the volume fraction of gas. So, the first
term is not necessarily not necessarily the actual gas volume fraction.
So, this term represents the ratio of the gas flow rate to the total flow rate, when it is
assumed or when the gas and liquid are homogeneously mixed. So, that means, β is defined
here as gas flow rate over Q L and plus Q G. But that is not generally the case in Taylor
flow because the bubble move faster. So, the gas velocity will be more than the liquid
velocity. So, this will also change the volume fraction of or the average volume fraction
of the gas in a channel. So, this is there is actual void fraction or the actual volume fraction
is gas is called ε G, or void fraction that is the actual volume fraction that you see in the
fully developed Taylor flow in a channel ok.
81
So, little bit of history. It is almost 80 years ago at least that is there I could find that
Fairbrother and Stubbs they looked at understanding of the velocity of bubbles in the
capillary. The idea was that they thought that a Taylor bubble can be used as an indicator,
to measure the velocity of the liquid in the channel. And eventually they found that the
bubble velocity is more than average liquid velocity in the channel.
Then in 1961, Taylor did the experiments. So, for a range of capillary numbers. So, to
𝜇𝑈
remind you the capillary number is equal to . So, σ is surface tension between gas and
𝜎
liquid phases. Μ is the liquid viscosity. And U should be UB bubble velocity, but because
in many cases, we might not know the bubble velocity beforehand. And there is not much
difference at least I mean the difference between the bubble velocity and the 2-phase
velocity case about 10 to 15 or 20 percent at max.
μ L 𝑈𝑇𝑃
So, this can often be written or this is often written as ok. So, he measured Taylor
𝜎
measured the film thickness for a range of capillary numbers experimentally. And after
him this bubble shape has been known as Taylor bubbles. then in 1961 the same year
Bretherton in his classic paper, he did the theoretical analysis to analyze, the shape of the
bubble calculate the thickness of the films surrounding it and from their find out the bubble
velocity and the pressure drop across the bubble.
Then because this particular configuration that a bubble flowing a capillary, this also can
mimic the flow of red blood cells in capillaries, to remind you that the red blood cells they
are not exactly like bubble, but they have a membrane and they are flexible particle;
however, as a first approximation one could understand as relatively simpler problem of
bubble flow in micro channels. So, they looked to understand with the motivation to
understand the blood flow in capillaries, they looked at Taylor flow. then in 1971 there are
the begg, he used it to form a continuous segmented flow analysis, or for a segmented flow
in the liquid region, and in the in 1973 Hovarth, he looked at radial mixing in the slug
flow, and then there have been a number of studies some of which I have written.
The Ratulowski and chang they looked at the shape of the bubble numerically and Kolb
and Cerro they had a quite a few papers for application in coating inside a capillary. So,
the liquid coating in the capillaries, and when you have the liquid filled with the material
or the capillary filled with the material with which it is to be coated. And then the air is
82
blown through the capillary and the front shape of the meniscus or the interface that forms
that is like in a bubble and that gives you what is the film thickness.
So, they looked at in the problem from this perspective, and in 2,000 Aussilous and Quere.
They looked at the film thickness at large capillary number, extending the Bretherton’s
correlation who developed a correlation for the film thickness. And after that in the past 2
or 2 decades, there have been a number of studies on Taylor flow in micro channel to
obtain film thickness to obtain heat transfer for mass transfer applications, or even simple
validation of the numerical codes, because when one can do the experiments one can
digitize the flow very accurately and it can be used to further validation of a of a say
volume of fluid method where the interface is captured.
So, some of the applications as I said it was earlier described to use for measurement of
liquid velocity. it the concept is still trying to be used or people are trying to develop
multiphase flow meters, based on this concept. then in chemical processing for example,
catalyst coating multiphase micro reactors. If you look at the gas-liquid reactions in
microfluidics or in micro channels or chemical micro processing, then about 80 to 90
percent of the work that you will observe is being done, it might be a gas-liquid flow all
liquid-liquid flow or sometimes even gas-liquid-solid flow most of the work is being done
in the Taylor flow regimes.
83
It is very important to understand the hydrodynamics integral of flow regime in biomedical
applications. The blood flow in capillaries as I said this can mimic the same problem of
RBC flow in capillary or in micro capillaries.
Then the lung airway opening problems. So, in the lungs there is mucous surrounding the
walls of the alveoli, and then in between those there is sometimes the mucus forms a liquid
plug. And this interface looks like the Taylor bubble. So, for example, professor grotberg,
his group has looked at a number of effects on say the characteristics of this structure, and
of what force is required to break this liquid plug, because it is not it can be harmful for
the for the kids in which this problem generally occur.
Then in last 2 decades, the gas sparging has been used for removal of fouling in the
membrane. and it has been shown that this fouling is most efficient if the gas-liquid flow
occurs in the Taylor flow regime. So, this enhanced the micro filtration efficiency in micro
filtration Nano filtration.
So, in the membrane fouling removal then in the electronics cooling we are one want to
have the advantage of the high thermal or heat capacity of liquids. And further enhance
the heat transfer. It has been it is being used and its application is being explored in
electronics cooling, and heat exchangers. in fuel cells one have gas-liquid flow. So, one
have this flow regime, and in the oil and gas industry, where the channels might not be the
straight channel. But there is flow of oil and gas and the displacement of one fluid by
another there one will observe Taylor flow regime might be in gas-liquid or liquid-liquid
flow.
84
So, it has a number of applications. Now let us look at some of the physics of it. So, what
we have drawn here is a channel in which the gas is coming from one inlet, let us call it
inlet 1, and the liquid comes from another inlet inlet 2. And we consider a control volume
that is represented by the dotted line and outlet of this control volume or the the other end
of this control volume. We take an arbitrary cross section in the channel A A. In this figure
we have taken it where both the phases are present, but one can take it there it is only liquid
phase.
So, let us look at and write the mass conservation equation here. One of the assumptions
that we make here is that the flow is incompressible. so, that means, ρG and ρL are constants
ok mass conservation we apply then we can write mass in the control volume minus mass
out of the control volume, and that will be equal to mass accumulation. We are not
considering any reaction. And so, no generation or some symptoms will be there.
So, if we look at the gas phase, then if the gas flow rate is QG1 and the inlet. So, we write
it as QG1, and then this density of gas is ρG. Minus, let us say the gas flow rate at A A. So,
Q is volumetric flow rate. So, QG1ρ G Minus QGA A ρG is equal to d by d t of gas volume
in the control volume and V represented by V gas in the control volume into ρ G ok. So,
the control volume V G or V C V is the total volume of the control volume. And V G V is
the volume of gas in the control volume.
Gas: -
85
𝑑(𝑉𝐺,𝐶𝑉 𝜌𝐺 )
𝑄𝐺𝑖𝑛 𝜌𝐺 − 𝑄𝐺𝐴𝐴 𝜌𝐺 = ……………………………(𝐴)
𝑑𝑡
Liquid: -
𝑑(𝑉𝐿,𝐶𝑉 𝜌𝐿 )
𝑄𝐿𝑖𝑛 𝜌𝐿 − 𝑄𝐿𝐴𝐴 𝜌𝐿 = ………………………………(B)
𝑑𝑡
……………………………………………………………………………….
𝑑
(𝑄𝐺 + 𝑄𝐿 )𝑖𝑛 + (𝑄𝐺 + 𝑄𝐿 )𝐴𝐴 = (𝑉 + 𝑉𝐿,𝐶𝑉 )
𝑑𝑡 𝐺,𝐶𝑉
Similarly, if we write down the same equation for liquid, then we have Q L1 into ρ L
Minus Q L at A A into ρ L is equal to d by d t, the volume of liquid in the control volume
into ρ L. Let us say this is equation A and this is equation B, and this we add them, but
after dividing them by density. So, what we do is A by ρ G Equation A divided by ρ G.
Plus, equation B divided by ρ L.
You can see that ρ G ρ G we cancel out from there um anyway. So, if you do that then
what we get is Q G 1 plus Q L1. All this is at the inlet. So, we can actually remove the
one, and that understanding that can be at the inlet minus Q G plus. So, actually we should
have written this earlier itself as at inlet ok, minus Q Q G plus Q L at cross section A A,
that will be equal to d by d t of gas volume in the control volume plus liquid volume in the
control volume ok.
86
𝑉𝐺,𝐶𝑉 + 𝑉𝐿,𝐶𝑉 = 𝑉𝐶𝑉 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑑
(𝑄𝐺 + 𝑄𝐿 )𝑖𝑛 − (𝑄𝐺 + 𝑄𝐿 )𝐴𝐴 = (𝑉 ) = 0
𝑑𝑡 𝐶𝑉
(𝑄𝐺 + 𝑄𝐿 )𝑖𝑛 = (𝑄𝐺 + 𝑄𝐿 )𝐴𝐴
𝑈𝑆𝐺 + 𝑈𝑆𝐿 = 𝑈𝑇𝑃 |𝐴𝐴
So, what we end up with is this equation that Q G plus Q L at the inlet minus Q G plus Q
L at cross section A A is equal to d by d t of volume of control volume. Remember, that
volume of gas in the control volume plus volume of liquid in the control volume, because
there are only 2 phases. So, that will be equal to volume of control volume. And volume
of control volume as our control volume is fixed. So, that is a constant it does not vary
with time. So, this term is 0. So, that means, that Q G plus Q L at the inlet is equal to Q G
plus Q L at any cross-section A A.
Now, if we divide by the channel cross section A here. And the A cross section A cross
section. So, this is U superficial velocity of gas plus, U superficial velocity of liquid. And
we had defined this Q G plus Q L over channel cross section. This is equal to U TP, or
mixture velocity at A A. So, this gives a important result; that at any place in the channel
the 2-phase velocity is equal to the gas and liquid velocities. In the assumption that we
have made in this derivation is that the flow is incompressible.
87
So, this is valid not only for the capillary or the slug flow or the Taylor flow, but for any
other flow also. So, if you have a flow in a channel, then if you measure the average
velocity at this cross section; this is equal to U TP. You major the average velocity at this
cross section this is also U TP ok.
So, this is an important result that can be quite handy and useful. as I said, that the Taylor
flow is periodic in nature. So, that means, it if you are doing the experiments and you are
in the laboratory or a stationary frame of reference. So, then you will observe if you have
a channel, and if you are looking at this point.
Then you will observe and what I am going to plot is let us say with time the volume
fraction. So, one will observe that the volume fraction keeps changing. So, and it is
periodic. So, one bubble comes and then another bubble comes then another bubble comes.
So, the flow is unsteady of course, in the bubble frame of difference flow is unsteady,
periodic in time at any time instant this is also periodic in space. So, if you take any
snapshot of Taylor flow in a channel, then you will have the bubble shapes are supposed
to be same in periodic pardon me for my poor drawing skills. So, there is periodic flow of
bubbles and this is also periodic in space.
Now, what if we are sitting on the bubble? So, if we change the frame of reference to the
bubble frame of reference, then the first thing will be that you will have bubble stationary.
So, when the bubble is stationary, or then what will happen that? You are looking at a
88
different picture. In this case you have a velocity if the slug is long enough maybe the
velocity profile is almost parabolic if not then, you will have some gradients in the liquid.
But all in the positive direction that is what it has been shown here
But if you are in the bubble frame of difference and the bubble is stationary. So, flow is in
this direction in the earlier case, but now the bubble is stationary, let us say there is no.
And before we describe the periodic flow let us say, because the flow is periodic. So, we
often take one-unit cell, and this unit cell is one bubble and one slug, it can be any choice
you can have the boundaries like this, or for convenience one can have the boundaries, that
the bubble is in the middle and one are from the slug in the back and one are from the slug
at the front.
So, in the bubble frame of reference UB is equal to 0. So, that means, you will have the
walls, the actual flow is in this direction, or in the laboratory frame of reference the flow
is in this direction. So, when you are sitting on the bubble you would see, the walls moving
with minus UB velocity in the opposite direction. And as a consequence, you will also see
the velocity profile to be something like this. Bubble is stationary, wall move or in opposite
direction with bubble velocity, and for a unit cell, which is like this; that we have unit cell
which has a bubble plus 2 halves of adjacent slug only liquid phase is present. and it has
half in half the channel when it half the cross section it is coming in and half the channel
is going out ok.
So, if you have if you have noticed in the previous slides, there was another statement, that
one can see internal recirculations in the liquid slug. So, that internal recirculation are
basically this which are which can be seen in the bubble frame of reference. So, it is
sometimes quite useful to look at or study the bubble flow or the Taylor flow, analyze it
in the bubble frame of difference, because the bubble becomes a stationary and then we
can analyze number of things on this along this stationary bubble.
89
So, we had defined 2 void fractions 𝛽 and 𝜀𝐺 . And generally, what we have at our disposal
is flow rate of gas, and flow rate of liquid from the experiment. so, that means, that we
have U superficial velocity of the gas, and U superficial velocity of liquid. And that gives
us U TP; which is equal to U superficial velocity of gas plus U superficial velocity of
liquid. But if you want to find out bubble velocity that is not known a priory. we can also
find out β very easily from the flow rates. This can be the inputs. So, β is equal to Q G
over Q L plus Q G, or we can also write this sorry Q G over Q L plus Q G, and this is U
superficial velocity of gas over U TP.
𝑄𝐺 𝑈𝑆𝐺
𝛽= =
𝑄𝐿 + 𝑄𝐺 𝑈𝑇𝑃
𝑄𝐺
= 𝑈𝑆𝐺 = 𝑈𝑇𝑃 𝛽 = 𝑈𝐵 𝜀𝐺
𝐴𝐶𝑆
𝑈𝑇𝑃 𝛽
𝜀𝐺 =
𝑈𝐵
Now, we can have a relationship from the mass conservation of gas, that U TP into β;
which is also equal to U superficial velocity of gas or Q G or channel cross sectional area;
that is equal to UB into ε G. So, if we are able to measure the bubble velocity. So, after UB
is measured, for example, visually then one can find what is ε G. So, ε G is equal to U TP
90
β over UB. So, this is another important relationship, which can be used for analysis of
Taylor flow.
𝑈𝑇𝑃 𝜋𝑅 2 = 𝑈𝐵 𝜋(𝑅 − 𝑏)2 + 𝑈𝐹 (𝜋𝑅 2 − 𝜋(𝑅 − 𝑏)2 )
𝑈𝑇𝑃 1 𝑅−𝑏 𝑏 𝑈𝑇𝑃

( )2 = →1− = √
𝑈𝐵 𝑅 𝑅 𝑈𝐵
𝑏 𝑈𝑇𝑃
=1− √
𝑅 𝑈𝐵
Now, moving on further, if we look at the bubble shape, the bubble has a front and the tail,
and this is a constant thickness film region. So, bubble shape is almost cylindrical here.
And this film region is let us say the film thickness is B. So, b is the film thickness. If we
apply mass balance then we can write that the also assume that not mass balance, let us
say. We have just found that average velocity equal to U TP. So, if we use that relationship
here. And let us say the channel radius is R, then we have U TP into pi R squared is equal
to U of bubble into pi R Minus B squared, plus U F which is flow in the film into pi R
square minus pi R Minus B squared.
Now, film is generally film has negligible flow. When we talk about the film this is, I am
talking about the constant thickness film region. So, that implies that U F can be neglected.
91
So, this term goes away. Then what we have is pi and pi also cancelling out. So, we have
U TP over U V is equal to R Minus B over R square or R min R minus B over R is equal
to U TP by UB raised to the power 1 by 2. So, 1 minus B by R is equal to root of U TP over
UB or B by R is equal to 1 minus root of U TP over UB. You might note that the bubble
moves faster than the average velocity in the channel. So, that means, this number is
always going to be less than 1.
So, this film this is the relationship for film thickness. So, we have seen that if we can find
out the bubble velocity, then we can know what is the film thickness. If we find out the
bubble velocity, then we can also know what is ε G, or in among B Ε G and UB. Out of
these 3 if we can find one parameter, then we can further analyze and find other 2
parameters by the relationships that we have just now developed ok.
So, let us look at a bit in the flow field in the liquid slug.
So, when we have flow field in the liquid slug, and I briefly talked about that we have the
slug, the velocity profiles in the slug also we have plotted. So, let us plot few bubbles here.
So, if we look at a velocity field in the slug. it will look something like this now the
question comes or people are often confused that it has been written in the literature that
it has internal recirculation with them.
92
So, the internal recirculation is so, this can be decomposed actually, that it can be the slug
velocity the velocity profile that one see is; slug velocity plus internal recirculation
velocity; that means, the velocity in this because the slug on an average, it has to move
with the bubble velocity in the front direction ok.
So, it’s total velocities such that that it move with the bubble velocity. If we look at a slug,
let us say, if we have a slug something like this. Then the picture in the slug will be
something like this. So, there is internal recirculation and that can be seen in bubble frame
of reference. So, you see that, and in along this there is also a small film surrounding the
slugs.
So, this film may or may not be of the same thickness; that is, the film surrounding the
bubbles. So, what we have is that the slug few things that we can note about the slug that
the slug move with bubble velocity. Internal recirculations can be observed in bubble
frame of reference. And the slug velocity is the sum of to the internally circulations as well
as the slug velocity.
Now, next we will talk about Bretherton’s problem. So, in 1961, in his classical JFM paper
journal of fluid mechanics paper Bretherton has analyzed the flow of a Taylor bubble in a
capillary. And what he was able to do that using lubrication approximation he analyzed
the flow behavior around the bubble, and used young Laplace equation for the pressure
jump across the 2 phases near the interface.
93
And from those 2 equations, he was able to find out the bubble shape. And more useful
from that he could find out the film thickness. And we as we have seen that film thickness,
we can also find out bubble velocity, and the volume fraction and he could also find out
the of course, the pressure distribution and the total pressure drop across the bubble ok.
So, we will look at this problem we will it in detail now.
So, first thing he looked at, he considered a long Taylor bubble, and by that what he meant
is long cylindrical region in the middle. If you look at the shape of the Taylor bubble, you
can fit the front part which is part of the circle and that comes because of a young Laplace
equation at the front. And at the back at very low velocities this might be almost same as
at the front, but a high velocity you will see that the bubble tail becomes flatter and flatter.
So, at the back once you will have the back of the bubble also you can fit part of a sphere
and in between the film thickness. So, what he was able to do or what he said or he
suggested that we can analyze the 2 parts of the bubble front and back independently.
Now he considered the front. So, for convenience, we will analyze the front part and then
use the result for the analysis of the back part. So, if we considered the front of the bubble,
then we will look at 3 regions in the bubble for example. So, one is the front; which is a
spherical. Then this region I have not been able to draw the films to be same, but they have
same here.
94
We are considering axisymmetric bubble. So, the constant thickness film region. So, this
is they say this is 1 2 and let us say 1 2 and 3. So, this is 1, this is 2, and the main region
of interest is 3. see, as we have said that the flow in the film is negligible. flow and it is
cylindrical in nature. So, we know that if we know, the film thickness then we know the
shape of the bubble front it is spherical. So, we need to find what we need to find with the
radius of with this is sphere.
Now, we look at this intermediate region or intermediate film; which is between the 2. We
also consider this in bubble frame of reference. So, when we are looking at the problem in
bubbles frame of reference, then the wall the bubble will be a stationary, and the wall will
be moving with a velocity minus U. So, that is why you see these uniformed vectors in the
constant thickness film region the channel radius is R the film thickness is b and the
thickness of the film in the intermediate region in this region, it is considered to be h and
it is varying along the axial direction. So, axial direction is x, and considering the origin at
the wall, we have x and y the coordinates near the wall. And this the coordinates are in the
film region.
Now, we can apply lubrication approximation. So, lubrication approximation is just

scaling, that if one of the approximation say it is because the problem is axisymmetric.
So, if we look at A cross section in the bubble, we have this channel wall, and this is bubble
interface wall bubble interface and this is channel cross section in the bubble region ok.
95
So, the film thickness is b, and the channel radius is R, or if we are in the intermediate
region then this channel thickness is h in the channel radius size R. So, then h is very, very
less than R, then for flow in the thin film we can use planar coordinate system. Also
remember that when we have looked at the problem in the bubble frame of reference then
we have been able to convert it to a steady problem. The fluid in the film the coordinates
are planar and now we look at the some of the further let us look at the pressure jump at
different points.
So, if we assume the pressure just at the front of the bubble in the liquid region at the
pressure at this point is 0, at the front of the bubble in liquid, if the pressure at this point
let us call this point as 1, at this point the pressure is 0. Then by young Laplace equation,
we have a sphere will have a radius of curvature or the sphere will have a radius R. And
the pressure difference across the sphere will be 2 σ by R. So, if the radius of curvature at
the front is R F, then we have pressure just near the bubble front, but in the bubble, it is 2
σ over R F.
Now, note that pressure is almost uniform in the bubble, when you compare the, that in
the liquid region. So, when the pressure uniform in the bubble; so, with respect to this
point we have the pressure difference 2 σ over R F, and this and the bubble front, and that
same pressure will be everywhere in the middle and back of the bubble.
96
So now look at the pressure jump, in the middle of the bubble, and if you analyze the
pressure jump in the middle of the bubble, then you have p in the liquid film in the
cylindrical region. So, this is this is cylindrical portion, and the pressure jump here will be
equal to 2 σ by R F is the pressure inside. And that is equal to the the difference between
the pressures is equal to σ over R minus b.
So, p bubble minus p liquid is equal to σ R minus b. So, that is why we have pressure in
the film region is equal to 2 σ by R F minus σ over R minus b, because the radius this is
cylinder. So, the radius of curvature for the cylinder in one direction it is infinite, and in
other direction normal to it it is radius of cylinder which is R minus b. So, 1 over R 1 plus
1 over R 2 for a cylinder is 1 over R minus b.
So, we have pressure jump and from that the pressure in the liquid constant thickness liquid
film region, 2 σ by R F minus σ R minus b. Similarly, at the back of the bubble, just near
the back we will have 2 σ over R b is the pressure difference, and pressure in the bubble
is to 2 σ over R F. So, the pressure difference is to σ over R F minus 2 σ over R b. So, that
will also be the pressure difference there delta p b
So, this is the pressure distribution at different points in the bubble and some of it might
be used in later.
97
So, using the lubrication approximation equations, as we said that we can analyze this
problem using the planer coordinate system, here I have written the Navier stokes equation
for an incompressible Newtonian fluid, and that steady state, this is continuity, and this is
of course, in 2D and this is momentum x momentum another one is y momentum.
So, what we want to look at is we want analyze is the film in the intermediate region,
where it is the film thickness is h and this direction is L. So, h is such that h is very, very
less than R, but it is bigger than B. h will also be smaller than the L, where L is the length
along the axial direction. So, this is x direction this is y direction.
Now, we look at the continuity equation and an order of magnitude analysis. So, the V x
over x which is L, the comparable with V y over h, and if we write that then V y is equal
to h by L V x, when h is smaller than L, then V y can be negligible with respect to V x.
So, we will not consider. So, we will consider or neglect the V y component and this. So,
if that is the case, we will also have the flow to be fully developed in such case. so, that
means, that del V x over del x will also be approximately 0. this stage from we have got
from continuity equation.
Now, let us look at y momentum equation. So, this term will be 0 because V y is 0 all this
is going to be 0, the only thing that we get from here is del p over del y is equal to 0; that
means, p is not a function of y or only a function of x ok. in this equation, this term is 0,
because V y is equal to 0. This term is 0 because del V x over del x is negligible. And this
is also 0, because del V x over del x is equal to 0. So, what we have is del p over del x 1
over ρ is equal to μ over ρ del 2 V x over del y 2. So, we have a simplified.
98
𝑑𝑣𝑥 1 𝑑𝑝
= 𝑦 + 𝐶1
𝑑𝑥 𝜇 𝑑𝑥
1 𝑑𝑝
0= ℎ + 𝐶1
𝜇 𝑑𝑥
……………………………………………………………
𝑑𝑣𝑥 1 𝑑𝑝
= (𝑦 − ℎ)
𝑑𝑥 𝜇 𝑑𝑥
equation μ del 2 V x over del y 2 is equal to del p over del x. Let us integrate this. So, what
we will get is; we can actually write this as total derivative. So, we will right now this has
total derivative, and d V x over d y is equal to 1 over μ d p over d x y plus C 1.
Now, the boundary condition at the interface will be this is gas-liquid interface. So, if it is
gas-liquid interface, then we will have 0 shear boundary condition or what we call free slip
boundary condition; which means that the shear of the gas at the liquid phase is negligible.
So, d V x over d y is almost equal to 0, at the interface. So, the interface is this. So, this is
y is equal to h at y is equal to h. So, we will have d V x over d y y is equal to 0 at y over h
1 over μ d p over d x h plus C 1. So, this gives us, if we subtract the 2 equations, we will
get d V x over d y is equal to 1 over μ d p by d x into y minus h.
99
1 𝑑𝑝 𝑦 2
𝑣𝑥 = ( − ℎ𝑦) + 𝐶2
𝜇 𝑑𝑥 2
Let us integrated further, and apply the boundary condition at the wall. So, we had this d
V x over d y is equal to 1 over μ d p by d x into y minus h. let us just confirm it. 1 over μ
d p by d x y minus h and we can integrate it to find out V x is equal to, 1 over μ d p over
d x y squared over 2 minus h y. plus, another constant c 2. For that we can use the boundary
condition at the wall. So, if we again draw the problem where we at y is equal to 0. and it
is in bubble frame of reference. So, we will have wall moving V w equal to minus U. So,
V at y is equal to 0 V x is equal to minus U. So, we substitute that here, then minus U is
equal to c 2. So, that gives us V x is equal to 1 over μ d p over d x y squared over 2 minus
h y minus U.
100
So, having obtained this relationship. Let us now calculate the flux in the film. So, if we
calculate flux in the film still in the planar coordinate system we will get Q, film is equal
to integral 0 to h V x d y. And remember, this is also equal to. So, when the constant
thickness film region, the film thickness is b. And the velocity is minus U. So, this Q is
equal to minus U b. So, we substitute this here.
So, we got minus Ub is equal to 1 over μ d p over d x, y cube over 6 minus h y square
divided by 2 minus Ub. And the integration limits from 0 to h. on simplification we will
get minus Ub is equal to 1 over, and this is sorry this is minus U into y here. 1 over μ d p
over d x this is h cube over 6 minus 0. So, this will be h cube minus 6 minus h cube over
2 minus U h.
So, if we substitute here, we will get you into h minus b equal to minus 1 over μ d p over
d x into minus h cube over 3. So, what we will get is d p by d x is equal to 3 μ U h minus
b over h cube.
101
So, using the young Laplace equation, we can calculate the pressure difference across the.
So, what this equation is? Pressure jump in the intermediate region. So, in this intermediate
region p bubble minus p, p is the pressure in the intermediate region or the region; where
we are looking at the flow, or where we are analyzing the flow is equal to. So, this will be
σ over R minus b. So, this will look something like this.
So, it has 2 radii of curvature, one is in this direction, and another one is normal to it, which
is in the say cylindrical region. So, this is R minus h. So, this have been approximated
approximated h like b is h is very small. So, this has to be σ over R minus h. But then
because R is very small as we say that R is large than h. So, we can write that p bubble, or
here p bubble is 2 σ over R F minus p equal to σ over R plus σ d 2 h over d x 2 the other
radius of curvature. So, this is Kemrock approximated this radius of curvature can be
approximated as d 2 h by dx 2. And the other radius of curvature normal to it is σ over R
minus h which we have approximated as σ over R.
So, if we differentiate it. We get d p over d x is equal to σ d cube h over d x cube minus
here.
102
𝑑3𝜂
𝑥 1
𝑑( (3𝐶𝑎)3 )3
𝑏
𝑥 1
𝜉= (3𝐶𝑎)3
𝑏
So, if we substitute that in the film region then we will get this equation this is d cube h
over d x cube is equal to 3 μ U into h minus b over h cube. And if we non dimensionalize
it by having 2 non-dimensional coordinate eta and psi. So, h h is approximated as b into
eta. So, let us substitute that here we will at get d cube eta b over d x cube is equal to 3 μ
U over σ is capillary number. So, you have 3 Ca into h minus b. So, we can write this as
eta minus 1 over eta cube.
And here you will have b square. So, if we further write this they will have d cube c eta
over d over b into 3 Ca power 1 by 3 cube is equal to eta minus 1 over the eta cube. So,
what we have done is approximated this this is x over b. So, if we look at x over b, over 3
Ca to the power x, this is psi. So, what we have defined as psi is equal to x into 3 Ca power
1 by 3 divided by b or x is equal to b 3 Ca power minus 1 by 3 psi.
So, we have got the equation the final equation
d cube eta, over d psi cube is equal to eta minus 1 over eta cube; which is very useful
equation, and we will look at it solution in the forward front, and bubble back region in
103
the letter classes. So, what we have looked at today is looked at what is the basic flow
characteristic or flow behavior in the Taylor flow. And the useful relationships between
film thickness bubble velocity, and ε G.
So, b ε G and UB relationship. We have also looked at that U TP is the average velocity in
channel; which is equal to Q G over plus Q and over channel cross sectional area
Then we will have looked at Brotherton’s problems in the Bretherton’s problem what we
did is applied lubrication approximation in film region. And substitute pressure from
Young Laplace equation. and then obtain a relationship between
eta over d psi cube is equal to eta minus 1 over eta cube where eta is h by b. So, this is the
radial coordinate or the or the coordinate in the transverse direction. And eta is and psi is
x over b 3 Ca power 1 by 3. So, this is a coordinate non-dimensional coordinate in the
axial direction. So, the solution of this equation will give us the bubble interface of course,
we will need to apply the boundary conditions to it ok.
Thank you.
104
Lecture – 06
Taylor Flow 2
So, in this lecture we will continue the discussion about the Taylor flow. So, in the previous
lecture we were talking about the Bretherton’s problem.
And we developed a relationship or the partial differential equation for the profile, for the
bubble profile or for the bubble interface in the non-dimensional coordinate 𝜂 and ξ. So,
this equation we developed for the front of the bubble, but if you again go through the
lecture, you will find out that there is no such assumption that stops us to use this equation
at the back of the bubble. So, we can analyze the front of the bubble and back of the bubble
independently using this equation. This equation is also known as Landau-Levich
equation, and has been used to understand number of problems.
So, 2 problems that we understand already is one is the flow of a long Taylor bubble, when
I say long; that means, that the bubbles should be long enough that it has a constant
thickness film region. So, if we look at the bubble ok so, this constant thickness film region
should be present for the bubble, to pass as at the long bubble. The other application of
105
this equation can be when there is no tail of the bubble. For example, air is passing through
a capillary, which is initially filled with liquid for example, for the coating applications.
So, there also the same analysis can be used to find out the thickness of the coating. The
other application which comes from Landau-Levich equation that if a plate dipped in a
liquid is brought out, then the thickness of the film. So, this plate is being pulled out with
a velocity U and the thickness of the liquid film that will be left behind can be analyzed
by the same analysis. As you can see that this is a non-linear equation. So, it is not possible
to have it is analytical solution. Nonetheless its asymptotic solutions are available and we
can find or we can try to understand the nature of the solution of this equation at different
regimes.
So, to start with we will consider a region; where η is greater than 1 and h over R is less
less than 1. So, to remind you we considered when we were analyzing this problem let us
look at again. What we did is we consider the front of the bubble. And we had that at the
the front of the bubble is spherical. And this is constant thickness film region; In the middle
this is the intermediate film region; where h is a function of x. So, and the film thickness
here is b; you might remember that η is equal to h over b. So, the film thickness non
dimensionalize by the film thickness in the constant film constant thickness film region.
So, we are considering the region where η is greater than 1, but it is a still sufficiently less
or sufficiently small than the channel radius. So somewhere in this region; where we have
the film to be of sufficiently thickness or sufficiently thick and it is very small when you
compared in the tube radius. So, in this region, if you look at the term η minus 1 η cube as
η will be large. So, if η is large, then η cube will be for the larger and then you can
approximate this to be equal to 0. So, that is why one can approximate this equation as d
η cube over d ξ cube is equal to 0.
𝑑2 𝜂
= 𝐴𝐹
𝑑𝜉 2
𝑑𝜂
= 𝐴𝐹 𝜉 + 𝐵𝐹
𝑑𝜉
𝐴𝐹 2
𝜂= 𝜉 + 𝐵𝐹 𝜉 + 𝐶𝐹
2
106
So, if you integrate this equation. So, let us integrate this equation, and we will get sorry
this is d cube η over d ξ cube, you might have this mistake carry it over everywhere. So,
please note that, and this will be; so, we will have d 2 η over d ξ 2 is equal to a constant.
Let us call that constant AF. Further if we integrate, it again we will get d η over d ξ is
equal to AF ξ plus BF, BF is another integration constant. And then we get η is equal to
AF by 2 ξ square plus BF ξ plus CF. So, that is the film profile and now we have 3 constants
AF, BF and CF. And remember this is in this particular region.
𝐴𝐹 𝑥 2
ℎ= (3𝐶𝑎)2/3 + 𝐵𝐹 𝑥 (3𝐶𝑎)1/3 + 𝑏𝐶𝐹
2 𝑏
1 𝑑2 ℎ
𝐾= +
𝑅𝐹 𝑑𝑥 2
𝑑2ℎ 𝐴𝐹
= (3𝐶𝑎)2/3
𝑑𝑥 2 𝑏
1 𝐴𝐹 2
𝐾= + (3𝐶𝑎)3
𝑅𝐹 𝑏
So, if we write this in the dimensional terms; so, we can write the previous equation that
we have just integrated h is equal to AF by 2. So, AF by 2 and b, this multiplied by ξ square
107
over 2. So, ξ square is x square over b square. So, this we can write is x square over b, and
this b will go away. AF by 2 x square by 2, and ξ is x over b 3 Ca power 1 by 3. So, we
will have 3 Ca power 2 by 3. Plus, BF into ξ into b. So, ξ will be x over b. So, b will cancel
out. We will have x into 3 Ca power 1 by 3 plus, b into CF. So, that is the equation in the
dimensional form and this equation is for the sphere.
Now, the mean curvature, if we look at this region, the region that we are talking about
here. This region will have 2 curvature, one along this direction; which will be given by d
2 h by dx 2. The another principal, curvature will be normal to it. So, if you are looking at
the cross section. Then, this will be given by the radius of the bubble so, the bubble can
the front this is at the front of the bubble. So, we can fit this at the front radius. So, this
radius will be above equal to front of the bubble.
So, the mean curvature in this region will be equal to 1 over RF. If we have this is the
surface, then this curvature is d 2 h by dh 2. And this curvature which where the sphere,
of this is circular this is the radius of this circle which is RF. So, kappa is equal to or the
curvature is equal to 1 over RF plus d 2 h by dx 2. And from this equation we have d 2 h
by dx 2 is equal to AF over b into 3 Ca raised to the power 2 by 3.
So, from this we could find what is the we have integrated, we found the profile in this
region, and then we saw that this the mean curvature in this region is the sum of these 2
curvatures.
108
Now if we look at this region the mean curvature in this region is constant. So, in this
region, if you look at the bubble; where you can fit a sphere near the bubble. So, in this
region, where we are analyzing. The curvature is almost spherical. And if you look at the
mean curvature this is 1 over RF plus AF by b 3 Ca to the power 2 by 3. Where RF is the
radius of this front of the bubble. And AF is a constant. B is the film thickness here; which
is a constant, and capillary number is also a constant.
So, that says that kappa or the curvature is a constant; that means, the surface tension is
still dominant force. Because, if the surface tension would have been changing that that
change is brought about by the viscous stresses. So, in this region what we see that the
curvature is constant; that means, this is dominant or surface tension is dominant in the
region, in this region and the viscous stresses are still negligible. So, for this Bretherton
suggested that this is a surface of constant mean curvature, which extends across the tube
with tangent nearly parallel to the wall.
So, when it says that you can fit sphere here, but this tangent is the slope of this line is
almost negligible. So, you can say that dh by dx is almost 0 in this region. It can also be
shown that one can fit a actually one can fit a circle, any circle can be fitted with a parabola.
So, this will be fitted at the apex on the parabola. So, in this region the circle can be fitted
with a parabola.
109
So, the equation that we had for h, we had h is equal to AF by 2 x square over b 3 Ca raised
to the power 2 by 3, plus BF x 3 Ca raised to the power 2 by 3, plus bCF. We can rearrange
this in such a manner, that we have h is equal to AF by 2 x minus xF squared or b 3 Ca
raised to the power 2 by 3 plus bCF.
So, what I have tried to do here, that we change or we select xF or the; is such that, the xF
has been chosen in such a manner that BF is eliminated. So, or that is what I have
suggested, here the center of the circle is located such that BF is eliminated, or in this
region if you look at a circle can be approximated by a parabola, and then you can look at
a publication my cherumuki et al. in micro nanofluidic. 2015, where they have described
this how this parabola can be fitted. But nonetheless, if we rearrange this equation in such
a manner; then we have eliminated the constant BF, taking into account the fact that as dh
by dx this will touch in this the circle is about to touch of the slope is negligible here.
So, the equation can be rearranged in this manner.
1 𝑑2 ℎ
𝐾= +
𝑅𝐹 𝑑𝑥 2
1 𝑑2 ℎ 𝐴𝐹
= = (3𝐶𝑎)2/3
𝑅𝐹 𝑑𝑥 2 𝑏
𝑏 = 𝐴𝐹 𝑅𝐹 (3𝐶𝑎)2/3
110
Now, in this place or as we said that there are 2 principal curvature, we had written the
curvature 1 over RF the first principal curvature. And the other principal curvature was d
2 h over dx 2. Now if this is a sphere, then for a sphere, the 2-curvature; curvature in this
direction, and curvature in this direction the 2-principal curvature will be equal. So, for a
sphere the 2-principal curvature are going to be equal. So, 1 over RF will be equal to d 2
h over dx 2, which is equal to AF over b 3 Ca raised to the power 2 by 3. That gives us b
is equal to AF RF 3 Ca raised to the power 2 by 3.
So, if we know the constants AF, and the radius of the front RF. Then we can approximate
this if we know AF and RF, then we can find out what is the film thickness. And the film
thickness from the film thickness we can further find out the UTP, not UTP by the
knowledge of UTP, we can find out UB bubble velocity and the void fraction. So,
Bretherton in his analysis he assumed that RF; that is, the radius of the sphere, that is fitted
here this is approximately equal to the channel radius R which is true for thin films. So,
that will reduce that b by R is equal to AF into 3 capillary number to the power 2 by 3.
So, if one can find out AF, then one know what is the film thickness this make things
simpler.
𝐴𝐹
ℎ= (𝑥 − 𝑥𝐹 )2 (3𝐶𝑎)2/3 + 𝑏𝐶𝐹
2
ℎ𝑥=𝑥𝐹 = 𝑏𝐶𝐹
111
𝑅 = 𝑅𝐹 + 𝑏𝐶𝐹
𝑅 = 𝑅𝐹 + 𝐴𝐹 𝑅𝐹 𝐶𝐹 (3𝐶𝑎)2/3
𝑅
𝑅𝐹 =
1 + 𝐴𝐹 𝐶𝐹 (3𝐶𝑎)2/3
𝑏 𝐴𝐹 (3𝐶𝑎)2/3
= 𝑣𝑎𝑙𝑖𝑑 𝑓𝑜𝑟 ℎ𝑖𝑔ℎ 𝐶𝑎
𝑅 1 + 𝐴𝐹 𝐶𝐹 (3𝐶𝑎)2/3
𝑏 1.34 𝐶𝑎2/3
=
𝑅 1 + 2.79 × 1.34𝐶𝑎2/3
So, recently a Klaseboer et al. in their physics of fluid paper in 2014. They have relaxed
this assumption of that one need to assume RF is equal to R. What they rather said that, in
the channel, near the bubble, where the sphere is being fitted, the channel radius R, at this
point where x is equal to xF is equal to RF, plus the film thickness at this location x is
equal to xF. So, recalling the h, we have h is equal to AF by 2. x minus xF square 3 Ca
raised to the power 2 by 3 plus b CF.
So, at x is equal to xF, we will have h is equal to bCF. So, that means, R is equal to RF
plus b CF, and from the previous relationship, we have b is equal to AF RF 3 Ca to the
power 2 by 3. So, we can substitute that here; we will have R is equal to RF plus AF, RF,
CF into 3. Ca raise to the power 2 by 3. Or that gives us RF is equal to R over 1 plus AF
CF into 3 Ca raised to the power 2 by 3. So, we can write b for the channel the film
thickness now becomes b is equal to AF into RF. So, RF we substitute by R. Or we can
write as b by R is equal to AF into 3 Ca raised to the power 2 by 3 into 1 plus AF, CF 3
Ca raised to the power 2 by 3.
So, this film thickness has an additional term in the numerator. You might check that the
numerator term is same. AF 3 Ca to the power 2 by 3 as obtained by the Brotherton, but
there is an additional term; which is 1 plus AF CF 3 Ca to the power 2 by 3 and which will
be valid for high capillary number also. So, that is the film thickness correlation that can
be obtained from there.
Now from their analysis the constants AF and CF were obtained by Bretherton by the
numerical integration and AF was 0.643. And CF was 2.79; so, one can get b by R is equal
112
to after substituting the values one will get 1.34 Ca raised to the power 2 by 3, over 1 plus
2.79 into 1.34 capillary number raised to the power 2 by 3. So, one can see from here, that
the film thickness is only a function of capillary number.
Now, right now we have looked at the solution of the equation d cube ξ over d cube η over
d ξ cube is equal to η minus 1 over η cube of for this equation. We have obtained the
solution in the region where η is sufficiently large than 1. Now if we consider another
region, which is near the film.
So, the region which is near the constant thickness film from where the film starts growing,
it might be a different or it might be considered even at the back off the bubble. So, they,
but they can be considered independently. So, in this region η is close to 1; so, η cube can
be approximated as 1 so that the equation can be linearized and one get the linearized
equation in this form; which can also be written as η triple dash minus η is equal to minus
1. And one can solve the characteristic equation of a this for homogeneous equation that
we get d cube minus 1 and one we get the characteristic roots.
𝜂′′′ − 𝜂 = −1
(𝐷3 − 1) = 0
𝜂 = 1 + 𝐶1 𝑒 ξ
113
√3 √3
𝜂 = 1 + 𝐶2 𝑒 −ξ/2 cos( 𝜂) + 𝐶3 𝑒 −ξ/2 sin( 𝜂)
2 2
And from that one can obtain the solution in this form η is equal to 1 plus C 1 is exponential
ξ plus C 2, exponential minus ξ by 2 and so on. So, C 1, C 2 and C 3 are constants here
and when ξ is increasing where at the front. So, when ξ is positive, then one can see that η
will be approximately equal to 1 plus C 1; e to the power xi, and this solution was obtained
by Bretherton at the back the solution that will be valid is η is equal to 1 plus C 2; e to the
power minus ξ by 2 cos root 3 by 2 ξ plus C 3 e to the power minus ξ by 2 sin root 3 by 2
xi.
So, by the nature of the equation one can see from this equation that the interface will be
at the front will be increasing exponentially away from the constant film thickness region.
Whereas, at the back there are some undulations in the film before, the film becomes on
the constant thickness. So, these 2 terms sine and cosine term will so, the oscillatory
behavior or the undulations at the back of the bubble. And this has been observed at the
experiment in the experiments. So, oscillatory or not let us say the undulations at the back
of the bubble. They can be explained by these 2 terms.
2𝜎 𝑅
Δ𝑝𝐹 = 𝑅𝐹 =
𝑅𝐹 1+ 𝐴𝐹 𝐶𝐹 (3𝐶𝑎)2/3
114
2𝜎
Δ𝑝𝐹 = [1 + 𝐴𝐹 𝐶𝐹 (3𝐶𝑎)2/3 ]
𝑅
2𝜎
Δ𝑝𝐵 = [1 + 𝐴𝐵 𝐶𝐵 (3𝐶𝑎)2/3 ]
𝑅𝐵
2𝜎
Δ𝑝𝐵 = −𝑝𝐵,𝑙𝑖𝑞 +
𝑅𝐹
2𝜎
𝑝𝐵,𝑙𝑖𝑞 = − Δ𝑝𝐵 = Δ𝑝𝐹 − Δ𝑝𝐵
𝑅𝐹
Now, so, we have obtained the asymptotic solution. But in this solution note that, what we
have not obtained or what we have not or we do not know what are these constants C 2
and C 3. So, this is only a qualitative representative how representation; however, one can
find out the wavelength of this undulations and then compare with the experiments. Further
if we look at the start of the Bretherton problem we discussed the pressure jump at different
regions near the bubble.
So, we will see that can this pressure jump be found out. So, if we look at the bubble front
the pressure jump at the front was 2 sigma over RF. And now what was not known is RF.
So, if you remember, we have obtained RF in terms of R and Ca on the constant. So, R is
equal to RF is equal to R over 1 plus AF, CF 3 Ca raised to the power 2 by 3. So, one can
write the delta pF is equal to 2 sigma by R into 1 plus AF CF into 3 Ca raised to the power
2 by 3.
Similarly, one can write the pressure difference at the back, that will be equal to 2 sigma
over RB, though we have not done the analysis, or we have not done the analysis for the
back separately. But from the similar arguments or from the similar analysis one will get
2 constants for the back which will not be necessarily same as at the front. So, we will
have those as subscripted as AB and CB. So, 1 plus AB CB 3 Ca raised to the power 2 by
3. That will be the pressure difference inside the bubble and outside the bubble near the
back in liquid. So, these are the pressure jumps that one can find out. And from this, one
can also calculate the delta pB is equal to p at the back of the bubble in liquid minus p in
the bubble which is 2 sigma over RF. So, one can find out what is pB in the liquid, that
115
will be 2 sigma sorry this will be because the pressure in the bubble will be higher. So, this
will be positive and this will be negative.
So, we will have pB in the liquid is equal to 2 sigma over RF minus delta pB. So, that is
basically delta pF. So, that will be equal to delta pF minus delta pB, and that will be the
overall pressure drop from the back of the bubble to the front of the bubble. And one can
find out knowing AF, A, AB, AF, CF and AB CB; AF CF we already know AB CB can
also be found out and looked at by the paper by Cherumuki.
So, till now we have looked at the Bretherton problem and in the Bretherton’s problems,
we have analyzed the shape of the bubble and from that we were able to find out the
important and useful relation for the film thickness. And once we have the film thickness,
we can also calculate the bubble velocity, we can calculate the void fraction. We can also
calculate what is the total pressure drop across the front of the bubble, and if we know the
constants for the back of the bubble, then we can also calculate the total pressure drop
across the bubble.
Now, we will have a look at the flow field in the liquid slug. So, if you have this picture
of the liquid slug, where there are 2 adjacent bubbles at the front and the back and in a
laboratory frame of reference, if the slug is long enough, for long slugs. If the slug is long
enough then one will get the profile to be parabolic, or at least near parabolic in the
116
channel. When the profile is parabolic in the channel, then we can approximate this profile
as 2 UTP 1 minus r by capital R squared in laboratory frame of reference.
So, when we write this in velocity profile in bubbles frame of reference. So, remember we
said that, because the bubbles move the velocity UB, the slug that is trapped between the
2 bubbles will also need to move with the bubble velocity. So, in the bubble frame of
reference the slug will also be stationary and we will see the internal recirculations in the
liquid. So, the velocity profile in the slug; in the laboratory frame of reference will look
like this, but if we plot the velocity profile in slug or bubbles frame of reference, then we
will see that the profile will look like this.
So, that means, there is recirculation in the slug, and the profile that we have drawn is in
the middle of the slug. You may also note that the velocity profile that we are talking about
will be observed in the middle of the slug not near the 2 ends at the front or back of the
slug; where they have a bubble front and bubble slugs close by. So now, if we do that then
we can have 3 conditions for UB and based on the relationship between UB and UTP.
So, let us say if UB is less than 2 UTP, then we will have v r is equal to, it will be the
profile look like something like this. This is the profile for a, this is the parabolic profile.
And this is 2 UTP; this value is 2 UTP; so U and this is where we have 0. So, UB will be
somewhere here; so we will have a recirculating flow in the slug.
If UB is greater than 2 UTP; that means, the velocity profile. So, this will be something
like this; in this case, because UB is more than UTP. So, the negative value will be
everything will be moving in the negative direction and one will not see any recirculations.
But it is unlikely to be; such condition to be present for air water flow for example.
So, in this case there is no; recirculations in the slug and if we have the third case, where
there will be equality UB is equal to 2 UTP, then one will have the profile to be exactly at
the same point.
117
So Taylor in his paper on the film thickness in the capillaries; he also drawn 3 different
profiles and this was for when UB is greater than twice the average velocity or mean
velocity. And these 2 were for UB is less than 2 UTP; so from our analysis what we have
done in the previous slide both of these have re-circulating flow in the channel; however,
this does not say anything about that, will there be 2 stagnation points on the bubble or
will there be only one stagnation point; so the recirculation will not be just close to the
bubble from it. So, these are the 3 hypothetical flow profile that were derived by or that
were shown not derived but shown by Taylor or three possibilities
118
𝑈𝐵 < 2𝑈𝑇𝑃
𝑟2
𝑉𝑥 = 2𝑈𝑇𝑃 [1 − 2 ] − 𝑈𝐵 [𝐵𝑢𝑏𝑏𝑙𝑒 𝑓𝑜𝑟]
𝑅
𝑈 𝑟0 2
𝑉𝑥 =0 → 2𝑈𝐵 = 1 −
𝑇𝑃 𝑅2
𝑈𝐵
𝑟0 = 𝑅 √1 −
2𝑈𝑇𝑃
𝑅 𝑈𝐵
𝑟0 = √2 −
√2 𝑈𝑇𝑃
So, another thing that we will discuss here that in quite a few cases, we have UB is equal
to or UB is less than 2 UTP. So, that means, we will have recirculations for many practical
cases, we have UB is less than 2 UTP; so they have a recirculations in the liquid slug. Or
one have recirculations in the liquid slug; then of course, one would like to know what is
the size of this recirculation? So in this; we will talk about the size of the recirculation. So,
we had written the velocity profile in the liquid slug, in bubble frame of reference 2 UTP;
1 minus r square by capital R square minus UB.
So, if the profile is recirculating also, there will be a point, at which you will have the
velocity to be 0. So, the center of the recirculation; so the ring that will form, so the radius
of the so, what we call r 0, let us say center of recirculation. So, at this point v x will be
equal to 0 and just remember that this is in bubble frame of reference.
So, when v x is equal to 0; that means, we have UB over 2 UTP is equal to 1 minus r 0,
square r 0 is the location where the center of recirculation is located divided by R square.
So, we will have r 0 is equal to 1 minus UB over 2 UTP square root, multiplied by R. Or
one can write in this form; r 0 is equal to R over root 2 into 2 minus UB over UTP.
So, that is the center of recirculation. The other parameter of interest will be the radius of
this the entire circulating zones.
119
𝑟0 𝑟1
∫ 𝑉𝑥 2𝜋𝑟𝑑𝑟 = ∫ 𝑉𝑥 2𝜋𝑟𝑑𝑟 => 𝑟1 =?
0 𝑟0
𝑟0
∫ 𝑉𝑥 2𝜋𝑟𝑑𝑟 = 0
0
𝑟1
𝑟2
∫ (2𝑈𝑇𝑃 (1 − ) − 𝑈𝐵 )𝑟𝑑𝑟 = 0
0 𝑅2
So, this r 1 how can we find r 1 here? So, there are two ways to look at it; if you look at
the flow that is happening from here to here. See if you look at the flow; Q or you can
write this Q in the integral form. So, if we write integral 0 to r 0 ;v x 2 pi r; d r. That will
be the flow that is happening in this direction. And the same flow will move during the
recirculation; so that should be equal to integral r 0 to r 1; v x 2 pi r; d r; so one can find
out from this what is r 1.
Alternatively, one can see that the net flow in this region integral 0 to r 1; v x 2 pi r; d r.
That is basically the sum of these 2, but the other one in will be in the different direction.
So, the directions will change that will be equal to 0 and once we substitute the values 0
to r 1; 2 UTP into 1 minus r squared by R squared, minus UB into r d r is equal to 0 2 pi
can be eliminated, because it will can be divided by divided to 0 that will have no effect
on the result.
120
So, after the substitution, one will get the value of r 1; which is the radius of circulation.
So, you can notice here that r 0 and r 1. The size of the recirculation they are function of
UB by UTP only. And UB by UTP is a function of b over R; which is the film thickness,
which is a function of capillary numbers. So, you can say that the size of the recirculation
zone also depends on the capillary number, which is mu U over sigma. So, if the liquid is
viscous, it means viscosity is high, then one might have a lower size of the re-circulations.
So, in summary in this section what we have analyzed is we looked at the Bretherton’s
problem.
And in the Bretherton’s problem, we looked at the flow in intermediate region. First, we
divided it into 3 regions; where in the other 2 regions near the front and in the film
thickness in the constant thickness film cylindrical region, we had some information about
the flow; flow was film are is stagnant.
So, we looked at flow in the intermediate region, use the lubrication approximation and
obtained what is the expression for v x. Once we have obtained this. This was in terms of
dp by dx and we obtained a case on from Young Laplace equation. So, Young Laplace
equation had pressure and h correlating or h double dash are correlated. Because the one
of the curvature much can be approximated as double derivative of the film.
121
So, from that we can substitute this pressure, and then we can find out what is p dash. And
that will be as a and then we got a relationship between. So, from then we got relation
between h and x; which are basically the transverse and streamline coordinates of the film
in the intermediate region. And when we non dimensionalized it was h was replaced by
eta, and x was replaced by ξ and we got it will associate d cube ξ over d η cube is equal to
η minus 1 over η cube.
And then we looked at because this equation is non-linear then we looked at the or the
solution of it for the 2 regions one is where η large, and η minus 1 over η cube can be
approximated to 0. And the other region near the film, and this was approximated by a
sphere. And in other region we looked at where η is about equal to 1. So, we had this as
approximated as η minus 1 over η cube is equal to η minus 1.
So, when it was sphere then we compared the 2-film thickness are 2 principal radii of
curvature, because it can be fitted in a sphere and these 2 curvatures will be equivalent.
We got what is the front radius of the front spherical portion and from that we could
calculate the film thickness. The other thing that we looked at is the flow behavior or the
flow in the slug; specially, the size of recirculation zone in the slug.
And from this all this analysis we have got expressions for film thickness, for UB, for
epsilon G or void fraction, for r 1 size of the recirculation zone and so on. So, we will
continue our discussion about the Taylor flow and transport processes in it in the next
lecture.
Thank you.
122
Lecture – 07
Computational Techniques
Hello, in this lecture we will be talking about the computational techniques that can be
used to model multiphase flow in micro channels. So, as this course is not predominantly
on computational methods, and there is no prerequisite that one should have a basic
understanding of computational fluid dynamics or other computational methods. So, we
will first look at what computational fluid dynamics is, and then we will talk about what
are the equations that are required to be solved for multiphase flows.
So, what we will be looking at that how our mathematical model can be formulated for
modeling multiphase flow in particular with respect to micro fluidics.
So, before we do that, let us looked at look at very briefly what computational fluid
dynamics is. So, if we a look at single phase flow, the single-phase flow under continuum
assumption can be modeled using Navier stokes equations; which momentum
conservation equation, and mass conservation equation, and the fluid is Newtonian as you
can see from this term.
123
So, the mass and momentum conservation equations in this; what we have as unknown is
the velocity vector, and pressure they are our unknowns. So, we have 2 equations in the
vector form you can say, and then we need to solve these equations for obtaining or to
obtain pressure and velocity. Now because this equation is non-linear, the equation
momentum equation, the convective term as you can see here is non-linear. So, it is not
always possible to get a solution analytical solution for this partial differential equation.
And we need to use some numerical method to do so.
Only under certain conditions, when we can eliminate or when we can linearize the
convective term, under those conditions it is possible to obtain an analytical solution for
this Navier stokes equation.
So, there are a number of techniques that can be used, or that have been used to solve
Navier stokes equations. We are just describing or I am just going to describe only one
technique very briefly, to give you an idea what are some of the terms that we might be
talking about later on so that you do not feel that you do not know those terms.
So, in computational fluid dynamics in particular the finite volume method, when you
solve a fluid flow first you need to choose your domain of interest or the computational
domain and in the commercial softwares, you need to create a geometry for this domain.
And then as we already have the system of equations. So, if our flow is incompressible
124
and isothermal, then we just need to solve the mass conservation and momentum
conservation equations.
If the flow involves heat transfer, then we also need to solve energy conservation
equations. And the things become more complicated when flow is turbulent, but we will
not go in the detail of turbulent flows; more so because as we discussed earlier in
microfluidics flow is often laminar. So, in microfluidics application we are generally not
concerned with turbulence flows.
The domain because it is always useful to have a smaller domain if possible; it is also good
if you can decrease the dimension of the domain from 3 dimension to 2 dimension. So, if
the things are not significant or if the velocity and velocity gradient in the third direction
are not significant, then we can solve the problem in 2D in the Cartesian coordinate, it is
in x and y coordinate and the third coordinate j Z coordinate can be neglected v z is equal
𝜕
to 0, and 𝜕𝑧 for all the variables will be 0.
Similarly, in cylindrical coordinate system, where if the problem is axisymmetric; that

means, if there is nothing happening along the azimuthal or angular direction, then v theta
or velocity in the angular direction if it is 0, and the gradients in this direction is 0, then
the problem can be solved as an axisymmetric problem and one need to solve only r and z
equations. And then reduces the number of equations that need to be solved. So, our
computational time and computational cost will be reduced.
So, that is about the computational domain. Now one can also take advantage of the
symmetry of the problem, and can just solve the problem in half of the domain for example,
a rectangular or square channel. Then one can solve the problem in one fourth of the
domain with the symmetry boundary conditions on these 2 phases. So, once we have
selected a domain, then we have domain boundaries and we need to define the appropriate
boundary conditions on the domain boundaries.
For the wall generally, we have no slip boundary condition no slip means, that there is no
slip between the solid wall and the fluid. So, that means, the velocity of the fluid adjacent
to the wall is equal to the velocity of the wall. So, if wall is stationary then the fluid velocity
is also 0. At the inlet we can define either a flow rate or a velocity profile or a pressure
profile. Now velocity it can be uniform or it can be or there can be a profile.
125
When you define a uniform boundary condition, say for example, if this is flow between
2 parallel plates and if you define the uniform velocity at this, then there will be a ill posed
boundary condition at these corners. Because the velocity will be 0 at the wall, but from
the inlet the velocity is non-zero. So, you will see some problems, there the pressure take
will take pressure will take unphysical values at those points, but generally this can be
handled.
So, it is always a good idea to define a velocity profile which defines the velocity, or which
has the velocity 0 at the walls. At the outlet one can have 0 gradient; which is generally
known as outflow boundary condition, or constant pressure or a pressure profile at the
outlet.
So, one can one has to define the domain, and identify the boundaries and give the
appropriate boundary conditions at the domain.
Once we have done that then we have formulated the problem; the equations are there and
the domain in which we want to solve and the boundary conditions are there.
Now, we need to discretize it; so discretization is the process of converting the continuous
partial differential equation into discrete algebraic equations. So, the process of converting
a continuous equation into discrete algebraic equation is known as discretization. So, when
126
we do it for the terms which have derivatives in a space, for those we need to divide the
domain or discretize the domain into a smaller parts.
Similarly, to solve in time, we need to have explicit or implicit time marching scheme; so
we need to discretize in time. And there are number of techniques for discretization in a
space; in computational flow dynamics 3 most popular techniques are finite volume
method, finite element method and finite difference method. So, FVM is finite volume
method, finite element method and finite difference method.
So, we are not going to discuss finite element and finite difference method we have been
briefly, look at finite volume method. So, in the finite volume method; the entire
computational domain, entire volume or entire area, depending on the problem is 2
dimensional or 3 dimensional is divided into smaller volumes or smaller cells. So, each
the domain has been for example, here has been divided into different cells as you can see
here; along these lines and the center of the cell, we call cell centroid and the intersection
or these points are called nodes, and there are faces of the cells; this is not simple
terminology.
So, in the continuous domain, or in when we have the partial differential equations, then
we have the values or all points. Whereas in computational fluid dynamics we are
discretizing and we get the values at the center of the cells.
127
Apart from that so, what is done in finite volume method, what we have here is a simple
diffusion equation with source term. So, this is the diffusion term, and this is source term
and in finite volume let us say they are 3 adjacent cells for a one-dimensional problem. So,
the length of the name of the cells or the cells enters are P which is point W a point on
west of it and E a point east of it; the faces are being represented by small w and E small
e for west and east sides respectively.
The distance between the points W and P and P and E is ∆ x; subscript w and ∆x subscript
e. Now let us integrate this equation d by dx and when we integrated and put the limits
from west to east. So, this integration when we write then we can put it in this form after
substituting the limits that ϕ d 𝜙 by dx e minus Γ d ϕ by dx w plus integral w t S dx.
Now, when we integrate along these lines, then we need to have a assumption for the
profile inside this, because what is the value of d ϕ by dx; so, we need to know that how
the or we need to make an assumption that, what is the profile variation of ϕ in this at; the
surface between W and P and between P and E. So, that can be used to find out d ϕ by dx
at E and d ϕ by dx at W. So, if our domain is sufficiently small if our mass size; if our
these elements are sufficiently small, then for any profile it can be approximated by a linear
profile. So, that is what has been done here and ϕ d ϕ by dx is replaced by ϕ E minus ϕ P
over ∆ x e similarly, for the other derivative and then it has been written.
So, when we collect these terms together and then wright; so all the coefficients of ϕ P has
been written into has a P and all the coefficient of ϕ E as a E all the coefficient of ϕ W has
a W. And all the constant terms have been collected in b; so what we have been able to do
that, we have been able to convert a differential equation to a algebraic equation, for one
cell.
And the same exercise needs to be done for all the cells; the treatment will be different at
the boundary nodes because the value at the boundaries will be known. So, we have to
take that into account so that the boundary conditions can be used in the equations and
then these number of equations are solved.
So, that more number of elements we have; the more number of equations we need to
solve; that means, we need more and more computational time. That is why we are always
concerned about reducing the number of elements or reducing the computational load. But
128
at the same time as I said that where the gradients are large, or where the curve is changing
sharply, there we need to have more grid points to capture the accurate information.
So, the first process was that we discretize, or first process was we find out a computational
domain, define the boundary conditions on this domain. Now the next step is discretize
the equations. Once the equations have been discretized then solve these equations.
So, you have a number of algebraic equations, and then we need to use generally an
iterative solver to solve these equations. And once we solve the equation, what we will
get? We will get the values of the 3 velocity components, or the 2 velocity components
depending on 2D or 3D problem and the pressure.
So, from that once we have solved the equations and got the values, then we can extract
the quantities of our interest for example, wall shear stress if the heat transfer is there then
maybe Nusselt number or pressure drop or vorticity or any other information that we can
construct from this parameters that we have at our disposal. So, this is in very brief what
the idea behind discretization is and how the partial differential equation is converted to
algebraic equation, and what the term mass or an element or grid element mean.
129
So now coming to 2 phase flow, as we discussed earlier that for a single-phase flow; we
need to solve a mass and the momentum conservation equation for an incompressible and
isothermal flow. Now, we have 2 phases then we need to consider one set of conservation
equations for each phase. One for the phase 1 and one for phase 2; so, we have the 2
equations here, the mass conservation and momentum conservation equations k is phases.
So, k represents different phase and it can be phase 1, phase 2; so, k is equal to 1 and k is
equal to 2 and depending on we will have these terms.
130
Now, at the interface because when we have 2 phase flow, they have the usual boundaries
of the domain plus 1 boundary is the boundary between the 2 phases and this boundary
would be moving; it is not a fixed boundary.
So, we also need to have the boundary conditions at the interface. So, as you might
remember from your if you have done a course on transport phenomena or then probably
you would have studied. That for 2 phase; 2 fluid system, the boundary conditions, the
kinematic boundary condition at the or the velocity boundary condition at the interfaces,
that if there is no phase change or there is no mass transfer at the interface. So, if this is
our interface between fluid I and fluid II.
And if then the fluid at this region is moving as a normal velocity component as v N; then
this is v 1 N and this will be; sorry this is v 1 N and this is v 2 N so, v 1 subscript N and v
2 subscript N. Now, if there is no phase change of mass transfer, then this interface will be
pushed by a velocity v 2N by this fluid, and then the same displacement will happen here.
So, the velocity at the interface or the normal component of velocity at the interface, or the
velocity component normal to the interface; they will be equal in case of phase change or
in case when there is no phase change, or there is no mass transfer. If there is mass transfer,
because of phase change or otherwise then we need to take into account also the flow rate
or the phase flow that there are flow that is happening from one phase to another; that we
are not going to discuss here.
Similarly, the velocity on the tangential velocity component v 1 t in fluid 1 and v 2 t for
fluid 2; they will also be equal. So, the tangential component are equal; the normal
components are equal, and if you combine the 2, then you can say that velocity vector 1,
and velocity vector 2 at the interface they are equal. So, velocity is continuous now; that
means, the velocity is continuous at the interface, if there is no phase change or mass
transfer in the fluid between the 2 phases.
131
Now, the boundary condition for the stresses; so, let us look at the normal stress now. The
normal stress, what we have pressure is also a normal stress. So, as you might remember
from surface tension discussion, that the difference in pressure for the static condition P 1
and P 2; the difference between is equal to 𝜎𝜅. So, the total jump in the extra tensor is a
normal stress tensor. So, pressure as well as normal stresses, that might be there you could
the flow that will be equal to σκ.
Now, if there is Marangoni convection which is caused by the gradient of surface tension.
So, this term is because of so, if there is a surface tension gradient, then we have jumped
this is wrong this is tangential. So, this should have been here, and this should have been
here anyway. So, jump in the stresses normal to the interface, is this where as in the
tangential direction to the interface and crossed t 2 minus t 1 that is the tangential
component of the stresses, that will be equal to ∇𝜎; in case of there is no and this will be
equal to; in case there is no surface tension gradient, when sigma is constant, then the
tangential stress will be continuous in the interface, but there is a jump in the normal stress.
132
So, for the 2 phase flows; we have written down the equations, and we have understood
the boundary conditions. Now there can be number of different approaches to model
multiphase flows. You might have heard about Eulerian-Eulerian multiphase flow
modeling method, or Euler-Lagrangian multiphase flow modeling method, or you might
have also heard about the volume of fluid method, level set method or in general the term
which is called interface capturing method which are used to model multiphase flow.
So, the question comes which approach one should take to model multiphase flow method.
Say in general for the industrial applications for conventional multiphase flows; the
number of bubbles, the number of droplets that are there in a system. For example, consider
a bubble column which is a column of liquid in which the gas bubbles are introduced and
the number of bubbles are millions. So, if we want to capture the bubbles in this column,
then we need to have the interface; if you want to capture the interface at each and every
bubble, then we need to have very refined mass.
So, in general in such cases the equations are averaged, and the exchanged between the 2
phases, momentum exchange and if there is a heat transfer or mass transfer exchange. That
is modeled by closure terms; so, that the computational time and computational cost is
reduced and a still one can find useful information about the flow. However, when it comes
to flow in micro channels, then the number of bubbles are limited.
133
Flow is often laminar and it is quite regular; so, in such cases we also see that the interface
is quite large, or the interface is quite big. So, one there is a motive or there is a motivation
to capture the interface and capture the information accurately; rather than having a closure
model which will come from either from the experiments or from some heuristic or
empirical models.
So, generally for most of the cases in micro fluidics one do not need to have one do not
use of Euler; Euler method which are generally based on the averaging of the equations.
Rather one uses interface capturing or interface tracking methods; so, let us look at what
those interface tracking methods are. In this case, a single fluid formalism is applied; that
means, only one set of conservation equations, even though we have 2 fluids, or more than
2 fluids, then let us talk about only 2 phase flows.
So, even though we have 2 fluids, only one set of conservation equations are solved, and
the interface between the 2 phases is captured explicitly. So, such things or such case will
be well suited for microfluidic applications when we have one droplet going in the channel
or few bubbles going in the channel, and we can capture the flow behavior by stream lines
and the bubble velocity, the droplet velocity, the heat transfer and so on.
So, when we have 2 fluids being modeled by a single set of equation, then we also will
need something by which we can identify the 2 phases. And we can distinguish between
the 2 phases; for that one additional equation is solved or one additional for the advection
of the interface, and this particular feature distinguishes different methods which are there
to model 2 phase flows. For example, in the volume of fluid method, the volume of fluid
that is present in a cell is used as a marker function and that is how one identifies the
interface.
In level set method, the interface is a level set function is solved or the level set function
is equation for the advection of the level set function is used. So, what is level set? That
means this is the interface, then distance from the interface is called level set function.
So, that means that the function when ϕ is equal to 0 that represents the interface. So, del
ϕ; so the equation is solved for the advection of ϕ and the ϕ is equal to 0 represents the
interface. For solving multiphase flows, one uses generally interface capturing methods
have been used, there are number of other methods, but we are going to describe the
134
general approach and the readers are referred or the students are referred to look at another
moocs course or look at further literature to know more about these models.
So, in the single fluid approach, we treat that all the phases are one material. Now if you
look at these equations, in this we have 2 properties. One is density, and another is viscosity
of the 2 fluids. So, one need to identify, when we have an interface and if this interface is
to be captured than in such case, we will have at any point in a space, except at the
interface; one will have that the fluid is either fluid 1 or fluid 2.
So, if it is in fluid 1, then we use ρ 1, if it is fluid 2, then we use 𝜌2; similarly, for mu. So,
the properties are variable, and we have to define it in such a manner, that if we have a
marker function C, which tells let us say if it is 0, in fluid 1 and it can be 1 in fluid 2. So,
once we have identified based on the marker function, that in which fluid we are; we can
use the properties of those fluids in these equations. Now comes So, basically by having
variable properties we are solving 2 equations. Now comes the interface the boundary
conditions at the interface.
So, the boundary conditions that we will use at the interface are if you remember the
boundary conditions that we discussed that the interface, we need to have a velocity
continuity. Because at the interface; in the absence of evaporation and mass transfer, the
normal components in the 2 fluids are equal tangential components in the 2 fluids are
equal; so, that means, velocity is continuous. In any way, in any case, the velocity for our
135
system of equation is continuous. So, the velocity continuity is inherently built in this
system of equations.
The next part is the dynamic boundary conditions at the interface. The first one is the
tangential stresses; so, if there is no Marangoni stress, then the tangential stress is
continuous which is inherently built in the system of equations, the stresses are going to
be continuous; so, fine. The other boundary condition which we had is; the normal stresses
or let us say P 1 minus P 2 is equal to σκ; which is becoming because of the surface tension.
So, and that will also take into account that the surface tension in the 2 fluids are different.
So, we have to have a model or a term which can take account take into account this jump
in boundary condition or in the normal stress or pressure at the interface. So, to take that
into account this new term has been added which we call as F SV and we will discuss this
in later slides. So, in summary what we have looked here is that in the single fluid
approach, we still solve the Navier stokes equation, only for only one fluid. Now for this
one fluid we have the velocity and the density oh sorry, the density and viscosity; these
are the 2 properties of the fluids.
Now, the question comes which fluid properties; so, we have a marker function, and we
need to know with this marker function; if we know let us say we know the values of the
marker function in each grid, then we will be able to define that what is the value of ρ, and
what is the value of 𝜇 at any cell. The cells that have interface in them, in those cells we
will have the velocity, the viscosity and density as the average volume fraction weighted
average of the properties of the 2 fluids.
136
∂C
+ 𝑉. ∇𝐶 = 0
∂t
𝐷𝐶
=0
𝐷𝑡
Now, there are a number of challenges, or number of issues that we need to address to use
a single fluid approach to model 2 phase flow. The first thing is that as we were talking
about that when we have a 2-phase system, and we will have a marker function, that C is
equal to 0 in one phase and C is equal to 1 in another phase. So, we need to have a marker
function; so, we need to solve a equation advection equation in terms of del C by del t plus
v dot; the velocity vector dot del C is equal to 0.
So, or we can say that the material derivative of this is equal to 0 for this marker function.
So, the marker function takes different values in different fluids. And each time the marker
function must be updated to account for the moving interface, when the interface is
moving. So, this is a critical and difficulty step in some cases; we also need to take into
account as we just discussed that the surface tension jump should be taken into account
and how this the term we had F SV in this system of equations how that can be modeled.
So, one need to take into account that the terms that we have in the Navier stokes equation,
if we go back to these equations every term is say for example, in the momentum equation,
we have force per unit volume. The unit for each term in the momentum equation; if you
137
look at the easiest to look at is 𝜌 g. So, if this term is in terms of force per unit volume or
for example, pressure is force per meter square and it is damage per meter; so Newton per
meter cube or forced per unit volume.
So, this is true for a body force the terms are for the different body forces. Now if you look
at the surface tension; surface tension is a surface force which acts only at the interface.
So, this is basically or it is not exactly a force it is a jump in the pressure or normal stress
at the interphase. So, how do we incorporate this in equation which has; which all the terms
are body forces? Or how do we convert this into a body force or a volume force?
So, this is done using Green Gauss theorem and the surface integral is turned is converted
into a volume integral using Green Gauss theorem, and then surface tension force is used
as a body force in the momentum conservation equation as a source term.
So, let us look at the marker function, marker function is that it has one value one in one
fluid and 0 in another fluid. And when this interface moves; so, when the interface moves,
the shape of the region changes, but each fluid particle has it is identity so; that means, this
material derivative is 0. So, that is why we have del C over del t plus v dot del C or DC by
DT is equal to 0.
∂C
+ 𝑉. ∇𝐶 = 0
∂t
138
𝐷𝐶
=0
𝐷𝑡
So, that is how one more equation; so, we had 2 conservation equations mass conservation,
momentum conservation plus 1 advection equation for the color function or the marker
function. This marker function also known as color function; so, we can write here that it
is also known as color function.
At the interface, when we want to implement something or when we want to implement

the surface tension force, then we need the curvature. If you remember the pressure jump
at the interface or Laplace pressure is defined as σκ; where κ is curvature of the interface.
So, we need to identify the curvature of the interface. And this information can also be
obtained from the color function or from the marker function. So, the unit normal to the
interface; that means, the vector normal to the interface at any cell is can be obtained by
del C over the magnitude of del C vector or the gradient of color pencil vector. Once we
have obtained the normal to the interface, the curvature is obtained by del dot n; if from
the relationship the curvature is del dot n. So, from this we can calculate the curvature.
139
So, lot of implementations of a volume of fluid method, the inaccuracies or the problems
that occur; they occur because of the inaccurate definition of or the inaccurate
implementation of n.
Having clarified that marker function or that we need to solve one equation for the marker
function, and how do we find the curvature of the interface. Let us look at how we can
include the surface tension.
So, the surface tension jump or the pressure jump at the interface is non zero at the interface
only. So, at the interface we have built a function which is 0 everywhere else, and non zero
at the interface. So, if we have we have plotting it an on a scale; so, the value of so, the
value of this force is non-zero near the interface, 0 everywhere else.
And in these methods, one need to appreciate the fact that the interface in reality is of
almost 0 thickness; however, the interface is fairly thick in several other cases. The
interface is quite thick in the in really interface is thin, but in the computational methods
it is at least one cell or one element thick because it is being discretized. So, the interface
is what is called the interface is diffused.
So, the surface tension force is also diffused over the area, or it is distributed over the area
in which the color function is between 0 and 1. This surfaces tension force is implemented
only at the interface for the σκ n gives the direction of the force. And del is Dirac delta
140
function, which is non zero at the interface only and this model is known as continuum
surface force models of the surface force is converted into a continuum forced and included
in the system of equations of the Navier stokes equation as a body force.
𝐶= 𝛼
∂α
+ 𝑉. ∇𝛼 = 0
∂t
𝛼1 + 𝛼2 = 0
So, what we have described is the general a structure of any single fluid formalism or
single fluid approach to model 2 phase flow. In volume of fluid method, this color function
or C is equal to alpha, where alpha is volume fraction of one of the phases. So, the equation
becomes del alpha over del t plus V dot del alpha is equal to 0, and if you look at this
equation what we get is the mass conservation of amount of phases. And alpha 1 plus alpha
2 is equal to 0; so, basically what this gives the conservation of mass for each phase.
𝜌 = 𝜌1 𝛼1 + 𝜌2 (1 − 𝛼1 )
𝜇 = 𝜇1 𝛼1 + 𝜇2 (1 − 𝛼1 )
So, the mass of each phase is inherently conserved which is not the case for other methods.
And in this case once we have obtained alpha, the ρ is equal to ρ 1, alpha 1 plus ρ 2; 1
141
minus alpha 1. Similarly,μis equal toμ1 alpha 1 plusμ2 1 minus alpha 2; while this is
accurate and it can be proved, this is just an approximation. But generally, works fine for
most of the cases. So, you can see that accepted the interface this ρ is equal to ρ 1, where
alpha is equal to alpha 1, and ρ is equal to where alpha 1 is equal to 1 and ρ is equal to ρ 2
and alpha 1 is equal to 0; this is about the interface capturing methods.
Now, there are number of other issues other problems that one need to look at when
modeling 2 phase flows. For example, the contact line or 3-phase contact line plays an
important role in number of micro fluidic flows. So, one need to take into account that how
do we model the contact line singularity near the interface. Because the contact line is
moving what is the contact angle and the contact angle it is a dynamic contact angle; so,
one need to have a model for dynamic contact angle.
Sometimes the films are very thin in micro predict flows. So, one need to have the mesh
very, very refined so that the thin film flows can be captured. Similarly, during coalescence
or breakup of the bubbles; the films that form between the 2 phases are very thin.
Sometimes of the order of few nanometers, and one cannot have over cannot afford to have
such small very such a small mesh, because that will require over the length scale changes
about 5 to 6 order of magnitude.
So, how one can model such thin films or the thin films that forms during breakup and
coalescence. Heat and mass transfer especially capturing the boundary layers near the
142
interface accurately when condensation or interface mass transfer or evaporation happens
and similarly, is of for the phase change modeling.
So, in summary in today’s lecture, we have first looked at specially for those who have not
been introduced to computational fluid dynamics. First, we have looked at the basics of
computational fluid dynamics, and then one is encouraged to read through books for; if
they want to study the computational flow in computational fluid dynamics of the book by
Patankar curve numerical fluid flow and heat transfer.
By S. V Patankar can be a good starting point to understand the basics of combinational

flow dynamics especially the finite volume method.
And then so, we have looked at the basics of or very briefly what computational fluid
dynamics is. And then we have looked at the governing equations for general multiphase
flows or general 2 phase flows and then what are the boundary conditions at the interface
for these flows. Then we have come to the interface capturing method; so, first we
introduced CFD. CFD, and then equations for 2 phase flow and then we have looked at
interface capturing method, specially the VOF or volume of fluid method, in which the
surface tension; how it is used as a body force in the Navier stokes equation. So, that’s all.
Thank you.
143
Lecture – 02
Bubble and droplet generation
Hello, so the main advantage of microfluidics specially, for applications in analytical

chemistry, biotechnology and chemical processing. It stands from the fact, that one can
control the size of bubbles or droplets precisely so, as to be able to create mono dispersed
bubbles or mono dispersed droplets, emulsion etcetera for a number of applications and
they can have a very precise control over the processes and can control and manipulate the
interfaces.
So, in order to be able to do that one needs to understand the mechanisms of the formation
of these droplets or bubbles. So, if one look at the literature the experiments that have been
performed in micro fluidics or the applications that have been developed, there is lot of
work that has gone through in understanding the formation of bubble and droplets. So, in
this lecture we will try to review some of the droplet formation and bubble formation
mechanisms and what are the factors that may affect them, what are the different regimes
that may exist during the formation of droplets?
144
So, as I said that many a processes that are relevant to micro fluidics They require very
highly controlled formation of interfaces that is bubbles and droplets one example is
formation of dispersions. Now, this can be achieved or this has been achieved by the design
of the channels. So, as you will see that number of channels that are of relevance in micro
fluidics. They are rectangular in nature and this is achieved by say T junctions.
So, in the T junctions the continuous phase which is generally a liquid comes from the
steady stream. And the dispersed phase enters from the and the size of these 2 channels d.
And let us say 𝑑′ they may or may not be same; the channel depth may or may not be same.
As you might have seen or might realize that a number of applications especially, because
of the ease in manufacturing number of applications in microfluidics the channels are
rectangular or square in nature. So, most of these channels are rectangular or square. So,
one of the configuration that is often used is T junction a variation of T junction is say Y
junctions and this angle theta between the 2 channels might vary. So, this is Y junction.
And another approach that has been used to generate flow is known as flow focusing
device. So, for example, dispersed phase comes from the center and the continuous phase
from the two sides and a neck formation can happen in that will may the bubble may
generate. So, these are 2 different configurations, there are actually 3. But, we can consider
Y junction to be a variation of T junction where the angle between the 2 channels is not
90 degree, but an angle theta. So, the mechanism of droplet break up or the bubble break
up in different channels is expected to different or it has been observed to be different.
Now, there are a number of factors that may affect the process the droplet may break up.
First of all, that it may depend on the flow rates in the Q of continuous phase and Q of
dispersed phase. It may also depend upon the viscosity of the dispersed phase and viscosity
of the continuous phase. Now, viscosity of the dispersed phase if it is gas bubbles then, the
viscosity of the dispersed phase can be negligible. So, it when gas right; negligible, it may
also depend rho d and rho c, but generally not observed to depend on such parameters it
may also depend on d, 𝑑′ and other geometrical parameters channel geometry. So, we can
say.
Now, let us look at the droplet breakup in these different channels.
145
First we will look at droplet breakup in T junctions and in T junctions, it has been observed
that depending on the velocity of the liquid or the viscosity of liquid different flow regimes
or different mechanisms are observed. For example, so, this can be grouped together in
capillary number, the capillary number is
𝜇𝑈
𝜎
μ U over σ. So, the at low velocities the capillary number is expected to be small for water,
for very highly viscous fluids one might see that the capillary number is large for highly
viscous liquids.
So, there are 3 distinct regimes that has been observed one is squeezing regime, then the
dripping regime and then jetting regime. So, in this squeezing regime it is observed at the
smallest capillary numbers. So, the capillary number less than 0.01 or 10 to the power
minus 2; you might want to clarify what is this μ? μ is μ dispersed or μ continuous. So, the
μ dispersed is not so significant here. It is μ continuous which is used in the definition of
capillary number.
So, at low values of capillary number the squeezing regime is observed; the dripping
regime, it is determined by the shear stress or by the interplay of the shear stress and the
interfacial junction, the balance of these 2 force will determine the size of the bubble. And
146
in the jetting regime, what one gets is a long jet of the 2-continuous fluid. So, the thickness
of jet or thickness of this thread is determined by the viscosity ratio of the 2 fluids.
So, let us look at the squeezing regime the squeezing regime, it can be described in a
number of the steps. So, in the first step, as the liquid enters from the T junction and it
starts forming a neck. slowly this develops because the flow is from left to right in these
cases. And as the droplet enters, it grows in size. Once it grows in size, the necking occurs
or this interface the size of this interface or this distance keep decreasing and eventually
the droplet break up occurs. So, this regime is squeezed.
So, the 4 steps that we have listed here that first, the liquid penetrates the main channel
and then forms a blob. because, the flow of the continuous liquid or the continuous phase
is happening in this direction. So, this deforms or this moves the interface towards the
downstream direction and this block moves to us or downstream direction. Then slowly a
neck is developed and this neck will depend on the on the flow rates of the 2 fluids. and
then eventually this neck elongates the bubble or the blob becomes longer and the neck
becomes thinner and thinner and then it break up breaks up and the droplet or the bubble
is generated.
147
Now, in the dripping regime the process is same, but the one can see that necking does not
occur at the junction itself, but further downstream. So, this is not determined as we will
see that the breaking or the necking process is not because of the flow rates of the 2 liquids
or the 2 fluids, but rather the shear stress balance ok. So, this is a typical picture of the
dripping regime and then.
The third one is jetting regime, so in the jetting regime what one observes that this the
thread the liquid thread becomes longer and longer and eventually. This the necking on
148
the droplet generation happens because of the capillary instability. So, this particular
experiment has had been observed higher capillary number Ca is equal to 0.05 and the
previous 1 at Ca is equal to 0.35.
So, most of the applications where the viscosity of the liquid or the dispersed phase is not
very high. So far of the gaseous for all the applications of the water and so on at low flow
velocity which is generally that is in micro fluidics. The capillary number is small so it is
important or the squeezing regime is the most important flow regime. So, low capillary
number as one can see the capillary number is ratio of μ U over σ or the ratio of μ U by
length scale.
So, μ U over d divided by σ over d. So, it ratio of viscous stresses or viscous forces to
interfacial stresses. and one can see that the at low capillary number the interfacial stresses
will be dominant and the shear stress will be relatively lower or smaller than the interfacial
stresses.
So, it has been observed experimentally that L where L is the length of the droplet or
bubble as the case may be. And so, volume of bubble or volume of the droplet that is
proportional to the L. because, the there will be a constant thickness liquid film
surrounding it as we have seen in Taylor flow examples of Taylor flow.
149
So now, it has been observed experimentally, that the length of this bubble of L by a is
proportional to or equal to 1 plus alpha the ratio of the flow rates of 2 phases. The
volumetric flow rate of discontinuous phase and the volume of flow rate of continuous
phase. The alpha is a constant which is of the order of 1. So, that means, that L by a is
greater than 1. So, the droplet; that means, the droplet is longer than channel diameter.
Now, this relationship has not been observed that it is always true for the entire range of
the ratio of the flow rates. It has been seen that for small values of flow rates or say for
this ratio. When this ratio is 1, L is observed to be a constant.
So, if we plot this picture on a graph, where x axis is the ratio of the 2 flow rates and the y
axis is the non-dimensional droplet or bubble. Then one observe from experiments this
kind of curve where we can have 2 asymptotes. So, at low values less than one observes
that L by alpha or L by a is equal to constant. And at high values of flow rate ratios L by
a is L by a is linearly varying with the ratio of the flow rates ok.
So, now let us look at this in.
150
Further details; so, if we look at the steps of the bubble formation again, in the squeezing
regime or the droplet formation what happens? That the discontinuous or the dispersed
phase enters from here. And because of the flow it will if there is no flow, then it might
grow this way, but because of the flow it starts moving in this direction and then eventually
it envelops the main channel; when it envelops the main channel, one of the things that
happens is that there is always a there is a small opening between the wall and the interface.
Let us say this is small opening as thickness ε and because this restricts the flow of the
continuous phase so the phase are builds up upstream.
Now, this forms this a blob blob this blob is of typical sizes a. Where a, is the size of the
channel. and let us say b is the depth, the channel of rectangular cross section. So, b is the
channel depth and if the dispersed phase flow rate or the discontinuous phase flow rate is
more than the continuous flow rate, then the droplet keeps elongating otherwise it
detaches.
151
So, this has been given that the elongation length the droplet will grow. So, in a case where
this is let us say, this is the neck be or a clear picture of neck will be. So, the droplet growth
rate will be determined by the dispersed phase flow rate, which is and divided by the cross
sectional area. So, the cross-sectional area is a is this dimension and b is the depth of the
channel.
So, the rate at which the droplet growth is Q displacement over a b. and the neck shrinking
rate the rate at which once this neck forms and the let us say that this thickness of the neck
is d and this shrinks with a when the dispersed of an the continuous phase pushes this neck
then the neck shrinking rate is continuous phase flow rate divided by the channel cross
sectional area. So, Q continuous divided by ab now, the time needed to achieve this
squeezing.
When the neck will shrink, when the neck shrinking rate is the velocity or it is say it has
units of meter per second and it also has units of meter per second.
Now, if the continuous phase flow rate is more than the dispersed flow rate dispersed phase
flow rate, then as soon as it develops the neck develops the dispersion happens actually
this neck will never be able to grow and what one will have is a picture like this.
So, this neck will be very thin because the rate at which it grows it is shrinking faster than
that. So, this will be the picture when Q dispersed phase is less than Q continuous. So, this
152
is Q continuous and the flow rate here is Q dispersed. And in this case the neck shrinking
rate which is proportional to PQ continuous is greater than the droplet growth rate. So, the
neck never develops there is never a neck which is enveloping the this mouth of the or the
junction of the channel of T junction.
whereas, for the other case, when we have the dispersed phase flow rate is higher than the
continuous phase, then.
We will see a neck forming which may look something like this. So, in this case, Q
continuous and Q dispersed, but dispersed phase flow rate is more than the continuous
phase flow rate. So, neck will develop and then once this neck has developed reach a steady
state, there will be continuous supply of the liquid or the gas in this blob.
At the same time, because of the continuous phase flow rate the neck will start sinking.
and eventually, when this the time based on which the squeezing time based on the
squeezing time, the length of the droplet will be determines. So, the total length of the
droplet will be when it is or when it is envelop the channel plus the rate of the droplet
growth rate into squeezing time. Now, this is squeezing time is neck width divided by the
Q continuous by a b; which is the neck shrinking rate. So, from this get the relationship L
by a is equal to 1 plus alpha Q This continuous phase divided by the Q continuous phase
ok. and this alpha will be of the order of 1. ok
153
Now, coming to the other regime the dripping regime in the dripping regime, the balance
of 2 forces. So, the viscous drag that is applied by the continuous phase on the dispersed
droplet and the interfacial force that will determine the size of the droplets. So, if we write
that the order of magnitude. So, the viscous drag will be
𝜇𝑈
𝜀
μ U by 𝜀.
So, in this case, as we have seen in the so this ε is the distance between the interface and
the droplet. So, that is the order of magnitude of the dispersed drag μ U by σ and that will
be equal to σ over r. R is the radius of curvature and from this one can have an order of
magnitude of that r is equal to σ ε over μ U. from this relationship, one can calculate the
volume of the bubble.
This; so this is the dripping regime ok we do not have further relationships for the droplet
volume in the jetting regime. Even in the dripping regime, what we have is the scaling law
and the dependence of the of the bubble volume on the velocity or the flow rate of the
phases and the viscosity.
Now, it has been observed that in the squeezing regime, what is important to note here that
in the squeezing regime, the volume of the droplet does not depend anything else except
154
Q of dispersed phase and the Q or the flow rate of the continuous phase. So, it depends
only on the flow rate ratio of the 2 phases.
And that is a very important and significant result, because it does not depend irrespective
of it is gas or liquid it does not depend on the viscosities or either the continuous phase or
the dispersed phase and. So, this flow regime it is possible to achieve such a capillary
number then, it can be used by manipulating the flow rates of the 2 phases. It can be used
to generate the droplets us or the bubbles of a desired size.
Now, at higher flow rates, one can have the effect of viscosity, may be effect of the
viscosity of the disperse phase this especially seen in the jetting regime where we have
discussed rates and so on and so, forth ok. So, that was about the T junctions and one can
develop similar relationships for a Y junction.
Now, for flow focusing devices; so, in the flow focusing device, a typical flow focusing
device that had been taken by taken from article by the Garstecki et al. And the discussion
we have is based around the article or the discussion in the article here.
So, in the flow focusing device, the dispersed phase comes from a channel in the middle
and the 2 from the 2 sides, the liquid phase comes in. So, what happens in this case? that
the this is a particular case, we have the emulsions of the droplet formation is happening
and the polydispersity is less than one percent.
155
So, polydispersity means, the size of the bubbles or the size of the droplet is not dispersed
one get mono dispersed droplets; that means, the size of all the droplets or all the bubbles
is same. This particular case for a gas liquid combination where, the gas is coming at the
center and which is the dispersed phase.
Now, if you look at this configuration, what they have given is the this is the W in and
Wm is varying that is the neck radius or neck diameter one can say. And there is another
parameter on the orifice diameter which is a geometrical parameter and these are the
lengths of the different. So, this is of the flow focusing device is very commonly used to
generate poly disperse the mono dispersed emulsions on the droplets or bubbles of same
size.
Now, there are 3 steps that have been identified in this. The gas enters the orifice and
penetrates into the channel as can be seen here the gas comes in, and it enters the orifice it
squeezes in the orifice there is a film surrounding it of liquid. And then it grows and there
is very thin film surrounding it here. and then as the this grows in here, the neck formation
happens. So, eventually the neck will be become neck will become thinner and thinner as
this grows and then finally, the neck will form.
Now, then when the neck is thinning; finally, at one particular time and the thread collapses
rapidly and breaks. So, the process of thinning of the neck and the then the collapsing in
that gives rise to that droplet break.
156
Now, what had been observed experimentally that the volume of the bubbles. So, this is
for the bubbles and they so one need not worry about viscosity ratios of dispersed and
continuous phase not a parameter right. Because the dispersion continuous phase ratio is
almost 0; so, one do not need to take into account it this into account in the bubble
formation.
So, one see or what they have observed that the experiment from the experiments the
volume of the bubbles is proportional to the pressure of the gas stream. and at this at these
flow rates the pressure or the gas Reynolds number is one calculates then it be in the
laminar flow regime and one can say that this pressure will be proportional to the gas flow
rate from the Hagen poiseuille equation.
So, anyway the volume of bubbles is proportional to the pressure in the gas stream. and it
is inversely proportional to the product of the flow rate so, the Q continuous phase in the
viscosity of the continuous phase. So, it is proportional to the two. it is proportional to the
pressure and inversely proportional to the continuous phase and dispersed phase. Flow rate
it is inversely proportional to the flow rate and the viscosity of the continuous phase which
is liquid. So, that is a relationship or scaling between the bubble volume and the pressure
flow rate and the viscosity.
So, in summary, in this lecture, what we have looked at? Is the mechanism of the bubble
formation in at a T junction and what we observed is there are 3 different droplet breakup
157
regime where different physical phenomena govern; or determine the droplet break up
process. These 3 flow regimes are dripping flow regime, ok the smallest capillary number
what we have is squeezing flow regime, at capillary number. More than 10 to the power
minus 2 one has the dripping flow regime and at further higher capillary numbers one has
the jetting regime.
So, the most relevant and most useful flow regime is the squeezing regime, because bubble
length or the droplet length in the squeezing regime depends on the ratio of the flow rates
of the two phases. in the dripping regime, the bubble size is governed by the shear stress
or the shear stress that develops on the droplet or the bubble that is responsible for the
breakup of the bubbles. In the jetting regime, the gas or the liquid it elongates as a
continuous jet and eventually the breakup happens from the capillary instability or
Rayleigh Plateau instability.
The mechanism of bubbling droplet formation in the flow focusing devices is not so well
understood where, however, what we have seen based on few articles that the droplet
volume is proportional to the gas flow rate or the gas pressure. And it is inversely
proportional to the continuous phase flow rate and the viscosity of the continuous phase.
And it depends on the 2 times scales the time scale the; at which the gas is filled up in the
continuous core or in the blob that grows. and the time scale at which the necking occurs
and the neck break up happens, so.
Thank you.
158
Lecture - 09
Interfaces, Surface Tension and Wetting (Contd...)
Hello. So, in today’s lecture we will be continuing our discussion of Interfaces Surface
Tension and Wetting. In the previous lecture or in the previous discussion on the same
topic, we focused on what interfaces what is the surface tension. And in this lecture and in
the previous lecture, we briefly described about the wetting or the contact angle and contact
line. In this lecture, we will focus more about the wetting the contact line between the 3
phases and the related phenomena.
So, to review we will look back at the Young’s law. So, Young’s law is that when we have
a solid surface, and there is a 3 phase contact on the surface. So, if we have a liquid droplet
and this outside the environment is gas, then the three different surface tension; we have
here is σGL, which is also called simply sigma; this is σsolid liquid, and this is σsolid gas.
So, by using a force balance, we can simply write that σs g is equal to if this angle is θ,
where θ is called static contact angle. in this case the solid surface is assumed to be
homogeneous. Now let us say, chemically homogeneous surface this surface is also
smooth.
159
𝜎𝑆𝐺 = 𝜎𝑆𝐿 + 𝜎𝐺𝐿 𝜃
𝜎𝜃 = 𝜎𝑆𝐺 − 𝜎𝑆𝐿
So, θ is the contact angle; by having the force balance we have σSG is equal to σSL plus
σGL cosθ or we can simply write that σcosθ, we can just get rid of the subscript GL; that
is equal to σSG minus σSL. So, that is Young’s law, that we have already seen and it
combines or it gives the contact angle in terms of the surface tension for the 3 phases.
Now, next we will look at the concept of work of adhesion. So, as you will know; that the
adhesive force or the adhesion refers to the attraction between 2 different phases, or 2
different set of molecules or to 2 different materials. So, here what we are looking at that
work of adhesion is the work required to separate 2 phases of 2 different phases. These 2
different phases might be 2 different liquids or it may be a liquid and the solid phase, and
then we have to separate them or the work required to separate them from one another is
what is called work of adhesion.
So, this is the energy that is released during the process of wetting. So, when you have a
solid surface, and when this solid surface is being wetted by a liquid film. Thus, energy of
the system changes; so the energy that is released during the process is work of adhesion.
And you can also think it as the work that is required to separate 2 bodies. So, you might
160
have seen say for example, in our bathrooms, we generally have a soap sometimes we keep
it on a smooth surface. And the soap when it is wet it gets struck to the solid surface.
And if the surface wet there is a small liquid film that forms between the 2 solid surfaces,
and then it requires significant amount of force to separate the soap from the solid surface.
So, that is a simple example from our day to day life for this adhesive phenomena. Another
example is if you have two glass slides, and a small amount of liquid is trapped between
them we generally need to do a lot of work to separate these 2 slides from each other, and
that is because of capilarity. So, consider that 2 bodies that are in contact with each other.
So, let us say this is body one, and another is body 2 and the area that they are in contact
is A as we have seen here.
So, the surface energy of the contact area, after the contact the surface energy for the
contact area is σ1 2 into A; that is the energy when 2 bodies are in contact. before contact
or after this contact has been removed and the bodies are separate, the energy will be E 1
plus E 2. So, that is σ1 plus σ2 into A. So, σ1 and σ2 are the surface tensions of body 1
and body 2.
And σ1 2 is the surface tension between body one or the fluid 1 and fluid 2, or the solid 1
and fluid 2. As the case may be; so the work of adhesion is the distance between 2, the
161
delta E. So, that means, it will be equal to delta E is equal to the energy, when bodies are
separated which is E 1 plus E 2, minus the energy when the bodies are in contact.
∆𝐸 = (𝐸1 + 𝐸2 ) − 𝐸
= (𝜎1 + 𝜎2 )𝐴 − 𝜎12 𝐴
So, this is the energy released and bodies are in contact. And these body may be a solid
surface and droplet or 2 droplets. So, this is equal to σ1 plus σ2 into A which is where A
is the contact area minus σ1 to A. Remember that the energy for the other surface area to
be same. So, when we get this energy per unit area of work of adhesion per unit what we
get is; σ1 plus σ2 minus σ1 2. So, that is one we had as work of adhesion.
Specifically, for the contact between solid and liquid, Young Dupre he gave an equation.
So, if we apply the same analysis considering that this is valid for a considering the case
of a solid wall and a liquid. Then we can write that this is work of adhesion for this case
is equal to σ1 plus σ2. So, we have σ1 plus σ2 minus σ1 2; where let us see this 1 refers to
liquid phase, and 2 refers to gas phase. So, σ1 is the surface tension, with the gas of liquid
phase, and σ2 is surface tension of solid with the gas phase. So, we have σ1 plus σ2 minus
σ1 2. So now, we have solid and liquid phases in contact so minus σL.
Now from the Young’s law we had just seen the σliquid gas 𝜃 is equal to σSG minus SL.
So, we can substitute this to σSG minus SL from here, and we get work of addition as σLG
162
or σliquid gas plus σliquid gas cosθ. So, that is simply σ1 plus cosθ where we have σis
denoted for σliquid gas if we remove or if we drop the subscripts. And work of adhesion
is the surface tension into 1 plus cosθ. We have let us remind ourselves. θ is the contact
angle between the 2 phases; so for a super hydrophobic surface.
For a super hydrophobic surface, the contact angle will be θ is equal to πor 180 degree for
a super hydrophobic surfaces. So, that means, cosθ is equal to minus 1. So, the work of
adhesion for a super hydrophobic surface is 0 as you can see from here. so, that means,
that no one is required for a drop of water to move it over the hydrophobic surface. So, the
drop of water roles freely over a super hydrophobic surface.
Now, work of cohesion? So, as we discussed earlier that the adhesion refers to the
attractive forces between 2 heterogeneous or 2 different materials. Now if we are talking
about the does a breakup of a liquid volume into droplets, then the work that will be
relevant here will be what we will call the work of cohesion.
So, in that case what we will have that σ1 and σ2 will be equal and σ1 2. So, if you look
at from the work of adhesion, the work of adhesion then we have is equal to σ1 plus σ2
minus σ1 2. If 1 and 2 are same, then what we have is this becomes work of cohesion.
And in this case because one and 2 are the same; so, we can say that σ1 equal to σ2 that is
σminus σ1 2. So, there is no surface tension in the same phase. So, that is 0; so that becomes
163
2 sigma; so, that is why we have work of cohesion is 2 sigma. And that then example for
this will be say and we want to produce droplets from a volume of liquid during the spring
process and number of other examples. So, that is work of cohesion.
Now, we will look at some of the examples of capillary sources. Especially, which are
relevant for microfluidic applications. So, first example in this case is the capillary forces
between 2 parallel plates. So, as we just described that the capillary forces between 2
parallel plates and there are number of real life examples. So, if we draw a picture of this,
let us say you have 2 solid plates which are separated by a distance and the distance
between the plates is h, because of capillarity you might have noticed sometimes that the
shape of this is curved.
So, if you look at this the liquid between the 2 phases you will have a interface something
like this, and solid this is in contact with the solid wall and same. So, we can say that the
radius of this is R, and the contact angle here is θ. it is just the distance between 2 as we
have already said that is h. So, what we need to know is the force required to separate the
2 phases. Now this is the force will be, there will be a pressure difference and to overcome
that pressure difference the phase the force that is required is to overcome the Laplace
pressure. So, this F is equal to delta P into the area.
So, our task is now to find out what is delta P and delta P is σkappa. A kappa is curvature,
we can further write this as σ1 over R 1 plus R 2 which edge as curvature is 1 over radius
164
of curvature and the radius of curvature of the surface is the sum of the 2 principal
curvature. So, for this surface the 2 principal curvatures, one principle curvature is this
one. And the other one is this. So, the one radius is about R so, we will take this as R, now
R 1 is equal to R what remains to be known as what is R 2. So now, R 2 is we want to find.
R 2 is this radius the distance.
From this point to the center of which this curve is apart; so, we know that this is h and
this is h by 2, and this angle is θ then this angle is 90 degree and this is πby 2 minus θ. So,
this angle will also be πby 2 minus θ. So, we can write sin πby 2 minus θ is equal to h by
2 divided by R 2 or cosθ is equal to h by 2 R 2 or R2 is equal to h over 2 cosθ. So, we have
the curvature as h over 2 cosθ.
1 1 1 2𝑐𝑜𝑠𝜃
∆𝑃 = 𝜎 ( − ) = 𝜎( − )
𝑅 ℎ/2𝑐𝑜𝑠𝜃 𝑅 ℎ
1
𝑎𝑠 ℎ ≪ 𝑅 → 𝑐𝑎𝑛 𝑏𝑒 𝑛𝑒𝑔𝑙𝑖𝑔𝑖𝑏𝑙𝑒
𝑅
−2 𝜎𝑐𝑜𝑠𝜃
∆𝑃 =
ℎ
So, we have delta P is equal to σkappa 1 over R, minus 1 over h over 2 cosθ. So, you might
be wondering why we have this as minus.
165
This is minus because they interface is the is concave in shape. So, to take into account
that into account this is minus. So, we have σ1 over R minus 2 cosθ over h and as h is very
small compared to R. So, 1 over R can be neglected. So, we have delta p is equal to minus
2 σcosθ over h ok. And when we multiply this by the cross-sectional area we get the force
required to separates the plates. So, let us just to have a feel of the numbers that what is
the force that will be required to separate the place. we can calculate F is equal to say for
water and air water.
The surface tension is 0.072 or we can write 7.2 into 10 to the power minus 2. In SI units
cosθ let us say take cosθ an average value of about half. So, half and h consider the plate
distance of about 10 microns. So, 10 into 10 to the power minus 6 meters into 3.14 into 10
to the power 10 raised to the power minus 4. So, what we have is F is about 10 to the power
minus 6. This all cancel out; what we have is 2 and 2 also, cancelling out and the. So, the
force in this case is about; the meniscus radius is πR square that is fine and plate distance
of 10 microns we have this.
Value coming among 7.2 into 3.14. So, let us say 7 into 3 about; so, 2.1. 25. So, let us say
this will be about 2.5 Newton. Quite a bit forced to separate quite large force to separate
the 2 plates. So, this is just to get the feeling of the order of magnitude of the forces that is
required to separate the plates. We can see that if the distance between the plates is more
than this force will be. So, if it is 100 micron, then for 100 micron. The force will be 0.25
Newton; so this is the force that is important or that becomes dominant only when the
distance between the plates which is more.
166
Another very popular and very common example of capillary is the capillary rise in a tube.
So, when we have this capillary rise in a tube, some common examples we always see that
the vig of a candle we see we burn the candle, and the liquid wets rises in the candle by
this capillarity phenomenon and then the flame is maintained at the top. Similarly, one of
a very complex a phenomenon that happens is when you lighten a Deepak which is a
lighten a earthen lamp, which in India different languages we would call it a Diya or a
Deepak it has a ghee in it.
And that ghee is in the winter let us say it is wet chilly winter, and then it is in the solid
state. Then the flame can only be maintained when the heat provided by the by the flame
by radiation or convection and convection combined, it is sufficient to melt this ghee so
that it rises in the by the capillarity phenomena.
So, this the burning of lamp or burning of a candle or burning of a Deepak they are a very
classical example of where the capillary rise is a common phenomenon. So, incidentally
this capillary rise is the first phenomena of first well known phenomena of capillarity, in
it was first observed by Leonardo da Vinci in 14th or 15th century. Now he hypothesized,
that the fountains that come out from the mountains, they are also he attributed and this
happens because of the capillary rise phenomena. But of course, as we know now that this
is not true.
167
So, after that several scientists have looked at the phenomena and had tried and tried to
understood. A Francis Hauksbee was the first one who studied this phenomenon in a
systemic manner. And he concluded the following. That the rise of liquid occurs in air as
well as vacuum. So, this might look like the trivial result, but before him one of the
scientists gave a notion that the rise of liquid of the it because in the capillary the air cannot
flow easily.
So, there is a small vacuum to fill that vacuum the liquid rises. So, he showed from the
experiments that the liquid rise occurs in air as well as in vacuum. He also showed that
this typical phenomena, because this phenomenon is easily observed in glass capillaries,
one because glass is transparent another is glass capillaries are very common in our day to
day life or in laboratories.
So, he showed that this phenomenon is not only limited to glass capillaries or not only
limited to cylindrical capillaries. It can also be observed it is also observed the. The liquid
rises we have 2 plates in parallel 2 parallel plates or 2 plates in between plates also the
liquid rise. he also showed that the height of the liquid rise, it does not depend on the
thickness of the channel. So, as your thickness of the tube is one mm or if the thickness of
the tube is 10 mm it does not depend on it. So, these are some of the observations that he
made in 18th century.
168
Now, the law that determines the height of capillarity rise was given by a physiologist an
English physiologist whose name was James Jurin in 1718. And he found that height that
is reached by the liquid. So, the height of the liquid in a capillary, if you have a vessel or
a tube filled with liquid and then it will capillary kept in that.
So the height of the liquid that rises is proportional to the 1 over R or inversely proportional
to the radius of the chain; so that is the first probably the first observation of capillarity
and then about a century later Laplace gain the complete understanding on the phenomena
and capillarity surface tension and so on so forth.
2𝜎 2𝜎 𝑐𝑜𝑠𝜃
𝑝𝐴 = 𝑝𝑎𝑡𝑚 − 𝑅 = 𝑝𝑎𝑡𝑚 −
𝑐𝑢𝑟𝑣 𝑅
𝑝𝐵 − 𝑝𝐴 = 𝜌𝑔ℎ
2𝜎 𝑐𝑜𝑠𝜃
𝜌𝑔ℎ = 𝑝𝑎𝑡𝑚 − 𝑝𝑎𝑡𝑚 +
𝑅
ℎ=
𝜌𝑔𝑅
So, we will look at the rise of capillarity from different angles and try to find the
relationship. So, capillary rise in a liquid if we consider the capillary and consider a point
B here and consider point just below it as point A. And the contact angle is θ. So, the
169
pressure at point A is atmospheric pressure minus 2 σ over radius of curvature of this
interface. So, it is a cylindrical tube and this interface it has a spherical shape, and it will
be a part of a sphere.
So, if we look at this the radius of curvature, then we can basically find the radius of
curvature from here. So, this angle between this and tangent to it this θ then the angle
between the 2 normals the normal to the tube, and the normal to the interface, will also be
θ. So, we will have this drawn here image, this angle is θ this is tube radius R, this is the
radius of curvature.
So, we have cosθ is equal to R over R curvature. So, we can substitute here PA is equal to
p atmosphere minus 2 σcosθ over R. now the pressure at the level of the liquid in this
vessel or in this tub is equal to the atmospheric pressure. So, the difference between the 2
is hydrostatic pressure. So, we can say that PB minus PA is equal to ρgh. So, we substitute
that here, then we have ρgh is equal to pgh atmospheric pressure minus p atmospheric plus
2 σcosθ over R So, these 2 cancel out and we have h is equal to 2 σcosθ over ρgR. So, this
derivation of this expression we have obtained based the arguments for pressure. Now the
same expression can be obtained by assuming the or by considering the force balance.
ℎ=
𝜌𝑔𝑅
170
So, if now we look at the capillary, the force acting this will be in this direction on the
contact line. So, there are 2 forces acting on it. The balance between capillary force, and
weight of liquid column; so if the height of this liquid column is h we have exaggerated
this here to make things clearer, it might not be that concave, so if this height of the liquid
column is h then we have weight πR square h is the volume this liquid into density. So,
that is mass into g. So, that is weight of this liquid column. Now capillary force is equal to
σinto 2 πR we can treat it as a tension in the line, but the direction of this force is not in
the vertical direction.
So, force balance in vertical direction. Let us say this is F capillary. So, that is equal to F
capillary this angle is θ. So, this angle is also θ. So, F capillary cosθ is equal to column
weight πR square h ρg. now please do not confuse that there will be 2 components of this
because the force is acting everywhere on this. So, that combined is F capillary in the
vertical component is F capillary cosθ. So, we substitute this F capillary. Here then we
have σ2 πR is equal to πR square h ρg. πand πcancel out R cancel out, and we have h is
equal to this is cosθ as so, this is cosθ.
So, h is equal to 2 σ2 σcosθ divided by ρgR. So, we can also obtain the height of the liquid
column from the force balance and what we have learnt and then the force in these cases
we want to apply the surface tension force. Σinto the radius not on the radius but the
circumference of the interface.
171
Now, you can also use the principle of energy minimization to obtain the height of
capillary rise. So, when we have a liquid rising in a tube then there are 2 changes in the
energy. The system gains some potential energy, which is equal to half πR square h square
𝜌g. And the system loses capillary energy which is equal to σSG.
So, to remember this when we had there was when there is no liquid, then this is surface
and gas or solid and gas. And when it changes then we have a liquid here. So, the energy
changes from solid gas to solid liquid in this case. So, σSL σSG minus σSL; so, the change
in the energy will be originally the energy was σSG, the surface energy σSG into area of
the cross section, and then it has reduced to σSL. So, the change in the energy is σSG
minus σSL, and this multiplied by 2 𝜋 rh which is the surface area of the wall.
So, the total change in the system energy E is equal to half πby 2 R square h square ρg.
now you might remember that σg minus σSL from the Young’s law. σSG is equal to σcosθ
plus σSL. So, σSG minus σSL is cosθ. So, this is equal to σcosθ. So, this is the gain in the
energy, and this is the σcosθ into 2 πRh. So, this is minus σcosθ 2 πR is capital here. And
the radius of the channel the 2 πRh. Now for minimum energy we have dE by dh is equal
to 0; that means, πby 2 R square 2 h ρg minus σcosθ 2 πR is equal to 0.
So, we will have 2; 2 is cancel out. then we can divide ρby R. So, we will have h is equal
to 2 σcosθ divided by ρg R ok. So, that is the expression of capillary rise in a tube using
the energy minimization.
172
Now, let us just put the numbers here to have an idea about the capillary rise. So, a typical
let us say we take a capillary of 10-micron diameter. So, the radius is 50 microns. oh sorry
um 50 micron or 50 to 10 to minus 6 meters. So, h is equal to for water 2 into we again
take cosθ is equal to half 2 into 7.2 into 10 to the power minus 2 is the surface tension for
air water cosθ.
We have taken to be half, and ρfor air is about 10 to the power 3 let us take g approximately
10 in meter per second square, and cosθ and R is 50 into 10 to the power minus 6. So, this
2 and 2 will cancel out, and what we will end of with 10 to the power minus 6, this becomes
minus 2. So this will cancel out, and we have this about 7.2 divided by 50 meters or 720
divided by 50 centimetre. So, that is about 14 centimetre. So, that is considering the length
of the capillary or considering.
The size of the capillary that is quite a large height, if R is equal to say 1 mm. Then this
number will reduce h is about one tenth of it because this is the thousand microns. So, h is
1.4 centimetre ok. So, that is to give you an idea about the numbers, if this height is let us
say if radius is about 1 micron.
then h for 1-micron capillary is 100 times of this; so, 14 meter of the rise of sap in the
trees; the rise of liquid water in the trees is determined by the capillarity phenomena. And
so, depending on the size of the capillaries that appear in the plants that determine the
maximum height that the tree can achieve.
So, we have looked at 2 examples of capillarity, one is the force that is required to separate
the 2 plates, and other is in the capillarity rise. Now we will look at some effect of surface
in homogeneities.
173
So, at the*t of this lecture, when describing the Young’s law, we said or we assume that
the surface is chemically homogeneous; that means, it has it is made up of only one
chemical material or one chemical substance. There is no dirt there is no inhomoge
inhomogeneities on the surface. And there is, it is smooth. So, there is no roughness on it.
ideally no matter what do we do, we will always have some amount of roughness. And
some amount of chemical inhomogeneities or chemical inhomogeneities caused by the dirt
on the surface of the solid.
So, we need to take into account on of these and see if the contact angle varies. Actually,
it has been seen it has been observed that the contact angle varies on such surfaces. So the
surface in homogeneities both physical surface inhomogeneity in terms of the roughness
and the chemical inhomogeneity in terms of the structure of the material or the different
chemicals that are there on the surface, or the different materials that are there on the
surface both affect the contact angle.
174
So, 2 different laws describe the 2 effects for the first for the physical effect or the
roughness effect, that is given by Wenzel’s law we are not going to derive it here. So, the
roughness, roughness is given Let us say here by r and it is the ratio of actual and the
projected surface area. So, if the material is rough and the actual area will of course, the
more than the projected surface area. So, r is always going to be greater than one, and
Wenzel derived a relationship that cosθ*. So, θ* is the contact angle on a rough surface.
So, he said he derived that cosθ* is equal to R cosθ now if R is equal to one. So, let us
consider the case when material is hydrophilic. So, θ is less than πby 2 and; that means,
cosθ* is less than cosθ. And as θ increase, cosθ will decrease so that means, θ*. So, that a
what does that mean the cosθ* is greater than cosθ.
So, θ* is less than θ. Similarly, one can show that the hydrophobic material the θ* is greater
than θ. So, for an hydrophobic material, θ is greater than πby 2. And cosθ will be negative;
so cosθ* will be negative and if it is multiplied by a value greater than one and it will be
less than cosθ; so, that means, θ will be θ* will be greater than θ.
So, this says that if the material is hydrophilic, it becomes more hydrophilic. And if the
material is hydrophobic it becomes more hydrophobic. So, that is what we have concluded
here that the roughness. it implies it amplifies or enhances the hydrophilic or hydrophobic
character of the contact.
175
𝑐𝑜𝑠𝜃 ∗ = ∑ 𝑓𝑖 𝑐𝑜𝑠𝜃𝑖
Now, coming to the chemical in inhomogenities; so, the material may be made up of a
number of inhomogeneities, but for simplicity let us consider a wall that is constituted of
2 different chemicals. So, the fraction of this the first material is f 1 and second material
is f 2 and of course, if there are only 2 matters and we have F 1 plus f 2 is equal to 1. So,
it can be shown Cassie Baxter law of they have shown some and that cosθ* is equal to f1
cosθ 1 plus f 2 cosθ 2. So, that is the apparent contact angle when the material is
microscopically inhomogeneous if we have a number of materials and we can write this in
general cosθ* is equal to σfi cosθ i.
Now, in this case, we have to assume that the roughness or the size of roughness is small
ok. So this property of the material that by changing the roughness or changing the material
physically or chemically the surface property.
176
Or the contact angle contact properties of the material can be changed. So, different
materials are used in microfluidic applications depending on what applications you are
using. So, for example, say plastic different kind of plastics which are hydrophobic
materials and the glass and metals are hydrophilic materials. And now different techniques
with the recent developments in material science and microfluidics Nano fluidics, different
normal materials are being made, and now they are used to change the properties of the
material on typical example is teflon coating on the frying pans that we have in our
kitchens. So, that can be done by chemically depositing a coating of the hydrophobic
material say if we want to make a hydrophilic surface to be hydro hydrophilic surface to
be hydrophobic and one can deposit a chemical coating on the surface of the hydrophilic
material, with the chemical coating which is hydrophobic in nature.
Or other way one can have a microstructure patterns or micro fillers or grooves on the
surface. For example, all of us have seen of the lotus leaves and seen that the water does
not stick to the lotus Leaf it just rolls through over it. So, lotus leaf is almost super
hydrophobic surface.
So, by inspired by this phenomenon in nature, people have a tried to develop in a surfaces
which are on which can achieve super hydrophobic nature. So, in such cases one need to
take into account they help in lot of research behind that what is the wetting and dewetting
nature on such surfaces.
177
So, we are not going to cover this. So, one of the questions for example, there will be that
can we directly use Wenzel’s law even than we have micro filler of; so, safe in size. So, I
suggest that you can read the book by De Gennes capillarity and wetting phenomena;
where this has been described in detail or digital micro fluidics and micro drops by Berthier
in both the books which has been describes in deatail.
So, the next topic that we come here is contact angle hysteresis. So, if you might remember
say probably in your high school you might have studied about magnetic hysteresis is that
the the phenomena that occurs in one direction does not occur in the same way or it is not
contained in the same way in the other direction.
So, that is what hysteresis generally refer to; so, contact angle hysteresis the typical
example think about a droplet being dropped on a solid surface using a pippet. And initially
what happens as the droplet volume increases as the volume of this droplet is increasing
the contact angle*t increasing, and then that the droplet does not move rather or the contact
angles*t increasing. And the volume of the droplet increases as we can see here and then
it achieves a Constant value which is what we call θ A or θ A is called advancing contact
Angle and if then this droplet is sucked back by the pippet as you can see from here, then
first the angle is reduced, and this angle is called receding contact angle. So, the angle
keeps reducing and then it achieves a value. So, and then that is θ R and this is θ A.
178
So, the static contact angle is between θ A between θ R and θ A ok. And the contact angle
hysteresis is θ a minus θ R.
𝜋𝑅 2 ℎ𝜌𝑔 = 2𝜋𝑅𝜎(𝑐𝑜𝑠𝜃1 − 𝑐𝑜𝑠𝜃2 )
2𝜎(𝑐𝑜𝑠𝜃1 − 𝑐𝑜𝑠𝜃2 )
ℎ=
𝜌𝑔𝑅
The difference between the advancing and receding contact angle is called contact angle
hysteresis. On a good surface which is say are typically smooth and not quite clean, the
hysteresis can be very small say about 5 degree. But it can exceed up to 50 degree on a
rough and dirty surface. So, the surface properties can change the hysteresis. So, hysteresis
can also explain say for example, many times we see that there is a in the capillary, there
is some amount of liquid that is that is remained there residually So, if the contact angle at
this is θ 1 and the contact angle the another direction is θ 2.
Then we can write the balance between the 2 here the height of this height of liquid column
can be obtained by the force balance. So, we can write the force or the weight of this is
equal to 2 or not 2 it is πR square h which is volume into ρinto g where h is the height of
this column, ρis the density g is the gravity and R is the radius of the capillary. And that
will be equal to the force in this direction will be 2 πR σcosθ 1 minus 2 πR σcosθ 2. So,
we can have h is equal to πand R are cancelled. So, here we have h is equal to 2 σcosθ 1
minus cosθ 2 divided by ρgr.
179
And we can see if θ 1 is equal to θ 2 then, h is zero here, the maximum that because θ 1 is
limited by θ R, the receding contact angle, and θ 2 by θ A the advancing contact angle. So,
the h maximum that can be is 2 σcosθ R minus cosθ 2 cosθ A over ρg R ok. So, that is
about the contact angle hysteresis.
Now, in the contact angle hysteresis we discussed about that the contact angle may move
when the droplet is being pushed and the droplet or the liquid has to move. Now all of us
have seen that for example, when you are travelling on a train and it is raining outside.
then the droplets on the train they move and the moving of the droplet or the size of the
droplet will depend on the speed of the train and so on and so forth. And you would have
wondered that what moves these droplets and what controls the motion of this droplet, and
this is governed by the wetting dynamics and the moving contact line.
So, in microfluidics because the size of the channels are small and we have seen the
number of efects when the size becomes small A are proportional to number, number of
capillarity effects they are proportional to 1 over R, and when the size of the channel
becomes a small these effects become dominant ok or so, in microfluidics the wetting
dynamics of the contact line also becomes important.
180
Now this is a multi-scale problem; so, it involves a number of scales the largest scale for
example, if we are talking about a droplet spreading on the surface, then the largest scale
is say the capillary length scale which is about 1 mm. And the minimum scale is the
relevant scale for the interface which is the molecular scale. So, which is of the size of 1
nanometre. So, we can see that this spread is about of 10 decades.
So, in any study it is difficult to resolve both or all the length scales. it is a challenging
task. Now the other problem with the moving contact lines is which is called contact line
singularity or contact line paradox. Now we apply all of us know that the no slip boundary
condition has been found to be valid on the on solid bound.
So, if it is the movement of a solid liquid solid, liquid on a solid or movement of a gas on
a solid we have the no sleep boundary condition which generally say that the fluid that is
it contact with the wall will have the same velocity as the velocity of wall. So, effectively
there is no relative motion between the wall and the fluid over the wall ok. So, that is fine,
now if we have a contact line. So, for example, on a solid surface we have a droplet of
liquid gas environment.
And this droplet is moving in this direction then the contact line moves. So, that means,
there is motion in the near the contact line, then only this fluid can move. So, this is what
the paradox says that the no slip boundary condition is not valid or has a discontinuity
here. So, we need to take this and of course, this is coming because the phenomena that
181
happens here is at the molecular level and the no slip condition what we are describing at
macroscopic phenomena; which is valid in that the continuum scale. So, there are different
models that has been suggested to take into account this contact line singularity.
So, we are not going to discuss those things in detail, and you are suggested to go through
this paper, on moving contact lines in annual review fluid mechanics to understand more
about it.
182
And we will just summarize some of the things here, that when the contact line moves
when we have a static contact angle, then the system is in equilibrium, but then the contact
line is moving. And of course, the system is not in equilibrium anymore and in that static
case because there is no velocity. So, there is no role of viscosity, but the viscosity becomes
important in these cases. And as we move near the contact point the viscous forces that
scale as 𝜇 u over R.
And then R if R is the distance between the contact line, and the solid surface then as we
move towards the contact line this becomes very dominant force. So, this phenomenon
this was studied by huh and scriven by assuming a wedge kind of corner flow. So, he
derived an analytical solution, for a planar liquid vapour interface. And then he showed
that on the free surface there is a viscosity induced pressure on the free surface, and this
pressure can only be balanced by the capillary pressure.
And when we have a capillary pressure, then we need that the interfaces carved, and the
curvature is 0 then the capillary force will be 0. So, that means, that the free surface needs
to be strongly curved close to the contact line. So, his assumption of his analysis was
limited only when the planar was, the interface was planer. But it came out from there that
this solution not full solution because one need to take into account the curvature of the
interface near the contact line.
So, one need to take different scales and into account. Some of the conclusions from the
paper that the motion of contact line can be reduced by the balance between power
generated by the capillary forces. And the energy dissipation that takes place at different
length scales. Most of the time this dissipation is the viscous dissipation and it is spread
over 5 to 6 order of different length scales from molecular from 1 nanometre to one
millimeter length scale.
And the molecular processes they become important only at the cut off to dissipation only
at the few nanometer length scale. Another phenomenon that can be important unrelated
to this is that; when inertia becomes important when the velocity of the contact line is very
high. then then inertial effects become important, and they become to the comparable to
the viscous effects.
So, in summery in today’s lecture what we have studied is different phenomena of contact
line.
183
So, the capillary rise and two plates having a thin film. So, we looked at the to a jump this
where we applied that principles of capillary force to find the liquid rise effect. Then we
also looked at the Wentzel. And Cassie Baxter’s law, for the surface inhomogeneities; it
may be chemical or physical. So, the Wenzel’s law takes into account the physical
inhomogenities of the surface and Cassie Baxter’s law takes in to account the chemical
inhomogeneities of the surface. Then we also looked at the briefly looked at the contact
line when it is moving. So, we looked at moving contact line and the scales into it.
Thank you.
184
Lecture - 10
Gas- Liquid Flow: Void Fraction and Pressure Drop
Hello in this lecture, we will be talking about the void fraction and pressure drop models
for gas liquid flow in micro channels. The two quantities; the void fraction and the pressure
drop or the pressure gradient; in the channel are of engineering importance, they are
required to design any engineering device.
So, it is important to develop correlation and get an estimate to for the pressure drop or
void frack fraction in the channel. In conventional channel or in large diameter channel, a
number of correlations most of them empirical have been developed to measure pressure
drop or calculate or credit or estimate the pressure drop as well as void fractions, in large
channel diameters or large diameter channels.
Now, the first thing that one would like to do is look at the applicability of these
correlations which have been developed for large diameter channels. And then one would
ask one a question that is those correlations, which are working fine for large diameter
channels are they good enough for small diameter channels or any modification to them
are is required. So, this is basically the theme of the lecture that we are going to have today.
185
So, before going into this we will first talk about void fraction and then later in the lecture
we will be talking about the pressure drop. So, a void fraction there are two different void
fractions, that we are talking about here one is 𝛼 and 𝛽. As you see here that a β is
homogeneous void fraction and α is the void fraction beta, if we know the flow rates of
the two phases.
𝑄𝐺
𝛽=
𝑄𝐺 + 𝑄𝐿
Then we will know beta, because β is defined as Q gas over Q gas plus Q liquid so; that
means, a where Q is volumetric flow rate of phase i. Now β is defined as Q G over Q G
plus Q L. So, if we know the flow rates of the two phases, then we know what β is β is
basically the volume fraction of the gas. If we assume that there is no slip between the two
phases if both the phases are moving with the same velocity.
So, then the flow rates at which the two phases are introduced they will be move with the
same velocity unfortunately in most of the cases it is not. So, for example, Taylor bubbles
they move faster than the; average velocity. So, there is a certain amount of slip in Taylor
bubbles, now void fraction measurements is not an easy task to do there are number of
techniques that have been developed to measure void fraction in large channels and more
recently the techniques are being employed or being modified to measure void fraction in
small diameter channel.
So, some of these techniques are say for example, Bao et al., they have used simultaneous
solenoid valve Mishima and Hibiki; they used neutron radiography apart from image
processing. So, of course, image processing is one of the techniques that where one take 2
D image of the flow, and then try to correlate from yet to get the void fraction in the three
dimensional channel ok. Then another method is electrical resistance method.
So, basically different phases will offer different resistance to the current flowing through
it. So, one can measure the voltage and then estimate the resistance and based on that
resistance overtime based on the resistance one can calculate or one can obtain the void
fraction or at any cross section in the channel. similar concept where one measures the
capacitance in place of resistances as electrical capacitors method ok.
186
So now, let us look at the correlations a number of correlations have been developed, and
it will be not possible to cover all the correlations. In this lecture, we will be talking about
those correlations the correlation most of them have been developed for large diameter
channels and then they have been modified for use in small diameter channels.
𝛼 = 𝐶𝛽
So, one of the correlations or one of the class of correlation are Armand type correlation.
So, they are α is equal to sum constants C into β Now if it is a homogeneous model; that
means, there is no slip between the two phases that will be probably true. When we have
very small size of the bubbles and then they follow the fluid flow faithfully then this C is
equal to 1.
So, we have α is equal to 𝛽. Now the next is Armand’s correlation so, in this C is 1 over
1.2 or 0.833, 1 over 1.2 is nothing, but 0.833. So, in some places you will see that C is
equal to 0.833. In other places β over 1.2 ok, now these two correlations or Armand’s
correlation is quite widely used.
So, this is just saying say for example, for slug flow, that the Taylor bubble moves 1.2
times faster than the average velocity. Now, another two correlations that have been listed
here by Chisholm and Spedding in Chen they also use. So, they have C, but C is not a
constant, but C is a function of β and the density ratio of the two phases. So, they take into
account the density ratio and β for both the phases.
187
So, these two correlations one can look at in any standard book on μ.ltiphase flow all in
the literature in different papers.
1
𝛼=
1−𝑥 𝜌 𝜇
1 + ( 𝑥 ) ( 𝜌𝐺 ) (𝜇 𝐿 )
𝐿 𝐺
So, then another class of correlations is Butterworth correlation in which the void fraction
α is defined as one over 1 plus A, where A is a constant into 1 minus x over x raise to the
power P ρ G over ρ L raised to the power q and 𝜇 L over m G raised to the power r. I have
tried to be as accurate as possible checking these correlations again and again; while typing
while, but; however, there might be some errors which I could have overlooked. So, I
suggest that when you use these correlations please verify these correlations with the
literature you can find most of them with the most of them, in the two phase flow boiling
and condensation in conventional and miniature channels by professor Mostafa ghiaasiaan.
𝑚𝑔 0
𝑥=
𝑚𝑔 0 + 𝑚𝑙 0
Most of these most of the discussion, that we are doing here you can find that in the book.
So, for the homogeneous model again if we define A, P, q, r all of them equal to 1, then
one can end up with having a Homogeneous model, which will basically be α is equal to
1 over 1 plus 1 minus x over x 𝜌 G over ρ L into μ L over μ G.
188
So, basically where I am sorry r might be equal to 0. So, this term will go away. So, one
can check and find out that α is equal to β Now x is a new term here. So, that it is not new
for most of us x is quality and as β is volumetric homogeneous void fraction or the
homogeneous volume fraction and x is homogeneous mass fractions. So, as x can be
defined as m dot gas over m dot gas plus m dot liquid where m dot is the mass flow rate
of a phase ok.
So, then in the zivi model they have assumed or they have they have found that A is equal
to 1 p is equal to 1 q is equal to 0.67 and r is equal to 0. So, they do not have any dependence
on viscosity please check this ok. Now the Turner and Wallis model have A is equal to 1
p is equal to 0.72 Q is equal to 0.4 and r is equal to 0.08. Lockhart and Martinelli correlation
A is unlike other three correlations: A is 0.28 here and as you see other two correlation
later also have value of A is equal to 1. So, out of all this only Lockhart and Martinelli
correlation has A is equal to 0.28, p is equal to 0.64 and q is equal to 0.36 r is equal to
0.07.
So, if you look at these correlation all three of them all two of both of them have the value
of p about 0.7 them Thom’s Correlation: p is equal to 1 and q is equal to 0.89, r is equal to
0.18. So, slightly stronger function of viscosity ratios and then Barcozy’s model p is equal
to 0.74 q is equal to 0.65, this q is equal to 0.89 and r is equal to 0.13.
So, there are basically different models and they have been grouped together as
Butterworth type-correlation. So, we have to looked at Armand type correlation which is
relatively simple A is equal to C β form where β is the volumetric flow rate. In this case
volumetric homogeneous volume fraction, whereas; in this case we have α is equal to the
function of x, which is quality the density ratio and viscosity ratio.
You might notice that the density ratio is will be of the order of 10 to the power minus 3
and viscosity ratio of the order of 10 to the power minus 2 or 10 to the power minus 1.
189
So, the third correlation is CISE group correlation. So, it has again a similar to Butterworth
type-correlation, that is; it uses quality 1 minus x over x is one group that they use then ρ
G over ρ L is another group. So, same the density ratios and the ratio of fractions mass
fractions of the two phases. Now, then they have 1 over 1 plus S; where S is slip ratio
which is the ratio of the gas and liquid velocities.
So, most of the equations that we see here are two model this slip ratio between the two
phases. So, S is equal to 1 plus B 1. We have two constants here B 1 and B 2 and y y is
nothing, but it has been defined as 1 minus β over beta, where β is the homogeneous
volume fraction or homogeneous void fraction with some simple considerations one can
always find a relationship between the mass volume fraction or homogeneous mass
fraction or not mass volume fraction mass fraction. So, one can find a relationship between
x homogeneous mass fraction and β which will also involve the density of the two phases.
𝑅𝑒𝐿𝑂 = 𝑙𝑖𝑞𝑢𝑖𝑑 𝑜𝑛𝑙𝑦 𝑟𝑒𝑦𝑛𝑜𝑙𝑑 𝑛𝑢𝑚𝑏𝑒𝑟
𝑑𝜌𝐿 𝑈𝐿 𝑑𝐺𝐿
𝑅𝑒𝐿𝑂 = =
𝜇𝐿 𝜇𝐿
So, I suggest that you should do this as an exercise, if you are already not if you have not
done it earlier ok. So, S the slip ratio as a function of 1 plus B 1 into this term which has
B 2 and y y we have already seen. So, now, what needs to found is B 1 and B 2. So, they
190
have defined B 1 based on the liquid only Reynolds number and the density ratios now Re
L o is liquid only Reynolds number and we will be using this quite often in this lecture.
So, this is defined as Re L o is equal to channel diameter divided by μ L ρ L into u L or
liquid velocity or you can define as d is equal to d into G L over μ L where G is the mass
flux. So, basically Re L o is the Reynolds number.
If only liquid phase would have been flowing in the channel and B 2 is equal to 0.0273
Waber number. So, if you do not remember Waber number let us just remind our self of
that waber number is ρ U square over σ over D, because it is the ratio of inertial forces and
a capillary forces. So, it can be defined in this manner, then we will have the properties of
and the velocities of liquid in this case. So, that is Waber number.
So, if you look at this in this correlation care has been taken to take into account the density
ratio. We can see the density ratio in a number of places it takes into account the viscosity
of liquid, using Reynolds number it also takes into account surface tension using waber
number and it takes into account the flow rate ratio of the two phases or the mass flow rate
ratio of the two phases using x.
So, this is a correlation which takes into account most of the physical effects that might
occur, now surface tension is an important parameter which is there in the micro channels
it also take into account the channel diameter by We L o and Re L o. So, one would expect
that this correlation might work for low Reynolds number or not low Reynolds number,
low diameter small diameter channels and. So, let us look at some of the literature that
people have found. So, in the small diameter channels for example, Bao et al., have
compared their results with CISE correlation for channel size ranging between.
191
About 0.74 mm to 3.74 mm; so, from about 1 mm channel size of that order and then they
found that CISE correlation is suitable for their purpose on the other hand Triplett et al.,
they found that their data correlated well with homogeneous model in 1 mm size channel
except for annular flow regime for which it over predicted; now an exercise has not been
taken up to look at the CISE correlation or at least I am may not aware where an extension
comparison with the CISE correlation has been made with all the literature data for void
fraction. In micro channels, it would be a really good to see if CISE correlation fits all the
data well.
192
Now, Chung and Kawaji and other coworkers they have looked at the effect of channel
diameter on the void fraction and you see an interesting trend in these graphs. So, there are
four channels that they have looked at or that they have been plotted, they have actually
looked at more channels in other articles.
So, if you look at here 530-micron channel diameter 250-micron channel diameter. So,
this is 0.5 mm 0.25 mm 96 about 100-micron channel diameter and 50-micron channel
diameter. All of three channels are circular among these and one channel about 100-micron
size channel is square channel and they found that for the millimeter size channel or.
So, you can look at that y axis is what is plotted α on the x axis is the homogeneous void
fraction of β has been plotted. So, they look at that homogeneous void fraction is good for
0.5 millimeter or 530-micron size channel as the channel size decreases, the ratio between
α and β it starts decreasing.
So, the slope of this line is decreasing and then, that is; why they say that α over β is equal
to 0.833 which is Armand type correlation or which is Armand correlation that predicted
well. At lower correlation they see a different train especially this is very clear at 50-
micron channel diameter that that the ratio between the α and β just small and then it
increases sharply at a high quality or high gas fraction.
193
So, based on their results and these plots have been taken from Ghiassian book based on
the work of Chung and Kawaji. So, they have developed a correlation which is that α is
equal to a constant C 1 β to the power 0.5 divided by 1 minus constant C 2 into β to the
power 0.5 and this costive is sensitive to channel size. So, they looked at for 50-micron
channel C 1 is, sorry; this is 0.02, actually 0.02 and C is equal to 0.98 for 100 micron and
96 micron. Channel C 1 is equal to 0.03, C 2 is equal to 0.97 and in both the case its C 1
plus C 2 is C 1 plus C 2 is equal to 1 ok.
So, this is one correlation that has been developed for flow in micro channels and, but this
does not take into account the size of the channel explicitly rather the; based on their
observations from the experiments. They looked at that there are different values for
different channel sizes; however, the difference between those is not very different as you
can see from here.
Now so, they further developed a that these constants they are sensitive to channel size
and then they developed a correlation C 1 is equal to 0.266 divided by 1 plus 13.6
exponential minus 6.88 D H.
So, where D H is the hydraulic diameter of the channel, which is the channel diameter if
it is DH is equal to 4, into area of cross section divided by perimeter. So, for a circles this
D H is equal to 4 into π R squared over 2 π R RR cancel out 𝜋 π will cancel out and one
will have 2 R. So, is same as the diameter of the channel for a square channel, if D is the
194
channel dimension one will again get D H is equal to 4 into D square, where D is the
dimension of the channel into divided by 4 D. So, 4 and 4 will cancel out and one will get
D H is equal to D, where D is size of the channel for other channels one can similarly find
out the hydraulic diameter.
So, the correlation the same correlation, but in this now C 1 is a function of the hydraulic
diameter of the channel and C 2 is 1 minus C 1.
Now, another correlation that have been developed to look at the effect of channel size is
where α is equal to 1 over 1 plus C into X to the power a. Now, we have this X is not
quality X is capital X here and now we have three unknown parameters here C X and a.
So, X is known as Martinelli or Lockhart Martinelli parameter which is defined as X square
is equal to d p by d z liquid or you can say liquid only divided by d p by d z gas only. So,
d p by d z liquid only or d p by d z gas only are the pressure drops in the channel if there
is only single phase flow.
So, liquid only refers that if the there is only flow of liquid and gas only refers if there is
flow only of gas. So, these are the frictional pressure drops which one can find using
Poiseuilles law for laminar flow for fluid development laminar flow or Blazeou's law for
turbulent flows, which we will be talking about later when we talk about the pressure drop.
So, in this correlation which was for the large channels the Lockhart Martinelli correlation
195
this it had C is equal to 0.28 and a is equal to 0.71 C constant 0.28 and constant a is equal
to 0.71 Turner and Wallis they had C is equal to 1 and a is equal to 0.8.
Now, Kawahara et al., they have modified this correlation for to incorporate the effect of
channel diameter in this. So, they had C is equal to 0.05 1 plus 55 exponential minus 6 D
over D star and D star is reference diameter they took as 250 microns. So, there is this
correlation has been modified again you can say that this was Lockhart Martinelli and
Turner and Wallis correlation which has been modified by Kawahara for implementing in
micro channels.
So, we have looked at that the void fraction measurement in micro channels depends on a
number of parameters or it might be the homogeneous void fraction or homogeneous mass
fraction. Then the density ratios viscosity ratios and the ratio of the two velocity is or the
slip ratio and surface tension and the channel diameter.
So, there are some correlation which we have discussed; in there are some more
correlations which we have not cited here, but people have developed those correlations
and most of those correlations are dependent on the data that has been generated by the
authors are some other data. And then they have compared their correlation with, but they
does not seem to be an universal correlation which can predict the void fraction over all
the channel diameters for all the fluids for all the flow conditions this is largely due to that
if one goes at lower channel diameters. For example, 50 micron or lower one sees that the
196
contact angle effects become important there which none of this take into account.
Moreover, a range of flow rate ratios, there is a range of their there are different flow
regimes that we observe and in each flow regime the arrangement of gas and liquid phases
is different.
So, the void fraction or the slip between the two phases is very different in the flow regime.
So, it is very challenging to develop a universal correlation which can predict all the data
correctly. So, now, let us come back to pressure drop in micro channels and this has been
investigated by several co researchers; in the past 20, 30 years and many of them have
compared the data with the correlation for the large channels and most of them are had
only limited success.
So, the data has been or the correlations have been modified to fit the data and rarely there
is a well agreed correlation which can predict like void fraction. In the same is the case for
pressure drops and there are number of uncertainties for pressure drop for example, when
one measures pressure drop the flow will not be fully developed immediately. So, one
might need to take into account the pressure drop in the developing regime.
So, one should be measuring the pressure drop one once the flow has reached at a fully
developed state and the pressure drop between the those two say; for example, in the test
section one can measure the pressure drop at these two regions, and it is not easy to have
probs in a micro channel at such distances most of the times people measure the pressure
drops and the inlet stream and then at the outlet or the outlet when it is open to the
atmosphere.
So, then one need to take into account the; loses that happens at the mixers or at the walls.
So, there exist uncertainty in the exit and entrance losses and this is more dominant in
micro channels then the wall roughness if you look at the roughness. Ε D is a length scale
for roughness and capital D is the channel diameter and looking at the accuracy of
machining and another manufacturing the 𝜀 D over D will be very small for large channel
diameter.
However, it will be not the case for micro channels. So, the roughness might have a
stronger effect and one might need to take into account. So, that ε D over D can have an
effect channel diameter channel geometry of course, will have any effect on the pressure
drop. Then it is also not possible to immediately extrapolate the correlations that have been
197
developed for large channels, because the flow in large channels is often turbulent where
as the flow in a small diameter or micro channels is of laminar flow. So, it is not possible
to immediately or to use the correlations for large diameter channels as it is for small
diameter channels ok. So if one look at it study state the pressure drop will be a sum of the
accelerational pressure drop the hydrostatic pressure drop.
Because, of the gravity and the frictional pressure drop. So, the accelerational pressure
drop is often important when one has phase change and the area of one phase is changing
with respect to the other. So, it might be boiling or condensation.
So, because, we are talking about here adiabatic flows; so, we will not be talking about the
accelerational pressure drop, hydrostatic pressure drop one can anyway take care by using
ρ G H using appropriate ρ here. Now we will be talking about mostly the frictional pressure
drop d p by d z some other pressured drops might come say; for example, when the there
is an abrupt change in flow area or flow path or pressure drop in control and regulation
devices as we discussed in the previous slide.
198
So, again for the pressure drop the pressure drop for two phase flow. In the correlations
are generally based on the pressure drop for single phase flow and from that the
correlations have been developed and the careful experiments. In last say past 10 to 20
years, it has shown that for 100-micron size channels or millimeter size channels of course,
one can use the correlations that have been developed for conventional channels.
∆𝑃 8𝜇𝑈
=
𝐿 𝑅2
So, for the laminar single phase flow, in a channel those correlations are very well
established for small diameter channels or. So, they can be easily used. So, let us look at
it that for a laminar flow in a channel we have a pressure drop can be defined as ∆ P over
L is equal to 8 μ U over R square. Now the pressure drop in the channel is often defined
in terms of friction factor and there are two friction factors you might remember and it is
confusing most of the time that there are two friction factors one is called Darcy friction
factor and another is fanning friction factor.
So, it the difference between them is the definition that they have the one, we have here
defined as f is equal to this is fanning friction factor and the Darcy friction factor is 4 times
fanning friction factor. So, one need; to be careful that which friction factor one is using.
So, the fanning friction factor for laminar flow is f is equal to 16 by Re, and this can be if
you substitute the ∆ P by L or d p by d z, then one can find out f is equal to 16 by Re and
199
for turbulent flows this is called Blasius correlation and it gives f is equal to 0.079
Reynolds number to the power minus 1 by 4.
So, these are the pressure drops relations for single phase flow.
Now, because as we just talked about that the roughness might be important. So, a
correlation, which has been used by several researchers to predict single phase pressure
drop for laminar transition and turbulent. So, basically all the flow regimes for single phase
flow taking into account the roughness of the channel is called Churchill’s correlation.
And it has a f is equal to 8, this in this there are two constants as you can see A and B plus
C 1. So, C 1 is 8 for circular channels and A is and again function of Reynolds number.
The channel roughness and Ct and Ct is 1 over root C t is 2.457. So, this takes into account
the roughness of channel and be constant B is 37530 over Re whole raise to the power 16.
So, this correlation can be used to obtain friction factor and from there the pressure drop
in rough channels for single phase flow.
200
So like void fraction the simplest thing is looking at homogenous flow model. So,
homogeneous flow model assume that the two phases move with the same velocity, there
is no slip velocity between the two phases the two phases are both well mixed and So, one
need to take into account the only factor that, what is the f there f as we will see f is as we
have seen the f is equal to function of Re.
So, I have just written G here, because f is here. So, f is a function of reynolds number
now one we will need to look at that what is ρ and what is the μ., that one should be using
here G of course, or the flow rate or the velocity one can use the mixture velocity.
201
So, the density there are two definitions of densities one can use that ρ is equal to ρ 1 α 1
plus ρ 2 α 2 or as it has been listed here that sometimes this definition of ρ is used as ρ is
equal to x ρ 1 or ρ gas plus 1 minus x ρ liquid which is harmonic average mass fraction
weighted harmonic average of the densities ok.
Then there are different definitions for viscosities say one given by McAdams is mass
fraction. So, 1 over μ for two phases is equal to x over μ G plus 1 minus x over μ L or
volume fraction weighted average of the viscosities. So, one can have the viscosity to two
phase viscosity is this basically will be, if we do not know α then we can use β here. So,
this will be σ i ρ i β i which is basically ρ 1 β 1 plus 1 minus β 1 ρ 2 and this two is liquid
and β is for gas. So, we have ρ gas.
𝜇 𝑇𝑃 = 𝜌𝐺 + (1 − 𝑥)𝜌𝐿
And mass fraction weighted average of viscosities. So, where μ T P is equal to ρ gas x plus
1 minus x ρ liquid. So, different combinations of volume fraction or mass fraction
weighted average of viscosities and densities have been used to predict the pressure drop
using homogeneous flow model. In my opinion it is appropriate to use if the if the liquid
phase is in contact only liquid phase in contact with the wall, then probably one can use
the viscosity of water and try to see how well does it do part of correlation ok.
202
Now for pressure drop another very famous correlation specially in large diameter
channels is known as Lockhart Martinelli correlation and then further down the constants
in that given by Chisholm so it is chisholm and Lockhart Martinelli correlation and. So, it
uses the concept of two phase multiplier. So, it gives that the pressure drop in two phase.
Is equal to a two phase multiplier, which has been given as phi L O square. So, a liquid
only into d p by d z L O, one can also have d p over d z two phase is equal to phi gas only
squared and d p by d z for gas only. So, there are different correlations for this also.
𝑑𝑝 2
𝑑𝑝
( ) = 𝜙𝐺𝑂 ( )
𝑑𝑧 𝑇𝑃 𝑑𝑧 𝐺𝑂
Now, the Martinelli parameter there is that we have already discussed for void fraction is
the ratio of the pressure gradient, if it is only liquid only flow and press the gradient if it is
gas only flow. So, the x square is the ratio of d p by d z liquid only and d p by d z gas only.
203
So once we have obtained Martinelli parameter then from Martinelli parameter one can
calculate the two phase multiplier and remember this two phase multiplier is required to
obtain the pressure drop in two phase. So, this is 1 plus C over X plus one over X square
now X is Martinelli parameter. So, we know this from here and we already know how to
calculate the d p by d z in liquid and gas cases one can calculate the Reynolds number for
liquid phase for gas phase and obtain and use the appropriate correlation either f is equal
to 16 by Re or f is equal to 0.079 Reynolds to the power minus 1 by 4 the Blasius
correlation.
So, one would have obtain X from there now the value of C. So, Chisholm he gave a table
that for turbulent. So, T here stands for turbulent and L is for laminar flow and for turbulent
flow, the constant C is 20 for liquid laminar flow and gas turbulent flow it is C is equal to
12 and for turbulent liquid flow and laminar gas flow C is 10 and both laminar C is equal
to 5.
Now the modification there has been a modification for micro channels and says that C is
equal to 21 1 minus exponential minus 0.319 D H where D H is in millimeters. So, one
has taken into account the effect of channel diameter into this. So, otherwise what we have
is the from this the effect of the flow rates of the two phases the viscosity of the two phases
and the density of the two phases in these correlations.
204
Now let us just look at that can we apply this Lockhart Martinelli correlation to obtain the
pressure drop for a quality of 0.2 in a capillary of diameter 0.5 mm for air water flow. And
we can take the density of gas is about 2.4 kg per meter cube density of water. Let us take
as 1000 kg per meter cube viscosity of gas is about 1.8 into 10 to the power minus 5 kg
per meter per second and viscosity of liquid is which is water.
So, again 0.001 kg per meter per second ok. So, if we have to calculate the pressure drop
in such case we have given the mass flux and we have been given quality. So, quality is
0.2 and mass flux G is 500 kg per meter square per second. So, we have G of liquid is
equal to 500 into 1 minus 0.2, which is 400 kg per meter square second and G of gas is
equal to 100 kg per meter square second. So, now to obtain two phase pressure drop,
remember the pressure drop that we will obtain using this is only frictional pressure drop.
So, that will be equal to phi.
Let us take liquid only square. So, this is two phase multiplier into d p by d z into liquid
only pressure drop. So, first we need to find out the liquid only pressure drop. So, we can
calculate d p by d z liquid only before that we need to calculate the Reynolds number for
the liquid phase. So, Reynolds number is d channel diameter G liquid over μ liquid. So, if
we substitute the values we will get 0.5; 10 to the power minus 3 into G L is 400 divided
by 10 to the power minus 3 and this will give us a value of about Reynolds number of 200.
So, basically the flow is laminar liquid flow.
205
So, we can say that for laminar flow d p by d z is equal to 8 μ U over R square and when
we substitute the values what we will get is 51200 Pascal per meter G one can obtain or
U, because U is not known. So, U one can obtained by G liquid by ρ liquid you can say
that this G L is k G per meter square per second and k G is k G per meter cube. So, units
will be meter square second ok. So, we have obtained the pressure gradient for the liquid
flow. Now, X square which is the martinelli parameter that is d p by d z liquid only divided
by d p by d z gas only.
So, we need to find out the pressure drop in the gas phase as well. So, we will need to find
that quality Reynolds number of liquid we need to find Reynolds of gas. So, that is again
5 into 10 to the power minus 3, which is channel its not minus 3 its 5 into 10 to the power
minus 4 into 100 which is the gas flow rate divided by μ of gas which is 1.8 into 10 to the
power minus 5.
So, we get the Reynolds number of gas phase is about 2778. So, let us make this calculation
from separated. So, that is the calculation for Reynolds number of liquid phase, this is
Reynolds number of gas phase and from that, because this is more than 2100. So, we will
use the Blasius correlation we will treat at as turbulent flow its not.
So, then we will have d p by d z gas only is equal to 0.0792778 raised to the power minus
1 by 4 and that will give us a value of about 1.8 its not d p by d z it is basically f.
206
Frictional pressure drop so, once we have obtained this fanning friction which comes out
to be about 0.011, then one can calculate d p by d z for gas only phase is equal to 4 f this
is fanning friction factor into G squared over 2 ρ and channel diameter. So, once we
substitute here we get 4 into 0.011 into 10 raise to the power 4, which is 100 squared
divided by 2 into ρ is ρ gas is 2.4. Channel diameter is 0.5 into 10 to the power minus 3
and when we substitute this we get value of 1.8 into 10 to the power 5 Pascal per meter.
So, we have got the pressure drop in the gas phase pressure drop in the liquid phase and
from that we have got X square which is 51 200 Pascal per meter divided by 1.8 into 10 to
the power 5 Pascal per meter and the ratio comes out to be 0.28 and we will get X is equal
to square root of 0.28 which is about 0.53. Now, we need to find out phi L O square or two
phase multiplier one plus C by X plus 1 over X squared. So, we know everything else the
only thing we need to know its C and we have seen that C is equal to 21 into 1 minus
exponential minus 0.319 D H
So, C is equal to 21 into 1 minus 0.319 oh sorry 1 minus exponential minus 0.319 into 0.5.
So, one gets the value of C to be 3.1. Remember this is in mm. So, one gets the value C is
equal to 0.31. So, phi L O square is equal to 1 plus 3.1 divided by 0.53 plus 1 over 0.28
and this comes out to be about 10.4. So, d p by d z for two phase is equal to phi L O square
is 10.4 μ.ltiplied by liquid only dp by dz which is 51200 Pascal per meter or this is about
5.33 into 10 to the power 5 Pascal per meter ok.
So, that is the pressure gradient for two phase flow and that we have used Lockhart
Martinelli correlation here with the correction by mishima and hibiki I guess which is for
the value of C here. So, I would suggest that you should be a doing some of such exercises
to obtain the feeling of numbers and to use these correlations to verify some of the
literature data with this correlation.
207
Another correlation which has been used in large channels is Friedel correlation and there
are two different correlations one is for horizontal and vertical upward flow and the other
one is for vertically downward flow I have not put it down here. So, this expression you
see that the two phase multiplier. So, we are still using the same correlation like this where
we have to have a two phase multiplier. So, this two phase multiplier is being defined by
Friedel which is 1 plus.
𝐼𝑛𝑒𝑟𝑡𝑖𝑎 𝑢2
𝐹𝑟𝑜𝑢𝑑 𝑁𝑢𝑚𝑏𝑒𝑟 = =
𝑔𝑟𝑎𝑣𝑖𝑡𝑦 𝑔𝐷
𝜌𝑈𝑇𝑃 2 𝐷
𝑊𝑒 =
𝜎
Sorry, a plus 3.24 and it is a function of x and 1 minus x, where x is the quality, then the
density ratios viscosity ratios and Froud number and weber number. So, Froud number is
ratio of inertia and gravity. So, that is basically u square over g D and one can use U T P
here and weber number we have already seen. So, that is ρ U Square D over σ ok. And one
can use U T P here ok.
So, in this correlation the while the Lockhart Martinelli correlation or the other correlation
they do not have the effect of their surface tension whereas, the effect of surface tension
has been taken into account in this case and. So, this might be a good idea to check the
208
validity of Friedel’s correlation for micro channels. So, which has been done by some
researchers ok.
So, in general the correlations that has been developed for large channels they are not able
to predict void fraction and pressure drop in micro channels. As it is some modifications
has been suggested; one of the flow regimes that is; most important or that has been used
for almost every application of micro channels is Taylor flow or slug flow region. So, it
makes sense to get the void fraction and pressure drop correlation for this flow regime. So,
we will just talk briefly about that correlation.
209
𝑈𝑇𝑃 𝛽 = 𝑈𝐵 𝛼
Now, for the slug flow regime, one can write a correlation between U T P we have already
seen it for when we discussed slug flow or Taylor flow. So, one can write the relationship
between the bubble velocity U B into α that is equal to U T P into β
𝑈𝑇𝑃 𝛽
𝛼=
𝑈𝐵
So, if the bubble velocity has been measured or bubble velocity has been obtained, then
one can obtain the void fraction the void fraction or some places it is also called; we have
termed it as ε G the void fraction. So, please do not confuse it is same as we have used in
the Taylor flow discussion this void fraction ε G. So, this α is equal to U T P β over U B.
𝑈𝑇𝑃 𝜋𝑅 2 = 𝑈𝐵 𝜋(𝑅 − 𝛿𝐹 )2 + 𝑈𝐹
Now we have also seen a relationship between U B as a function of film thickness that can
be obtained as U T P into π R square is equal to U B into π R minus ∆ F squared plus U
film, which is often neglected into π R square minus π ∆ F square π r square minus π R
minus ∆ F square ok.
210
1.34 𝐶𝑎2/3
𝛿𝐹 =
1 + 3.34 𝐶𝑎2/3
So, one can obtain U B as a function of film thickness. Now we have a number of
correlations for film thickness the prominent or most popular of them being a Bretherton’s
correlations 1.34 Ca to the power 2 by 3 and there is Aussillous and Quere modification
which has been derived from klaseboer. So, this is 1 plus 3 by 4, 1.34 Ca power 2 by 3
divided by 1 plus 3.34 Ca to the power 2 by 3.
So, one can use this correlation for film thickness and from this one can obtain bubble
velocity and once one have optimal bubble velocity one can substitute here of course, one
will know the flow rate of gas and liquid. So, one knows what is U T P and beta? And that
is basically the superficial velocity and gas. So, one obtain what is α . Now a small error
in the film thickness you can make this you can test this for yourself that is small error in
film thickness can have a large error in the velocity and consequently in the void fraction.
then for the pressure drop in the slug flow regime we can divide we take a unit cell which
consists of a gas bubble. So, if we consider a unit cell consisting of one bubble and one
liquid slug, then we have the there are different pressure drops that we need to consider
the pressure drop in the continuous slug phase. So, ∆ P slug, then pressure drop in the
cylindrical region or in the film region because this flow regime occurs in gas liquid and
liquid liquid flow.
211
So, one need to take into or one can make a generic or general correlation which can be
valid for both the cases. So, one can say that that is the pressure drop in the film region
which will be negligible for gas liquid flow then we have bubbles and then the interface
pressure drop. So, the pressure drop and the front and back of the bubble.
So it has been observed that if the slugs are long enough then one can use the single phase
flow.
So, let us remind our self that the slug have a velocity profile which is close to parabolic
velocity profile here, and if, except near the ends of the bubble. So, one can use for long
slugs. So, one can use the hagen poiseuille correlation from where one can obtained that
the pressure drop is 8 μ U L over R square where C is the continues phase U T P is the two
phase velocity some of the costume the two phase is L is the length of the slug.
So, the pressure drop or the in the slug phase is given by this. Now pressure drop in the
cylindrical region one can assume an ideal smooth interface and over flow which we will
be discussing in a separate lecture. So, from that one can calculate the drift pressure drop
in the film region and that comes out to be this for high value of lambda where lambda is
the droplet or bubble viscosity to continuous phase viscosity.
So, this can be negligible for gas liquid phases.
212
And then the third is in the pressure drop at the front and rear end of the bubble so it will
depend on the curvature and the two phase. So, if you consider the point here and point
here the pressure drop at the front the; because the pressure in the gas is almost uniform.
So, the pressure you can say the pressure in gas here and pressure in gas here is same.
Now, you can say the pressure at front and pressure at rear. So, ∆P because of the interface
is ∆ P and the interface is basically P R minus PF and that one can write that PR minus PG
minus PR minus PG minus PG minus PF or PR minus PG plus PG minus PF.
So, that we can give that the pressure drop at the front is two σ or one can have this as σ κ
and the front and this is σ 𝜅at the rear. So, based on this one can calculate the pressure
drop at the front and back and using Bretherton’s model following Bretherton this is the
pressure drop that has been calculated because of the bubble.
213
So, if we sum these two and this has been done by Kreutzer and he has obtained a frictional
pressure drop or the friction factor for two phase flow in the slug flow regime which is 16
by Re 1 plus a D over where D is the channel diameter L is the length of slug and Re over
C a raise to the power 0.33 and this value of a, what he obtained from his correlation was
0.17, and then he compared this with the numerical simulation and he obtain found that
0.07 fit well his numerical simulation whereas, from experimental data it was 0.17.
So, this is one correlation which is used or which can be used for pressure drop in the slug
flow regime.
214
So in summary what we have looked at today is the void fraction and pressure drop
correlations for large channels, and how they have been modified for small diameter
channels. Please note that we have not discussed all the correlations that are available,
there in the literature for large channels.
Neither have we discussed the correlations all the correlations that have been developed
for micro channels or that have been modified; all the modifications are there for micro
channels. We are not put together all the message is that one can modify these correlations
and look at the; what are physical factors which are being taken into account into the small
diameter channels and try to see a universal correlation can be there which seems to be
almost impossible to me.
I would suggest some of you can take up the task of verifying the correlations. Especially;
the once which include all the effect of all the factors such as surface tension and the
channel diameters for example, once that come to my mind that CISE equal to relation for
the volume fraction and Friedel correlation for pressure drop and look at their validity for
a large bunch of data; if that can be those can be good enough for the small diameter
channels ok.
And then we have looked at briefly the pressure drop and void fraction, how it can be
measured? How the Taylor flow regime there has been some descent modifications in the
in the recent paper by klaseboer for pressure drop in the Taylor flow regime. So, I suggest
215
that you look into it and that that generally mostly talk about the pressure drop the
interfacial pressure drop that is there so.
Thank you.
216
Lecture – 11
Liquid Flow Regime & Plug Flow
Hello. So, in today’s lecture we will be looking at the flow regimes that occur in liquid-
liquid flow in micro channels. And in addition we will also look at the; pressure drop in
the plug flow regime or in the plug flow. So, as we have seen for gas-liquid flow there are
different flow regimes that occur in gas-liquid flow in micro channels.
Primarily we have listed as bubble flow Taylor flow or bubbly flow Taylor flow slug
annular flow annular flow annular flow and channel flow. Now the objective of the liquid-
liquid flow in micro channels is quite different in most of the cases or the application of
liquid-liquid flow in micro channels is very different when compared with gas-liquid flow
regimes. The primary objective in most of the applications is to generate mono disperse
droplets in micro channels or to carry out liquid-liquid reactions in very small size droplets.
So, the objective here is to create droplet or plug flow.
So, in most of the cases what has happened that the flow regimes that people have analyzed
or people have looked at is limited by the velocity or the flow rates in which one gets a
droplet or one gets the plug flow. So, most of the cases what one will have is the occurrence
217
of droplet and plug flow. Now when we compare gas-liquid flow and liquid-liquid flow,
then there are few differences that in liquid-liquid flow. We have wetting properties of
both the phases. So, there are more chances of formation of plugs; that means, both the
phases.
Might be in contact with the channel walls then in the gas-liquid flow the viscosity of the
gas phase was negligibly for example; however, in liquid-liquid flow this will not be the
case as the viscosities of both the phases will be non negligible and comparable plus unlike
gas which is a compressible fluid the both the phases in liquid-liquid flow are
incompressible So, as I have said that there are different applications of liquid-liquid flow
in micro channels formation of mono dispersed droplet formation of or production of
emulsions. And the large number of encapsulated biological objects this is also known as
droplet micro fluid people want to have reactions in small droplets which acts like a small
slug reactor and can have efficient mixing, because of the re-circulations in the liquid plugs
or liquid slugs.
So, they have number of application; in addition, say for heat transfer the gas-liquid flow
in Taylor flow regime has been used, why? Because the re-circulations internal re
circulations in the liquid slug they enhance the mixing and there by heat transfer on Nusselt
number; however, in gas-liquid flow the gas bubbles, because they have very low thermal
mass very low mcp value the or rho cp value the cp might be comparable with that of the
liquid, but the rho is about three order.
Of magnitude for gases lower than that of liquid whereas, in liquid both the phases will
have equivalent thermal mass. So, immediately when we use liquid-liquid flow for heat
transfer applications the Nusselt number is bound to increase. So, that is another
application that people have been looking into for liquid-liquid flow in micro channels.
Now unlike gas-liquid flow, the where a number reviews and a number of papers are
available on the flow regimes in micro channels this is not the case. So, so what to the best
of my knowledge there are two reviews available one is by zhao and middleberg where
they have reviewed two phase.
Micro fluidic flows and in that there is a; quite a bit of section devoted on flow regimes in
liquid-liquid flow in micro channels. So, most of the discussion that we will have in this
is based on that paper and the other paper is by Gunther and Jensen in lab and chip by the
218
name multi phase micro fluidics from which the name of this course is inspired this. So,
they have also looked at review of liquid-liquid flows, but very small and short and crisp
review about liquid-liquid flow in micro channels ok.
So, one of the things to understand is that liquid-liquid flow in small diameter channels or
in micro channels is dependent on the structure of the channel or a structure of the channel
inlet so.
Gas-liquid flows they are affected by the inlet condition, but not to the extent that is
observed in liquid-liquid flow, they are very sensitive to the channel inlet conditions. So,
there are number of different devices that have been used for liquid-liquid flow in micro
channels three main are: T-junctions, Flow focusing devices and co flowing devices. So,
T-junctions one can have T-junctions where the flow is happening like this there are two
phases coming from two different streams or another example of T-junctions can be that
one phase and that the two phases come head on and. So, the variation of T-junctions is
also y junction were people have looked at the effect of the angle between the two streams
on the flow regime.
And another is flow focusing devices. So, the difference here is that there is a the dispersed
phase come through the center and the continues phase comes from here and before that
there is a construction. So, the velocity that it has is normal to the velocity of the dispersed
phase flow in the flow focusing devises co flowing devices on the other hand have a similar
219
structure, but in this case the dispersed phase again comes from the center and the
continuous phase comes from the sides.
But, the flow is parallel in this case. So, in here we can say the flow is perpendicular we
can say that continues phase flow is perpendicular to the dispersed phase when phases
come in contact. So, in this place the flow is parallel, but when they come into contact the
flow in the flow focusing device the continuous phase flow become normal to the dispersed
phase flow in the co flowing devises the flow is ok.
So, now let us look at the flow regimes in these three different configurations. So, in T
junctions there are actually three modes of operation in T junctions the first one is what
we call cross flowing. So, we have a T junction; in this case the cross flow. So, the
continuous phase is coming in the straight channel and the dispersed phase comes from
here. So, when the dispersed phase comes from here as the bubble generation or the droplet
generation the droplet generation. In this case will be like this and the droplets will be
moving from here. So, that is for cross flowing in the perpendicular flow where the channel
configuration is same, but the.
Now, the continuous phase comes from the; the branch where as the dispersed phase comes
from the straight channel. So, in this case the dispersed phase will be experiencing a force
or because of the velocity of the continuous phase, and then it will it will break off into
220
droplets the other configuration, where we have T junction, but in different configuration
where we have a T junctions and the liquids.
The two liquids continuously in dispersed phase, they come from two different arms of the
channel and then, they come head on depending on the inertias of the two fluids probably
one of the fluid will enter earlier than the another and then So, this might be the case for
other way round. So, we can have the dispersed phase breaking around the dispersed in
such ok, probably by changing the flow rates of the two phases and depending on the
hydrophobicity of the two liquids one can have one can change the continuous and
dispersed phase in such a configuration ok.
So, now some of the examples from the literature for T junctions. So, in this case what is
happening this is from Tice et al., and there is cross flow, where the dispersed phase
coming from here and the continuous phase straight through it is glycerol water is the
dispersed phase and FC 3283 FO is the is the continuous phase and we can see three
different flow regimes and they have named as and the distinction between the these flow
regime is coming from the capillary number 3, different values of capillary number which
have of course, will varied by changing the velocity of three phases the density and surface
tension and the sorry the viscosity and surface tension for same So, the channel size is 50
by 50 micron square channel and one can absorb the formation of droplets or plugs or So,
those plugs when they say the plug can be of two types where no.
221
Contact with the wall and we will in we will be focusing or we will be saying the plug
more to the; flow regime where it is in contact with the wall. So, in this case we can say
that there is there is a thin film that separates. Now these three different flow regimes; what
is different here? That in such cases the break up in the first flow regime which they have
named as squeezing the breakup happens just at the T junctions. In the second case which
is known as dripping the breakup happens downstream of the T junctions not at the T
junctions itself, and then in third case we see a jet of this liquid coming into and then longer
neck forms in the droplets forms in we can see that the size of the droplet which reducing
has been moved from there to here in the three cases.
Then another work is in due to Thorson et al., and what they have looked at the flow
regimes in a square channel of 60 by 60 and then again it is a cross flow type of
configuration they have looked at in the square channels as well as circular channels. So,
in the square channel for the; now what they have specified is the pressures for the two
phases and these pressure these numbers are in psi. So, water pressure and oil pressure and
one has the droplets forming in this case.
So, depending on the pressure one always get the droplets, but the size of the droplet the
distance between the droplets. The distribution of the droplets changes with the pressures;
in all the cases the oil pressure is same and the water pressure has been increased. So, as
222
we see that water pressure is increase the droplets are becoming more and more close to
each other ok.
Now, the same configuration, but for a sound or circular channel for water oil continuous
flow they have got a very different flow regimes in the circular channel.
And they have drawn a map based on the pressures of the two phases on the x axis, what
is drawn is the average pressure in psi. So, oil pressure plus water pressure divided by two
whereas, on the y axis the difference between water and oil pressure divided by the average
pressure. So, basically the pressure difference here has been plotted. So, when oil pressure
is very high when one see or when oil pressure is greater than the water pressure and one
see only the flow of oil no water when as it is increased then we see three different flow
regimes here H, B and C.
So, if we identify this H, B and C and depending on the pressures and the average pressure
values one can see that; there is formation of droplets and as the size of or as the pressure
is increased. The average pressure is increased size of this droplet which seem to increase,
the water pressure in all three cases is am difference of the two pressure which negative
so; that means, water pressure is lower than the oil pressure.
Now, as the water pressure is increased and they are almost equal when we see two
different configurations depending on the average pressure G, J, A, K, E and D. So, if we
223
look at A and J where they have droplets and they are forming a neckless kind of a structure
J and A and then K again and then you have sometimes two droplets and interesting a
structure that two droplets horizontally followed by two droplets vertically to get that again
a very kind of regular structure here and then E, where we have long droplets connected
together and then E again the droplets of different size.
So, there are different structures very small change in the pressure can change the flow
structure, but what we have essentially is the droplets 1 or 2 droplets and it can be long
droplets higher pressure at higher pressure we observe; when they are both of them are
equal and they are non droplets, but just a small decrease or slight tilt over water pressure
and one see a spherical droplets. So, a number of patterns one can see here that solid water
stream or solid streams should not solid, but a continuous stream what you should say.
And then elongated droplets triple droplet layer. So, we can see that higher water pressure
where water pressure has increased now in this case we have F I and L. So, here we have
continuously two droplets with each other in the L configuration in the triple droplets in
the F configuration. So, water pressure is higher in this case So, depending on the pressure
of the two phases interesting patterns have been formed in this case ok.
Now, this is an example for perpendicular flow. So, where the dispersed phase is coming
from the straight channel and the and the continuous phase.
224
Is coming from the channel normal to it and then they have different snapshots the image
has been taken from Xu et al., has been described in Zhao and middleberg paper. So, one
can see the two different configurations the droplet flow or in this case the circular droplet
and here; the long droplet. So, we have droplet spherical as well as Taylor droplet which
look like which has a shape like Taylor bubble.
And. So, this is for recently by Yagodnitisyna in chemical engineering journal they have
published their findings and they have conducted the experiments in 200 by 200-micron
channel and use three different liquid combinations kerosene water and different other oils.
Now they report different flow regimes they have done over a range of flow rates and
different liquid. So, they So, that they have observed parallel flow then plug flow. So, this
plug flow you can see that there is no film surrounding the wall both phases are in contact
with the water and now that is bound to happen when we go at smaller channel size,
because at the smaller channel size what is going to happen the contact angle will play a
major role.
And. So, the and the and the hydrophilicity and hydrophobicity of the channel will also
play a role. So, one can have a plug flow there they have termed the base flow regime and
slug flow which is different from the notion that they have about slug flow, you may also
notice the configuration that they have is had on flow that two flow regimes comes from
the two branches of T-junctions. And then they match at the centre this the channel size is
225
200 by 400 micron. And these two channels are 200 by 200 microns and they also get
rivulet flow.
So, one do not get a continuous phase and where as one has a small stream of one phase
and then the on the other side you have the other phases. So, this is from the wall. So, this
is called the rivulet flow and then they have an interesting flow pattern which is called
serpentine flow So, the configuration in 3D; that they have described in that which the
there are very structured or very regular continuous ways on the two phases which one
might relate with Kelvin Helmholtz instability.
So, Kelvin Helmholtz instability in generally seen when there are two different fluid
streams with velocities different velocities U 1 and U 2. Originally, it was observed for at
large derived for the two fluid streams which are flowing parallely and the two steams are
inviscid when the flow streams are inviscid there is no profile in the two and then miles
looked at the effect of the velocity profiles on the instabilities. So, the instability causes
the; the way from the interface which was originally have which initially had a smooth
interface and then it was perturbed ok.
So, this looks very much like the serpentine flow; what they describe? This might be
caused by Kelvin Helmholtz instability its might be playing a role there
226
So, then coming to the next configuration the flow focusing device in this case the disperse
phase as we said the disperse phase come from the center and the continuous phase comes
from the two sides of the channel.
And in this case what is observed that? It generally generates droplets rather than plugs.
So, in the T-junctions we generally see that one of the phases for example, in this case the
cross through one of the phases in contact with the wall. So, the wall contact angle is bound
to play a role in the generation of the droplets or plugs whereas, in this case the separation
happens either here or afterwards. So, this is the contact with the channel walls can be
avoided in this case and the.
Co flow devices also.
So, one observe two flow regimes have been observed, in this one is called dripping flow
regime which is observed at low flow rate. Another one is called jetting regime which is
observed at higher flow rate.
So, in the dripping flow regime one observes the; at low flow rates or at the low velocities
the droplets of uniform size mono disperse droplets are being generated. When the flow
rate becomes higher then there are droplets of different sizes inertia has become higher
and inertia might play a role when these droplets break off from the jet So, in the first case
of one can get regular mono disperse droplets which are spherical in say
227
Now, this kind of a structure where a continuous jet is being broken into droplets as we
have seen.
We can a relate it with the say for example, we have all of us have taps are in our kitchens
in our bathrooms and then we see that there is faucet coming out of the tap, and then it
disintegrates into a smaller droplets and that generally attributed to what we call capillary
instability or Plateau-Rayleigh or Rayleigh plateau instability. So, in this case what the
instability says that a long cylinder a static long cylinder of liquid it breaks up into small
droplets and then the lambda at which this breaks up is equal to the 2 π r of the of the jet.
So, the same instability capillary instability might be playing a role in this regime, but it
will also have the effect of viscosity effect of inertia playing a role on this disintegration
the similar phenomena also happens in an ink jet printer; where one need to have a droplets
generating of course, there are or active actuators also which control the droplets
generation in the ink jet printers ok.
228
Now, next is coflowing device.
So, as we discussed earlier in the coflowing device the flow happens in the same direction
both the continuous and the disperse phases and So, in this case the disperse phase is
introduced by a capillary or by a needle in the continuous phase and like flow focusing
devices one observed dripping and jetting regimes. In this case also and there are different
one can see that sometimes one can droplets of two different sizes at the droplets are of
mono dispersed droplets. So, this transition between jetting and dripping regime they have
attributed to the capillary number which is basically inter play of
𝜇𝑈
𝜎
So, the discussed forces and the intra phased forces.
So, in all the cases in fact, the T-junctions coflowing and flow focusing devices and any
other focusing devices the or any other mixing devices for liquid-liquid flow gas-liquid
flow the capillary number is the mazor factor the interplay of surface tension or the
interfacial and the and the viscous forces is an important factor. So, they have suggested
then when capillary and Weber number are small.
So, when velocity is small the dripping occurs as we have seen for the flow focusing
devices and then jetting occurs, when either of that increases and then they have given a
229
criteria when capillary plus Weber number is of the order of one then one observes the,
then when observes the jetting regime say if we look at this the capillary number is the
ratio of viscous effects divided by interfacial effects and Weber number is ratio of inertial
effects and interfacial effects. So, when this becomes order of one.
So, that what they suggest that when the viscous forces and inertial forces they are
equivalent to the interfacial forces than the jetting regime takes over in the smaller regime
one would have or at the lower velocities the inertia will be negligible. So, and that
interfacial forces will be dominating then over the viscosity forces.
Then So, we have what we have looked at is the different flow patterns that may occur or
that has been observed in the literature for flow focusing devices or coflowing devices or
T-junctions. Now there is another work by Dreyfus et al., where they have used a flow
focusing device a configuration like this the tetradecane and water tetradecane is the
disperse phase which is coming from the center.
And water from the two sides and it is in 200 by 200 this is micron 200 by 200-micron
channel. And they observed different flow regimes isolated droplets pearl necklaces
stratified or parallel flow and pear like drop. So, we have isolated droplet seen here
stratified flow which is as there the what is termed as parallel flow that one phase is coming
at the centre and the other two phase is on the sides of it and then with they have pearl
necklace type that; there are two droplets which are connected by smaller shapes hare then
230
they also have pear shape droplet different interesting shapes and this has been done when
the channel is wetted fully wetted.
Now, the other case when they have had a that they observed that if the phases are
interchanged, then the same regimes are not observed and then they say that it will depend
then the system of course, this we have already said that they said that the from the flow
regimes it depends on the way fluids are injected. So, the system does not forget its initial
condition. So, it is difficult for the system to forget its initial condition probably because
the viscosity of the fluid is high.
231
So, what they have seen in the previous case that was for fully wetting regime. So, it was
hydrophilic. Now it is partially wetting. So, hydrophobic channels and then they see that
they have observed of that well defined structure as was observed in the previous case are
not observed they have observed some stratified flow, but not smooth one some
corrugations are there on the interface and then they have droplets of irregular shapes, as
one can see here they are observed. So, basically if it sensitive to the way flow rates are
varied and enough stress, what they have suggest that enough stress is not generated.
So, that the fluid advects the capillary number in these cases was of the order of 10 -3. So,
what they have I mean, what the objective of this? that the hydrophobicity of the channel
can play an important role on determining what flow regimes are being generated
everything else being the same
232
So, Gunther and Jensen in their work they have put together the flow regimes from a
different researches in different channel sizes. And they observed that the segmented
isolated drops or the segmented which is Taylor kind of flow or stratified or parallel flow
and pears like safe F, G have been observed and this flow regime if they have drawn in for
disperse phase and continuous phase as the coordinates of the graph.
Now, the next topic in this.
233
So, as we have just seen that depending on the hydrophobicity hydrophilicity of the
channel one can have that both the phases are in contact with the channel wall. So, our
general notion of slug flow is that there is a long droplet like, Taylor bubble shape they
are moving in the channel, but at lower channel diameters one may observe or one as on
has observed some of the researches may observed that, there are plugs of different phases.
So, you can have phase one and then another plug of phase 2 and then plug of phase three
sorry plug of again phase 1 and then plug of phase 2. So, we can color it differently to
distinguish the two phases and in such a case the contract angle phase going to play an
important role and, because the interface is moving. So, the contact angle is not going to
be static, but it will be a dynamic contact angle.
So, Hoffman proposed an expression for dynamic contact angle, but which was simplified
by Voinov and Tanner; which says that the dynamic contact angle can be given as a
function of static contact angle and capillary number and a constant A which has a value
of 94, if the angles are in radian. So, Ca we already know is
𝜇𝑈
𝜎
and 𝜃d is dynamic contact angle and θs is static contact angle.
𝜃𝑑3 = 𝜃𝑠3 + 𝐴𝐶𝑎
234
𝜃𝑑 = (𝜃𝑠3 + 𝐴𝐶𝑎)1/3
𝐴𝐶𝑎 1/3
= 𝜃𝑠 [1 + ]
𝜃𝑠3
Now, this law which is θ d cube minus θ s cube is equal to a Ca, if Ca is small which it
generally the case of the order of 10-3; especially for low viscosity fluids then we can
linearize it.
So, what we can write is that θ d cube is equal to θ s cube plus A Ca or that can be written
as θ d is equal to θ s cube plus A Ca raise to the power 1 by 3; then we can write this as θ
s into 1 plus A Ca over θ s cube raise to the power 1 by 3. Now we can approximate for
low Ca we can suggest that A Ca over θ s cube will be small. So, we can write this as θ s
into 1 plus 1 by 3 ACa over θ s cube or we can have θ d is equal to θ s plus 1 by 3 ACa
over θ s squared. So, that is the linearized version of now one need to take into account the
sign of the velocity and as we have seen or as we know that Ca is equal to μ u over σ.
So, for advancing contact angle what will have a θ d for the advancing case and U is
positive. So, one will have θ s plus 1 by 3 A Ca over θ s square we can put this in terms of
θ s advancing. So, this is static advancing contact angle and this is dynamic advancing
contact angle. Similarly, when U is negative for the receding case one will have θ r or θ d
receding is equal to θ s receding minus 1 by 3 ACa over θ s receding squared. So,
immediately one can see that θ d advancing is greater than θ d receding from this
expression is of course, θ d s advancing is greater than θ s receiving sufficiently that is ok.
235
So, based on this these correlations now, we will try to look at the pressure drop when we
have plugs in Microchannels.
8𝜇1 𝑈𝐿1 8𝜇2 𝑈𝐿2

∆𝑃𝑓𝑟𝑖𝑐𝑡𝑖𝑜𝑛𝑎𝑙 = +
𝑅2 𝑅2
So, if we have a picture of the plug that we just described that was considered one in itself
we have the plugs of two phases 1 and 2. So, this total pressure drop will be the sum of
frictional pressure drop plus the pressure drop caused by the capillary force or the capillary
action near the wall or by the contact angles. So, the frictional pressure drop is if we assume
the flow to the laminar and use Poiseuille solution then we have delta P friction is equal to
8 μ 1 L 1 by R square plus 8 μ 2 L 2 by R square. So, then you will also have U where U
is the velocity of the plug and the both the plugs will have to have the same velocity that
steady state.
So, that is the frictional pressure drop now let us turn our attention to the capillary pressure
drop.
236
2𝜎
∆𝑃 =
𝑅𝑐𝑢𝑟𝑣𝑎𝑡𝑢𝑟𝑒
So, for the capillary pressure drop if we analyze one plug then we will have the interfacial
pressure drop say that the two junctions P 1 and P 0. So, what we will have? The pressure
difference and the front is equal to 2 𝜎 over R curvature and if we say that this is θ
advancing contact angle.
𝜋 𝑅
𝑠𝑖𝑛(𝜃𝑑𝐴 − ) =
2 𝑅𝑐𝑢𝑟𝑣
−𝑅
𝑐𝑜𝑠𝜃𝑑𝐴 =
𝑅𝑐𝑢𝑟𝑣
2𝜎𝑐𝑜𝑠𝜃𝐴
(∆𝑃𝑐𝑎𝑝 )𝑓𝑟𝑜𝑛𝑡 = −
𝑅
Then we can draw a geometrical configuration this is the radius of the channel the angle
being θ and the radius of curvature is this and the term is R curvature and this angle is 90
degree. So, the angle here is θ advancing minus 𝜋 by 2 and this angle will then be θ d
advancing minus π by 2. So, what we can write is that sign θ d A minus π by 2 is equal to
R over R curvature. So, that will give us minus So, we will have this as cost θ d advancing
is equal to minus R by R curvature. So, we can substitute that from here that delta P cap is
237
equal to minus 2 σ cos θ d advancing divided by R similarly. So, that is at the front and
the similar expression for be written for can be written for the back.
2𝜎𝑐𝑜𝑠𝜃𝐴
(∆𝑃𝑐𝑎𝑝 )𝐵𝑎𝑐𝑘 =
𝑅
2𝜎
∆𝑃𝑐𝑎𝑝 = (𝑐𝑜𝑠𝜃𝑑𝑅 − 𝑐𝑜𝑠𝜃𝑑𝐴 )
𝑅
1 𝐴𝐶𝑎
𝜃𝑑𝐴 = 𝜃𝑆𝐴 + 2
3 𝜃𝑆,𝐴
So, one can have that delta p cap and back of the plug is equal to 2 σ cos θ d receding
divided by R and that will be the minus sign will be because of the concavities. So, we can
have the total pressure drop delta p at cap, because of the capillary force is the sum of the
two. So, we will have 2 σ over R cos θ dynamic contact angle receding minus cos θ
dynamic contact angle which is for advancing So, now we can recall what we had for θ θ
dynamic for advancing we had an expression that θ static advancing plus 1 by 3 ACa over
θ S a square as we take cos of this say we substitute this here then we will have cos θ S a
plus 1 by 3 A Ca over θ S a square. So, that will be cos a plus b. So, cos a cos b minus sin
a sin b.
1 𝐴𝐶𝑎 1 𝐴𝐶𝑎 1 𝐴𝐶𝑎

cos (𝜃𝑆𝐴 + 2 ) = cos 𝜃𝑆𝐴 cos ( 2 ) − sin 𝜃𝑆𝐴 sin ( 2 )
3 𝜃𝑆,𝐴 3 𝜃𝑆,𝐴 3 𝜃𝑆,𝐴
238
𝐴𝐶𝑎 1 𝐴𝐶𝑎
2 <1 small → 𝑐𝑜𝑠 (3 𝜃2 ) ≈ 1
𝜃𝑆,𝐴 𝑆,𝐴
1 𝐴𝐶𝑎
𝑠𝑖𝑛 ( 2 ) ≈1
3 𝜃𝑆,𝐴
So, cos θ S a cos 1 by 3 A Ca over θ S a squared minus sign θ S a into sin of 1 by 3 A Ca

over θ S a square. So, if we have assumed in when while linearizing this that A Ca over θ
S a squared is small. So, we can say that cos 1 by 3 A Ca over θ S a squared is about 1 cos
0. We know that it is 1 and at low value sin θ is equal to θ. So, we can say that sin of 1 by
3 A Ca over θ S a square that is equal to about 1 by 3 A Ca over θ S a square. So, we can
write that as θ from here.
2𝜎 𝐴𝐶𝑎 2𝜎 𝑠𝑖𝑛𝜃𝑆𝑅 𝑠𝑖𝑛𝜃𝑆𝐴

∆𝑃𝑐𝑎𝑝 = (𝑐𝑜𝑠𝜃𝑑𝑅 − 𝑐𝑜𝑠𝜃𝑑𝐴 ) + ( 2 + 2 )
𝑅 3 𝑅 𝜃𝑆,𝑅 𝜃𝑆,𝐴
We can write that delta P capillary is equal to 2 σ over R the same substitution can be done
for the receiving angle with the minus sign there and we will get 2 σ over R cos θ S R.
So, static angle receding minus cos θ static angle advancing plus A Ca by 3, 2 σ over R
into sin θ S R divided by θ S R square plus sin θ S a divided by θ S a square.
239
𝜎𝑆,𝐴 = 𝜎𝑆,𝑅
2𝐴 𝑠𝑖𝑛𝜃𝑆𝑅 𝑠𝑖𝑛𝜃𝑆𝐴
∆𝑃𝑐𝑎𝑝 = (𝐶𝑎. 𝜎) [ 2 + 2 ]
3𝑅 𝜃𝑆,𝑅 𝜃𝑆,𝐴
𝐶𝑎𝜎 = 𝜇𝑈
2𝐴 𝑠𝑖𝑛𝜃 𝑠𝑖𝑛𝜃
(𝐶𝑎. 𝜎) [ 2 𝑆𝑅 + 2 𝑆𝐴 ]
∆𝑃𝑐𝑎𝑝 3𝑅 𝜃𝑆,𝑅 𝜃𝑆,𝐴
=
∆𝑃𝑓𝑟𝑖𝑐𝑡 8𝜇𝑈𝐿
2( 2 )
𝑅
∆𝑃𝑐𝑎𝑝 𝐴𝑅𝑠𝑖𝑛𝜃𝑆
=
∆𝑃𝑓𝑟𝑖𝑐𝑡 12𝐿𝜃𝑆2
And now what we can have is if there is static no static hysteresis. So, that means, that θ
S a is equal to θ S R. So, the first term in this will go away this term may be 0. So, what
we will have as delta P is equal to 2 A over 3 R Ca into σ into sin θ S R over θ S R squared
plus sin θ S a divided by θ S a square. Now one can have this the ratio of just to compare
the two force phase of the delta P capillary divided by delta P the frictional drag. So, this
Ca into σ you can see that Ca in σ is μ into U. So, and we assume that μ 1 is equal to μ 2
for simplicity and L 1 is equal to L 2. So, we can write this as the frictional pressure drop
and will be eight μ L U by R square multiplied by 2.
240
And this will be 2 A over 3 R μ 1 U into sin θ S R by θ S R square plus sin θ S a divided
by θ S a square and if θ S R and S R a equal. So, one will have this substituted by 2 sin θ
S divided by θ S square. So, μ 1 and μ 2 are same. So, we will have this μ to be cancelled
out 1 of R will cancel out and we will have this as 8 into 2. So, this 2 into 2 will canceled
out and we will have 4 into 3, 12 L θ S square at top we will have A R into sin θ S. So,
that will be the ratio of delta P capillary over delta P friction. So, one can see from here
that if L is smaller than delta P capillary can be significant.
So, if θ S is equal to let us say π by 2, then this number will be A R by 12 L π by 2 square

and then one can where A is 94. So, one can estimate what this number is and one will
have that these forces are quite equivalent and the ratio of L by R. Similarly, one can have
a magnitude when considering when there is static hysteresis. So, the idea here is that one
that how to analyze the effect of the wall contact angles by the simple of Voinov tanner
model and then appreciate the fact that at low R the capillary force might be important
when compared with the frictional forces and that is what we see in smaller channels ok.
So, in summary in today’s lecture what we have looked at the different flow regimes that
has been observed in the literature in different inlet configurations T-junctions
configuration. Flow focusing devices coflowing devices people have observed the parallel
flow mainly the two flow regimes that has been observed at droplet flow and the plug flow
or Taylor droplet flow and then people have also observed the parallel flow or what they
call is stratified flow and then pearled flow pearl necklace flow and the. So, which are
generally the droplets which are interconnected with each other with bridges around them.
So, different interesting observation of are there then we have looked at the plug flow in
micro channels where we analyze the pressure drop when there is no film that separates
the droplets from the channel wall and the pressure drop.
In such cases and appreciating the effect of capillary pressure drop. Now it has been
observed in some cases that at low channel diameters people also get similar kind of plug
flow in gas-liquid flow also. So, the same can be same analysis can be applied for gas-
liquid flow plus when one have say triangular channel or rectangular channel, then the
Taylor droplet may not be it may not have a film everywhere there might be a contact
angle or there might be a contact with the wall of the channel at certain places especially
at the centre of the channel. So, in such cases also one need to take into account capillary
forces when analyzing the pressure droplets and channels. So, that is all from this lecture.
241
Thank you.
242
Multiphase Micro fluidics
Prof. Raghvendra Gupta
Lecture – 12
Smooth Interface Laminar-Laminar Annular Flow
Hello, in today’s lecture, we will be looking at annular flow. In particular, we will look at
an ideal case of annular flow, where the interface is smooth and there is no, when the
interface is smooth and the film thickness or the size of the core, or the radius of the core
is same throughout the channel. Ok?
So, let us draw a configuration of an ideal annular flow. We have a channel. Let us say
this is a cylindrical channel and in this cylindrical channel, the interface is located
somewhere here. So, this is phase 1. Let us consider gas-liquid annular flow.
243
So, gas is at the core and liquid is on the sides. Ok? And the radius of the channel is capital
R and the radius of the interface, let us name it as Ri and we have properties of liquid 𝜌 L
and Ρ g and 𝜇 L and μ g are the density and viscosities of gas and liquid. Now, this
configuration, when we look at the flow regimes in gas-liquid flow in micro channels, we
often encounter annular flow, but this is rarely the annular flow that we see in gas-liquid
flow, in micro channels is rarely a smooth interface annular flow.
We generally see the presence of waves on this annular flow, but to understand this, let us
consider a case, where for which we can obtain an analytical solution and can get
relationship, say for example, between pressure drop and film thickness or interfacial
radius and the flow rate of the phases. So, this is the case that we are going to considering
this lecture.
There can be annular flow or what is termed as in general, core annular flow for in the
reference of liquid-liquid flow. So, there depending on the viscosities and densities of the
two fluids plus hydrophobicity or hydrophilicity of the two liquids. One phase will be at
the core and another phase will be these properties will determine which phase will be at
the core and which phase will be near the wall. Ok?
So, we are looking at this. We will use the terms for ρ L and ρ g or μ L and μ g; however,
the same terms can be used with reference to liquid 1 and liquid 2. Ok? So, we will solve
the mass conservation and momentum conservation equations and because this is a
244
cylindrical configuration, so we will prefer cylindrical coordinate system. So, where we
have r, 𝜃 and z? So, z is the direction, r the radial direction and when we say that the flow
is axisymmetric; that means, there is no flow in the angular direction and no gradients in
the angular for azimuthal direction.
So; that means, v θ is equal to 0 and del over, del θ is 0. So, this is the symmetric
∂
configuration, axisymmetric configuration at steady state; that means, the ∂t term is equal
to 0 steady, means as you will know that steady means the properties, the things, the
configuration does not change with time. So, all the variables with respect to time are
∂
constant. So, that is why ∂t term in the conservation equations will be 0, flow is laminar
and we do not need any models for turbulence and we consider fully developed flow here.
So, when the flow is fully developed, we can say that del v z over del z is equal to 0; that
means, the velocity profile, whatever the velocity profile be obtained at one axial location,
a similar velocity profile will be obtained at another location. So, the corresponding
difference delta v will be 0 or delta v z will be 0. So, the flow is fully developed or that is
the definition of a fully developed flow. Now, let us write the conservation equations. So,
coming to the continuity equation, let us write a continuity equation in the cylindrical
coordinates.
𝜕𝜌 1 𝜕(𝜌𝑟𝑣𝑟 ) 1 𝜕(𝜌𝑣𝜃 ) 𝜕(𝜌𝑣𝑧 )

+ + + =0
𝜕𝑡 𝑟 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧
So, that is del ρ by del T plus 1 over r, del over del r ρ v r plus 1 over r, del over del θ ρ v
θ plus del over del z ρ v z is equal to 0. Now, because this is at steady state, so, this term
will be 0 and axisymmetric. So, this term will be 0. Now, what we have is del v z over del
z is constant, because of fully developed flow. So, this is because the flow is axisymmetric
and in this case, this says that fully developed flow. So, del v z over del z is 0 the flow.
We considering this case that the flow of both the phases is incompressible. So, we can
add the assumption of incompressible flow. So, ρ is constant in these cases. So, that is why
we can neglect this term.
𝜕(𝜌𝑟𝑣𝑟 )
=0
𝜕𝑟
𝑟𝑣𝑟 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
245
Now, what we have from this, that del over del r of ρ v r is equal to 0 or v ρ sorry, this is
ρ r v r, here, ρ r v r and; that means, r v R is equal to a constant and v R is equal to constant
over r, but v r needs to be finite. So, that R is equal to 0, v R is finite. So, the term v r will
be 0; that means, that is obvious from the flow configuration itself, because if the flow is
happening in the radial direction; that means, the interface radius will change, because
some flow will happen along this direction and the radius interface radius will have to
change to accommodate that flow.
So, because we do not have any such acceleration flow, so, the radial component of
velocity will be 0. So, the only component of velocity that we have non zero is, v z is non
0 and v r and v 𝜃, they are 0. Ok?.
𝜕𝑣𝑟 𝜕𝑣𝑟 𝜕𝑣𝜃 𝜕𝑣𝑧 𝑣𝜃2

𝜌( + 𝑣𝑟 + 𝑣𝜃 + 𝑣𝑧 − )
𝜕𝑡 𝜕𝑟 𝜕𝜃 𝜕𝑧 𝑟
𝜕𝑝 𝜕 1𝜕 1 𝜕 2 𝑣𝑟 𝜕 2 𝑣𝑟 2 𝜕𝑣𝜃
= − +𝜇[ ( (𝑟𝑣𝑟 )) + 2 + − ]
𝜕𝑟 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝜃 2 𝜕𝑧 2 𝑟 2 𝜕𝜃
Now, let us come to the r momentum equation. So, for the cylindrical coordinate, the r
momentum equation is ρ del v r over del t plus v r, del v r over del r plus v θ over r, del v
r over del θ plus v z, del v z over del z minus v θ squared over R is equal to minus del P
over del r plus μ del over del r 1 over r, del over del r of r v r plus 1 over r squared del 2
v r over del θ 2 plus del 2 v r over del z 2 minus 2 over r squared del v θ over del θ.
246
We do not consider gravity. So, we can say right gravity effect is not considered, if we
take horizontal flow, then the flow will not be axisymmetric. If the gravity effect is
considered, but remember that in micro channel, where the channel 5 is sufficiently small,
the effects of gravity are anyway negligible. So now, we can neglect or we can strike out
the terms which are not relevant for this is at steady state. So, this term will be 0, v R is 0.
So, this term is 0 v θ is 0. So, this is 0 del v z over del z is 0. So, this is also 0 and v θ is 0.
So, this is 0 and because v R is 0, so, this term is 0. This is again 0 because v r and this is
because v θ is 0.
So, what we have is del p over del R is equal to 0; that means, p is not a function of r or
pressure is same at a particular cross section in the radial direction. So, you take the
pressure here and the pressure here, it is independent of the location, but remember for the
two phases, what these equations we are writing general equation for phase 1 and phase
two, but when we consider the boundary conditions, we will have a pressure jump at the
interface. So, you will discuss that, but for a phase inside one phase, pressure is not a
function of radius.
𝜕𝑣𝜃 𝜕𝑣𝜃 𝑣𝜃 𝜕𝑣𝜃 𝜕𝑣𝜃 𝑣𝑟 𝑣𝜃

𝜌( + 𝑣𝑟 + + 𝑣𝑧 + )
𝜕𝑡 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧 𝑟
1 𝜕𝑝 𝜕 1𝜕 1 𝜕 2 𝑣𝜃 𝜕 2 𝑣𝜃 2 𝜕𝑣𝑟
= − +𝜇[ ( (𝑟𝑣𝜃 )) + 2 + − 2 ]
𝑟 𝜕𝜃 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝜃 2 𝜕𝑧 2 𝑟 𝜕𝜃
Now, for θ momentum term, what we have is , ρ del v θ over del T plus v r, del v θ over
del r plus v θ over r, del v θ over del θ is v z, del v θ over del z plus v r v θ over R is equal
to minus 1 over r del p over del θ plus μ del over del r 1 over del over del r r v θ plus 1
over r squared del 2 v θ over del θ 2 plus del 2 v θ over del z 2 plus 2 over r squared del v
r over del θ. Ok?.
So, in this again, we neglect the terms. This term is 0 v θ, is 0. So, this term is 0 v θ is 0 v
θ is 0. So, all these terms are 0 and again, we will have this is 0, because v R is 0. So, we
will have that pressure is not a function of θ so; that means, pressure is, if it is, then it is a
function of z only. So, let us look at the z momentum equation.
247
𝜕𝑣𝑧 𝜕𝑣𝑧 𝑣𝜃 𝜕𝑣𝑧 𝜕𝑣𝑧 𝜕𝑝 𝜕 1𝜕 1 𝜕 2 𝑣𝑧 𝜕 2 𝑣𝑧

𝜌( + 𝑣𝑟 + + 𝑣𝑧 )= − +𝜇[ ( (𝑟𝑣𝑧 )) + 2 + ]
𝜕𝑡 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧 𝜕𝜃 𝜕𝑟 𝑟 𝜕𝑟 𝑟 𝜕𝜃 2 𝜕𝑧 2
When we write the z momentum equation Ρ del v z over del t plus v r, del v z over del r
plus v θ over r, del v z over del θ plus v z del v z over del z is equal to minus del p over
del z plus μ 1 over r, del over del r, del v z over del r plus 1 over r squared del 2 v z over
del θ 2 plus del 2 v z over del z 2.
While I have been very careful while writing these equations; I would request you to
recheck these equations from a standard book. The book from which I have taken the
equations is transport phenomena by Bird Stewart and light foot from the appendix B, you
can see all those equations are in the chapters where the Navier Stokes equation have been
derived. So, again del v z over del t, because it is at a steady state and this term will be 0,
because v R is 0, this term will be 0 because v θ is equal to 0 and this term is 0, because
del v z over del z is 0.
Now, this is 0 because gradient in θ direction are 0 and this is 0 because flow is fully
developed. So, del v z over del z and its derivatives are zero. So, what we end up with is
this, it is minus del p by del z is equal to μ 1 over r, del over del r, r del vz over del r. So,
we will, for clarity or for convenience, we will assume that minus del p by del z is equal
to B. Note, that this can be written. Now as minus dp by dz, because it does not depend,
the pressure depends only on z. So, it is equal to the partial derivative is equal to the total
248
derivative and the v z, we will write as v only, now onwards because for our convenience.
Ok?
So, we can write this equation later on for both the phases. Let us look at the boundary
condition. So, the configuration we look at again. So, this is our interface in the center. So,
at R is equal to capital R, which is general wall. We will have v of liquid, because this is
liquid phase and this is gas phase. So, v L is equal to 0 at R is equal to 0, which is at the
axis. We will have v gas to be finite. So, if you use this condition or then, we will also
have an R is equal to Ri which is at the interface.
There are two boundary conditions, the velocity continuity. So, v liquid at Ri is equal to v
gas at Ri and we will also have shear stress continuity, μ L, del v L over del z at Ri is equal
to μ g, del v G over del z Ri. We will also have a boundary condition for normal stress and
that will be a jump in pressure. So, we will have pL minus pG is equal to sigma, kappa and
kappa is 1 over Ri in this case. So, we can have sigma over Ri, but Ri is constant
throughout. So, we can say that d p L over d z minus d p G over d z is equal to 0, so; that
means, because Ri is not a function of z.
So, that we can say the gradients in the both the phases are same. So, we can have B L is
equal to B G from here.
249
Now, the equation, if we look at this equation again, μ over r del over del r r μ over r, del
over del r, del v over del r R is equal to minus del p by del z or d p by d z, which we have
named as B now. So, we will have, when we integrate this equation, we will get del r, of r
del v r over del R is equal to B R over μ which will give us del, sorry, this is not v r, del v
over del R is equal to this, will be r and this will be equal to B R square over 4μ
So, we can write del v over del is g plus. A constant of integration, let us say, this at C 1.
So, del v over del R is equal to B R over 4 μ plus C 1 by r and now, we can write the same
equation for gas and liquid phases, using the shear stress continuity boundary condition at
the interface, we will have μ gas del v g over del r at Ri is equal to μ L, del v L over del r
at Ri. So, if we substitute here, we will have four constants, basically when we integrate
this fellow.
So, we will have this term, written for two, so we will term these constants as C 1 L and
C 2 L for two different things. So, when we multiply by μ we can have B Ri over 4 and μ
mu will cancel out plus μ G C 1 G divided by Ri, similarly we can write B Ri by 4 plus μ
L C 1 L by Ri. So, B Ri by 4 B Ri by 4 will cancel out and what we will have as the
boundary condition, that μ G C 1 G is equal to or from this boundary condition, we will
get μ G C 1 G is equal to μ L C 1 L. Ok?
So, we have a relationship between the two constants for two phases, C 1 G and C 1 L.
Ok? Now, we can integrate this equation.
250
So, we will have del v over del R is equal to B r over 4 μ plus C 1 by r, when we integrate,
we get v is equal to, sorry, this is, this will be B r by 2 𝜇. So, when we integrate this further,
we will have B r squared over 4 μ plus C 1 L n r plus C 2 and now, we can write the same
equation for gas phase and liquid phase.
So, if we use the boundary condition for gas phase at the axis at R is equal to 0 v G is finite
or defined. So, then because L n 0 is not defined for C 1 G has to be 0 so; that means, we
have B G is equal to B r square by 4 μ plus C 2 G. Ok fine! So, now, we need to find out
the constant C 2 G. Ok?
251
For liquid phase again, we can write the same velocity profile. So, if C 1 G is equal to 0,
this will make C 1 L is equal to 0 too. So, we will have v L is equal to B r square over 4 μ
L plus C 2 L and at R is equal to capital R, that is at the wall, we have a v L is equal to 0.
So, the substitute 0 is equal to B Capital R square by 4 μ L plus C 2 L.
So, by subtracting these two equations, we can get the velocity profile or we can get rid of
C 2 L v L is equal to B over 4 μ L small r square minus capital R square. Now, to obtain v
G, we will apply velocity continuity at the interface; that means, v L at Ri is equal to v G
at Ri and we substitute that, we will get the liquid velocity, we have already got.
So, the gas velocity, we will get after substitution and manipulating the algebra, we will
get v G is equal to B over 4 μ gas small r square minus capital minus capital Ri square plus
B by 4 μ L into Ri square minus r square. So, I request you to go through this step and do
the required algebra. So, we have got the velocity profile in the gas phase and velocity
profile in the liquid phase for a smooth interface annular flow.
252
Now, once we have got that, then, we can get the by integration, we can get the Q L and
Q G. So, Q L will be integrating v G 2 π r d r from Ri to R Q G will be equal to 0 to Ri,
sorry, this is v L and v Q G, will be equal to v G 2 π r dr integrated from 0 2 Ri. Now, from
this we can also find out the superficial velocities. So, when we do the substitution and the
superficial velocity of the liquid phase will be Q L over π r squared.
So, 1 over 𝜋 r square and U G is equal to Q G over π r square. So, 1 over π r square and
by substituting the values and do the necessary integration, we will get U G is equal to
minus B r square over 8 μ L into μ L over μ G minus 2 into Ri by R raised to the power 4
plus 2 Ri square over R squared.
Similarly, U L can be written as minus B R square over 8, μ L into 1 minus Ri squared

over R square. So, now, in general for any annular flow or for any experimental conditions,
in any experiment, what we will know is the gas and liquid flow rates, if we want to obtain
the velocity profile; the problem for us for the velocity profiles are that, they are functions
of B, which is pressure gradient and Ri which is interfacial radius. Now, we do not know
both of these. So generally, what we have Q L and Q G or U L U G, these are the superficial
velocities.
So, you can just note that U L U G are superficial velocity are generally known for a flow.
So, by knowing this we have two equations and we can solve an equation, let us say, A
and equation B to obtain B, which is pressure gradient and Ri.
253
So, we can look at this in the ideal annular flow, these are the U L and U G, the two phase
flow rates and the superficial velocities, which we have solved, just solved for and now,
when we solve for pressure gradient and interfacial radius.
So, we can rearrange these two equations to in such a manner that it becomes a quadratic
equation, in sigma i R square, sigma i square where sigma i or ξ i is equal to Ri by capital
R. Ok?
254
Sorry, this is not sigma, this is ξ and so, ξ I is equal to Ri by R. So, once we solve that, we
can obtain what is ξ i square and from that, we can obtain in the roots of it. Generally, with
my experience, I have observed that one root is always greater than 1. So, we know that is
unphysical. So, we will have to choose ξ i with the condition that ξ i is between 0 and 1.
So, we can get what is ξ i and when we substitute back in the pressure gradient, we will
get the pressure gradient.
So, basically by solving what we have done today, is solve the ideal annular flow equation
starting with a basic Navier Stokes equation, we have neglected the terms the relevant
terms in the R θ and z system of the equations and in the continuity equation and finally,
after integrating we got the velocity profile, which had 4 constants; two constants for the
liquid phase and two constants for the gas phase.
And then, we had four boundary conditions, one is at the axis, that the velocity gradient
should be finite and L N, it should be defined at the axes. Then, we had two boundary
conditions at the interface, the interface continuity and shear stress continuity and one
boundary condition at the wall that the liquid velocity at the wall is 0. So, by solving that
we could obtain the velocity profiles in terms of the pressure gradient and the interfacial
radius. Once, we have done that, then we integrate it and obtained the flow rates of the two
phases or from the flow rates, we can also obtain the superficial velocities and from solve,
by solving these two equations, we could get the pressure gradient and Ri or interfacial
radius.
So that we can get the velocity profile for any given gas and liquid phase flow rates. Now,
this ideal annular flow, there is lot of literature on the stability analysis non - linear and
linear stability analysis of the annular flow, this is subject to a different interfacial
instabilities.
So, if you remember the capillary instability, which we discussed in the previous class. So,
in the capillary instability a jet disintegrates into smaller droplets. So, following that, the
annular flow is at lower velocities, it becomes Slug Flow or Taylor Bubble Flow and there
are generally, what we see in practice that, there are always waves present at the interface
and that is because of probably the Kelvin Helmholtz instability, which is that, when there
are two parallel liquids flowing. There are waves generated or there are small
perturbations, they grow up to become wave to make the interface wavy ok,
255
Thank you.
256
Lecture – 13
Taylor Flow: Heat Transfer
Hello. So, in this lecture we will be talking about heat transfer in the Taylor flow regime.
So, as we know that in microchannels, the surface area to volume ratio that is surface area
density is large when you compare with large diameter channels or conventional channels.
So, what happens that one can get large surface area for heat transfer or large interfacial
area for heat transfer, and that results in enhanced heat transfer performance in
microchannel, additionally the diffusion length, because in a channel when the flow is
happening during the convective heat transfer the radial conduction happens from the
channel center to the wall.
So, that is the diffusion length or the, that is the conduction length and as the channel size
becomes smaller, this conduction length becomes smaller. So, the heat transfer rate for the
same conductivity will be increased. Now looking at some of the applications in which the
heat transfer in microchannel is being used increasingly are say electronics cooling. So, in
electronics cooling, the cooling of different parts of computer or other hardwares as we
257
know of from Moore’s law that in about a year a number of chips on a motherboard on a
number of chips increases almost every year
So, that increases the requirement for the cooling of the electronic equipment that increases
that the larger heat flux, when the number of chips are increasing, then the largest heat flux
need to be removed. So, at one stage the number of chips that can be put are limited by the
heat flux that can be removed from the surface. So, it is important to develop technologies
which can enhance the heat transfer performance for electronics cooling, and one of those
technologies is based on microchannels. So, in that case when we have single phase flow.
Now the performance can be further increased if in place of gas cooling one uses liquid
cooling, because thermal mass of the liquid is more than that of gas.
Further if we want to increase the performance further, then additional means of heat
transfer mixing is required, because in microchannels we always have flow to be laminar
or in most of the cases, the flow is generally laminar. So, there is no mixing offered by the
turbulent nature of the flow. So, people have devised different active and passive means
of mixing one of them being using the, say serpentine channel which causes the flow to
have a secondary flow in the secondary direction, say if this is the cross section of the
channel, then in addition to the flow in the streamline direction. There is another flow in
the cross section of the channel at that and that enhances the mixing in the channel. So,
these are called dean vortices
So, that can be one of the solutions. Another solution is where people use baffel in different
arrangements to enhance the radial mixing by convective means and so on so forth. So,
further on if we want to increase the further heat flux or for a single phase flow or liquid
only flow, if the heat flux is increased at one point, it will start evaporating and in that case
we will have a boiling in microchannels.
So, that becomes important, or one can have another means of mixing that the flow can be
segmented which is generate a Taylor flow regime, using bubbles or droplets. So, if one
uses bubbles than one is just segmenting the liquid flow and enhancing the leak internal
recirculation, and the heat transfer increase significantly 2 to 3 times depending on the
flow conditions
In comparison to gas-liquid segmented flow or gas-liquid Taylor flow the heat transfer is
further increased, if one have liquid-liquid flow, because now both the phases will have
258
internal recirculations and both the phases will also increase or will contribute to the heat
transfer. So, the heat transfer can be further increased, when one compare with the gas-
liquid flow ok..
Another application is exothermic gas-liquid reactions. So, a number of applications which

require or for which the microchannels are being used for, are for the reactions which are
highly exothermic, because it is relatively safer to carry out these reactions in
microchannels.
One of the reasons being that higher heat flux can be handled in small diameter channels,
and additionally when we have gas-liquid reactions, the mass transfer is also enhanced by
the, by carrying out these reactions, because the interfacial area has increased significantly,
when we compare with normal channels or conventional channels, then they have
applications as compact heat exchangers in aerospace engineering applications. As we can
guess that the design of aerospace equipments is limited by that one, should have the
highest performance with the minimum mass. So, that more and more payload for a space
applications on a more and more load in the aeronautical applications can be taken there
So, one would like to have the lowest mass possible, and that will be achieved that, can be
achieved by having high interfacial or high surface area density. So, it is being explored
for application in aerospace engineering, then offshore petroleum engineering, where
again one would like to have compact heat exchangers. So, the offshore industry is using
the compact heat exchangers. One of the companies that supply the heat exchangers is
name as Heatric, and it is a part of a company called Meggitt PLC. So, you can look at
some of the applications of these compact heat exchangers, the heat exchangers that
Heatric offer are known as PCHE or printed circuit heat exchangers ok
So, printed circuit heat exchangers, what they are that in that is one of the way to a scale
up the microchannels for industrial applications. So, in a metallic plate, a number of
channels are hatched, which are of the millimeter size or smaller dimensions, and then
another plate which has the same kind of channel it diffused, bonded on this plate and then
this flats plates are stacked together. They may be thousand of channels in a plate and
thousands of plates or hundreds of these plates can be stacked together and then the flow
happens in this
259
Now, one of the challenge, it is such cases will be the distribution of the fluids in the
manifold, but that is not our topic of discussion today. So, these, these printed circuit heat
exchangers are already being used for a number of applications; say for example, of shore
applications right different companies ok.
So, coming to heat transfer in Taylor flow regime. Generally, it is when we are looking at
it, it is convective heat transfer that is, because the flow is happening and, the heat is either
entering in the channel and being removed by the fluid or it can be the other way around
that the heat is being taken away in case of say exothermic reactions
So, they can offer better heat transfers than liquid only flow as I just said, because the
typical structure in a Taylor flow is, we have the dispersed phase this maybe bubbles or
droplets and in between these bubbles, there is internal recirculations in the liquid slugs..
So, this internal recirculation causes that the hot fluid from the wall, that comes towards
the centre and them efficient mixing happens. So, it enhances the heat transfer with respect
to the liquid on the flow of course subject to the condition that if these dispersed phase is
bubble and then the volume of bubble should be smaller. So, that the efficient heat transfer
is happening
Now, because in gas-liquid flow, they do not contribute to the heat transfer and as I said
earlier that, this heat flux that can be removed, can be further increased by phase change,
260
where one get the QL or the latent heat that can be removed. So, one can remove further
heat flux or higher heat flux by evaporation and boiling ok.
So, let us look at some of the basics, because this course is not about heat transfer. So, let
us remind ourselves some of the basics of convective heat transfer. So, when we talk about
forced convection, it is often represented by a non dimensional number which is known as
Nusselt number and it is abbreviated as Nu, given as
ℎ𝑑
𝑁𝑢 =
𝑘
where h is heat transfer coefficient, d can be taken as channel diameter or in case of non
circular cross sections, one can take as take this as hydraulic diameter and k is conductivity
of the fluid..
Now in case of two phase fluids, it is important to understand which phase of

conductivities should be used in this expression, because this is a comparison of. If you
can rearrange this h over k over d. So, the convective heat transfer divided by the
conductive heat transfer in the radial direction. So, the conductivity of the phase in contact
with the wall; that is what should be used now.
So, we know what d is the diameter of the channel, and what k is the conductivity of the
fluid which is the property of the fluid. and now we can find it out or we can calculate by
261
using different formula available in transport phenomena. Now h is the heat transfer
coefficient which is defined as
𝑄𝑤𝑎𝑙𝑙
𝑎𝑟𝑒𝑎
Q wall over area, area of the wall, it is important to remember that this is not the cross
sectional area, but the area of the channel. So, for a cylindrical channel, it will be 2 π R L
Q w over A w into T wall minus T bulk or T bulk mean. So, T bulk or T mean
So, generally there can be different kind of boundary conditions one can have on the wall.
So, in general one encounters the constant wall heat flux constant or constant h boundary
condition, which is called this constant wall heat flux. So, it is called h boundary condition
that the heat flux is constant. The another boundary condition is constant wall temperature.
So, in that, in this case it is called T boundary condition or the third boundary condition is
mixed boundary condition, where one will have a relationship between Q wall is equal to
h A say T wall minus T and
So, in this case. If h known then this can be used as a mixed boundary condition, one can
see that Q wall from Fourier’s law of heat transfer is equal to k dT over dr at the wall. So,
this is basically the boundary condition in which the gradient of the temperature is being
defined. Now the question comes when we are defining the heat transfer coefficient T wall
is wall temperature, and this T which is bulk temperature or mean temperature.
What this temperature is, because if we look at the flow in a channel you can see here is
heat entering at the channel wall the, the temperature profile may look something like this
which suggest that the temperature will be highest near the wall, and then it will be
varrying and become minimum at the centre and then again increase towards the wall.
So, the, clearly the temperature varies according to the with the radius of the channel. So,
the which temperature, should we take say in this case the mean temperature is defined as
equal to T mean or T bulk mean is equal to integral ρ u Cp T dA divided by integral ρ u
Cp dA..
262
So, where ρ is the density of the fluid, u is the local velocity of the fluid, Cp is the heat
capacity and T is the local temperature of the fluid and dA is the area of the cross section,
it has to be integrated over the area, it has to be integrated over the area. So, what you can
observe from this that the only difference between the integrals at the top and bottom is T.
So, this is basically mass weighted average of the fluid ok. So, that is bulk mean
temperature.
Now, because this is flow in a channel. So, we will remind ourself the temperature profiles
for, to often use boundary conditions for h boundary condition and for T boundary
condition. So, in a constant wall heat flux case as one can see or one can speculate that the
temperature off the wall will be increasing and the temperature of the fluid will be
increasing.
We have this in the fully developed regime from the, when say a fully developed region,
this is thermally fully developed region. So, in the thermally fully developed region, we
will a little later, we will talk about how do we define the thermally fully developed region
in a in a channel. So, this is the upper line is for T wall and the line below is T bulk and
we can say that they are parallel to each other, and the slope of these lines can be found by
writing the heat balance equation. So, the temperature of the wall and temperature of the
fluid both increases linearly, and the difference between them is constant everywhere for
the fully developed flow case
263
Now, for a constant wall heat flux case, this will not be the case, but what we will have is
the temperature and x. So, the temperature of the wall will be a constant, because that is
what the boundary condition is, and then as the fluid start flowing into it. The temperature
of the fluid will increase and then approach towards the temperature of the wall, and that
will be achieved only at infinity.
So, in this case we can see the difference between the temperatures keep decreasing, as we
𝑑𝑇
move in the axial direction and Q is equal to 𝑘 𝑑𝑟 at the wall Q wall. So, we can see that Q
wall in this case will keep decreasing almost exponentially ok. So, these are the typical
temperature and heat wall, wall, heat flux profiles for single phase flow in a channel.
Now, for thermal heat fully developed flow. So, if we remember or if we remind ourselves
that hydrodynamic fully developed flow refers to that
d𝑣𝑧
=0
dz
𝑇𝑤 − 𝑇
𝜃=
𝑇𝑤 − 𝑇𝑚
dθ
=0
dz
264
d v x or the velocity in the axial direction. Let us say d v z over d z is equal to 0. So, the
velocity profile is not changing along the axial direction. So, if we look at the velocities at
two different locations x1 and x2 in the channel x1 and x2, then we will find that the
velocity at the corresponding radial locations are same in both the cases
Whereas if we look at the temperature profiles, here it clearly tells us the temperature has
to increase, because it absorbs more and more heat. So, we cannot have that the thermally
fully developed flow in analogy with the hydrodynamically fully developed flow, we
cannot have d T over d z is equal to 0; that is not possible for thermally fully developed
flow. Whereas, a non dimensional temperature. So, the temperature do increase, but the
non dimensional temperature which can be defined as T wall minus T divided by T wall
minus T mean and d theta over d z is equal to 0 for thermally fully developed flow. So, let
us remember all this
Now, what we talked about that, when the flow is has developed, what is the, what are the
temperature profiles and how we define the non dimensional temperature and so on and so
forth. So, it can also be seen, shown that for laminar fully developed flow the Nusselt
number in a channel in for fully developed flow and fully developed heat transfer, the
Nusselt number is a constant. So, for h boundary condition Nu h is equal to 4.36 and for
constant temperature boundary condition it is about 3.6 around them ok
265
So, for developing flow as one can expect that the heat transfer or the or the heat transfer
coefficient or the Nusselt number will be higher, when the flow is developing. And then
eventually it becomes constant, it has been plotted the Nusselt number, the local Nusselt
number Nu x, it has been plotted along the channel length when I say local Nusselt number;
that means, it is being calculated along the length of the channel.
So, Nu x is h d over k and h is Q w x, because this is constant and T wall at the x location
minus T bulk at the x location. It does not consider the entire the thing, the in the heat
transfer from the inlet to this point, this has been plotted only the heat transfer at the cross
section. So, that is why it has been termed as Nu x.
So, the heat transfer coefficient keep decreasing as we move in the axial direction. Notice
the non dimensional parameters at the x axis what we have here is, x over d Re Pr. So, all
of you will know that Pr is defined as Cp 𝜇 over k. Now we can rearrange this as μ over 𝜌
divided by k over ρ Cp. So, μ over ρ is known as dynamic viscosity or momentum
diffusivity and k upon ρ Cp is known as thermal diffusivity.
So, this is basically the ratio of momentum diffusivity and thermal diffusivity in the Prandtl
number is the ratio of momentum diffusivity and the thermal diffusivity. So, one gets this
the, the non dimensional number that is important in this case is x over Re Pr and this
Prandtl number is a function of the property of the fluid only, and this decreases there.
There are different correlations that has been given and one of them is that it the Nusselt
number decreases as x over d Re Pr of this non dimensional number raised to the power 1
minus 1 by 3, when this is less than 0.03, and then it decreases according to this equation
when x by d Re Pr is greater than 0.03 and this number inversely. This is written as, we
can change it as d small d for uniformity
So, the Graetz number is written as d over x Re Pr. So, this number or this term is also
known as inverse. Graetz gave an analytical solution for a hydrodynamically fully
developed and thermally developing flow and it is a series solution. So, to simplify the
solution in number of approximation has been presented, and this is one of those. So, what
I would like to emphasize here that the developing Nusselt number is a function of inverse
Graetz number or Graetz number ok, which we will use later on when we develop the
model for Taylor flow heat transfer.
266
So, coming to heat transfer in the Taylor flow regime, the flow in Taylor flow regime is
pulsatile. So, the flow changes with time, but it is periodic. So, it repeat itself after a certain
time or you can see this periodicity in space as well. So, if you are standing at one place,
you will see one bubble coming along and then one slug comes in and then you have
another bubble coming along and this will be periodic. So, you can see if you are staying
at one place and you can see the, the volume fraction varying with respect to time and this
variation maybe periodic
Similarly, if you are looking at in space that one time instant, if you look at you will have
number of bubbles at different at regular intervals, the distance between the two bubbles
will be same and the shape of the two bubbles will be same. So, the flow is periodic in
space as well. Now to do experiments for this for Taylor flow, one need to make sure what
boundary condition is being used, and one also need to look at that what kind of heat
transfer one is looking at, developing or simultaneously a developing hydrodynamics and
heat transfer or they are looking at the heat transfer. The hydrodynamics of the flow has
already developed and the and the heat transfer is developing and then it becomes fully
developed.
Now, when we do simulation or when we are studying, we can decouple phase change
easily. You can say that we are not going to include a phase change model in our
computational model saying CFD is, but when this happens experimentally, then there is
267
no way to suppress or it is difficult to suppress the heat the phase change that might happen,
because of evaporation. So, what one can do that, it can be minimized to the maximum
extent possible and that can be done. If one uses a nonvolatile gas-liquid combinations or
nonvolatile liquid and the temperature of the fluids is at slightly lower range. So, the
volatility or the evaporation rate is negligible or very small.
Now, So, then the question comes when we measure what, what is to be measured. So, it
is Nu is equal to h d over k. So, if one have specified a constant wall heat flux, which is
generally the case during the experiments, then what is need to be measured is T wall and
T bulk. So, the T wall can be easily measured by putting the thermocouples at different
locations on the wall or by different wall temperature measurement techniques.
However, it is not easy to measure bulk fluid temperature, because if one put a
thermocouple inside the channel, then the flow behavior is going to change, because of the
presence of the temperature sensor or the that is put there. Then in most of the cases the
temperature in the channel walls. So, if let us say this is the channel, this is the internal
wall of the channel and this is the outer wall of the channel and this is the channel is
thickness of the channel.
So, this is Ri and this is Ro. Now the sensors will be put somewhere here on the wall
temperature will be measured. So, one need to take into account the heat transfer; that is
happening from here to the liquid solid interface. So, in general when one put the
temperature sensors here, then one can obtain an analytical solution for the conductive
heat transfer in this direction, or if you want to take into account the axial heat transfer
also, then one can have a full 3D simulations to look at the effect of conjugate heat transfer
ok. So, all these factors needs to be thought about when performing the experiments.
268
So, this is a typical experimental setup in which the liquid is coming from a liquid tank,
and the gas nitrogen, gas is being supplied and then the inlet flow temperature is being
measured. Then they come into a mixing section and the fluid move along the test section,
and the temp, wall temperature is being measured at different sections. The wall is being
heated A by A power supply.
So, this is a test section that was used by Miss Leone in her experiments at the University
of Sydney and then the outlet temperature is measured. They have also arrangement for
looking at the and confirming the flow regime in the channel and then it is being brought
together, been brought to outside and released to the atmosphere
So, that is a typical experimental setup and what is being measured, they have a h boundary
condition. In this case and the wall temperature is being measured at different sections ok.
one, I advice to further look into the details of the test section, because this is, there is lot
of detail in while designing the test sections, especially the h boundary condition or T
boundary condition in the, in this blocks ok.
269
So, the another test section that has been used by walls et al. it is a similar test section and
they have made sure. Here actually they have also made sure that the flow is
hydrodynamically fully developed, and then there is heat being applied on it, and this heat
is being applied by joule heating, and the local measurement of surface wall temperature
is being done and using an IR camera probably.
And then this is different tube cross sections they have taken and wall thickness and the
length of the heat section. So, then they are also looking at the fluid behavior or the flow
regime and the exit
270
ℎ𝑑
𝑁𝑢𝑥 =
𝑘𝑐
𝑞𝑤
ℎ=
𝑇𝑤 ,𝑥 − 𝑇𝐵 ,𝑥
∫ 𝜌|𝑈𝑥 |𝑐𝑃 𝑇𝑑𝐴

𝑇𝐵 ,𝑥 =
∫ 𝜌|𝑈𝑥 |𝑐𝑃 𝑑𝐴
So, while looking at the Taylor flow behavior. Now we have two phases gas and liquid or
liquid-liquid phase. So, one need to define a Nusselt number or one need to revisit the
definition of Nusselt numbers. So, we can define a local Nusselt number Nu is equal to h
d over k, and because it is always the continuous phase, which is in contact with the wall.
So, k can be k of the continuous phase, h is the heat transfer coefficient and d is the
diameter of the channel. Now h is defined as q wall over T wall minus T bulk
So, we will be defining these definitions for h boundary condition. Most of the discussion
that we will have in this lecture will be based around the constant wall heat flux boundary
condition. So, this q wall is available from experiment and it is an input T wall is measured.
Now generally T bulk is a problem ok. So, what we are looking at here is the local
definition of Nusselt number. So, Nu x, everything else is going to be same and everywhere
on the cross section q wall and the temperature of the wall is at that particular location.
Now we also need to define T bulk x.
271
So, T bulk, we remember the definition of bulk temperature, then it can be defined integral
at the area ρ ux, if the area or if the axial direction is x then ρ ux Cp T at the particular
location dA divided by integral over the area ρ ux Cp dA. Now because the there can be
backflow in the liquid. So, if we take in do not take into account, only the magnitude of
the axial velocity then the terms will cancel out, and we will have a wrong definition or
sometimes a very small bulk temperature which does not represent the bulk temperature
correctly.
So, we have put that the absolute value of the axial velocity is being taken into account
here. So, that is our local bulk fluid temperature; however, this is good, if you want to do
computations and we can get the local Nusselt numbers at different locations in the
channel, but if we want to compare this with the experimental data, we do not have, we
might have with the techniques available. Now we can get the wall temperature locally,
but it will not be possible to get the bulk temperature with the same definition. So, this is
generally not used for the design calculation.
ℎ𝑑
𝑁𝑢𝑈𝐶 =
𝑘
𝑞𝑤
ℎ=
𝑇𝑤 − 𝑇𝐵
272
𝑥+𝐿𝑈𝐶
∫𝑥 𝑇𝑤 2𝜋𝑟𝑑𝑟
𝑇𝑤 = 𝑥+𝐿𝑈𝐶
∫𝑥 2𝜋𝑟𝑑𝑟
∫𝑈𝐶 𝜌|𝑢𝑥 |𝑐𝑃 𝑇𝑑𝑉

𝑇𝐵 =
∫𝑈𝐶 𝜌|𝑢𝑥 |𝑐𝑃 𝑑𝑉
Now, So, for design calculations we also need to have overall Nusselt number, now
because Taylor flow is periodic. So, if we consider the flow, when the heat transfer also
have become periodic, then we can analyze flow in one unit cell which consists of one
bubble or droplet and the slug around it. And once we have obtained this we can use the
same Nusselt number for the entire channel.
So, it is easier to understand the behavior in one unit cell, which consists of one bubble or
droplet which is the dispersed phase and two halves of the liquid slug on the two sides of
the unit cell ok. So, when we define the Nusselt numbers for a unit cell, we need to see Nu
unit cell is equal to h d over k will be k liquid, no change there, d will be the channel
diameter h is equal to q wall over T wall minus T bulk
So, now remember this is for the entire unit cell which consists of the bubble or droplet,
and two halves of the liquid cells. So, the thermal conductivity as we discussed already;
that it is always the continuous phase or the liquid phase, which is in contact with the wall.
So, the conductivity of the continuous phase will be taken into account the wall heat flux.
If we have the any boundary condition which is not constant volume flux and we can get
area, average area weighted average of the heat flux and the wall temperature that will be
measured and for calculations.
If we want to get the, because here what we want is, the area weighted average. So, on the
wall temperature in the entire location. So, we can do it integral area of the wall T wall dA
divided by integral area of the wall dA. So, remember this is for a circular channel, it will
be integral x location to x plus L, which let us say this LUC and this is T wall 2 π r d x
divided by integral x 2x plus LUC T wall 2 π r d x on T wall.
This will have only 2 π r d x. So, 2 π 2 𝜋 will cancel out and we can get the area weighted
average. Now the bulk mean temperature T bulk will have the same definition, but now
the integral will be over the entire unit cell. So, we will have integral UC. Remember that
273
is that in this case also we have 2 phases. So, we will have the gas phase and the liquid
phase, depending on and what location what phases present.
So, you will have a integral ρ ux Cp T dV divided by integral ρ u x Cp dV. So, that will
give us the bulk mean temperature. Now, one can see and confirm for himself that this
bulk temperature will not be same, if one sums up the temperature at different axial
locations or it will not be simply integral TBx dx x dx, that will not be the case. So, one
need to be careful when defining this.
Now, so in the experiment the Nusselt number estimation is done, based the axial lengths.
So, if this is the channel from where it is being heated, the channel wall is being heated
and the total heat flux to this is, let us say Q in. So, we can write a heat balance at a location
x from inlet. So, it can be written as m dot cp and m.dot Cp is for 2 phase and TBx or bulk
temperature at x location minus T inlet, where T inlet is the temperature at this location is
equal to Q in x by L.
So, this is the heat flow rate and if this the Q in is for the entire length, and then what we
are taking the how much heat has gone in this that will be equal to the heat absorbed by
the liquid or the gas-liquid flow. So, that will be m dot cP m dot cp will have to take into
account that, there are 2 phases present in it. So, TBx minus T in and. Of course, the
thermal conductivity will be the gas and liquid phase and one can use and see the variation
274
of k in the temperature range between the minimum and maximum temperature, and
judiciously use an average value
The wall heat flux is one of them, either the wall heat flux or wall temperature is constant.
Often It has been observed that the experiments are done with the constant wall heat flux
boundary condition, and the wall temperature is measured along the axis, and this is the
way by which we can calculate TBx, as one can see that one will know what is the heat
being input it to the system and what location one wants to measure the temperature, and
the flow rate and the properties of the two fluids will be known. So, one can calculate what
is the bulk temperature at the location
Now, this is how it is been. Now what we have described just and the bulk mean
temperature can be measured from heat balance can be obtained by heat balance, but it is
difficult to measure it; however, recently it the technique using laser induced fluorescence,
where the particles which are following the flow faithfully, and they are quoted by a
fluorescent material, which is temperature sense, which gives temperature sensitive
fluorescence.
So, the efforts are being made to measure the local temperature in the fluid. Efforts are
also made to put the thermocouple inside the channel, which may affect the flows slightly
and then look at the temperature field in the, and the channel and obtain the mean, bulk
mean temperature from there ok.
275
So, let us look at some of the results, this is a typical temperature profile in a unit cell. So,
why you see her from the bubble, because an axisymmetric channel, then this is from a
CFD computational, where this is done in an axisymmetric channel. The black line shows
the boundary of the bubble, the other bubble will be on the bottom half and you can see
the temperature, there is a constant wall heat flux boundary condition.
So, the heat flux is, heat is entering from the wall and the fluid as hot as you can see from
this temperature scale, which vary from 301 to 305.5. So, the temperature is hot, hot is the
yellow color and the red color represents the hot fluid and then this hot fluid is being
brought towards the centre. So, the heat is being brought by the fluid using convective
convection and the, the mixing of the fluid is happening, the temperature of the bubble
seems to be very high, but remember that very small heat very little amount of heat is
required to increase the temperature of the bubble. So, this is a typical temperature profile
and this also shows us clearly the presence of internal recirculations in the liquid slugs.
276
Now, this is a typical wall heat flux for T for the constant wall temperature boundary
condition. So, you can see that the wall heat flux is high in the liquid slug zone, and it is
small in this region, because that that will be q wall1 is equal to k dT over dr at the wall.
So, the temperature difference will be small in this region now. So, the heat flux will be
small ok.
So, then Nusselt numbers and these are local Nusselt numbers. So, you can see that one
fine or one obtain very high Nusselt numbers in the film region. So, let me remind you at
277
the outset that Nu that has been defined is equal to h d over k, where we have taken this
length scale and diameter; however, if one look at the heat transfer is being, is because of
the film only.
So, the actual length scale that should be used here, is 𝛿F and if one take that into account
this will be very small, because delta F over d is of the order of say 10 to the power minus
2 or smaller. So, the Nusselt numbers are not high. In this case this is just because of the
length scale being used here and the Nusselt numbers in this case is about the as one can
see from here on the y axis the Nusselt number is about 10. So, this is significantly high
when you compare with liquid on the flow in the channel. So, both the boundary
conditions.
So, now let us look at a bit about the heat transfer mechanism. So, we have the bubbles
and the internal requisition in the liquid slug, and the heat can find different routes to go
from the wall in the liquid, which is, which has the high heat carrying capacity. So, the
one route is that the heat goes from wall enters from the wall, and it goes into the bubble
fill or the film that is surrounding the bubble, because the heat carrying capacity of this
bubble is very small
So, it cannot carry much of the heat. So, it waits until the next slug comes in and then it
supplies this into the slug. So, that is first route. Now the second route is that the bubble
film, it passes on the heat to the liquid slug. So, and thirdly that wall to the recirculating
278
liquid slug zones of the heat and that enters from the slug region that passes the heat from
wireless film to the liquid slug directly.
So, one need to remember or the general picture is that, because the heat enters in the film,
but the bubbles, if it is gas-liquid flow then the bubbles do not carry much of the heat and
this heat is taken over by the liquid slug which comes behind. If it is a gas-liquid or if it is
a liquid-liquid flow, then this heat will be redistributed between the two phases. So, we
will look at, when we develop a heat transfer model, we will look at this correlation from
the ground.
So, let us look at some of the correlations that have been presented in the literature for
predicting the Nusselt number in Taylor flow regime. So, this Nusselt number is defined
as Nu is equal to 1.615. And now you can remember the inverse Gratez number which is
Rr Pr channel diameter by L raised to the power 1 by 3
So, this what they have done is modified the Gratez Leveque solution, which is for
thermally developing flow, and in this solution they have included the effect of bulk
viscosity, its like seider-tate equation for single phase turbulent fully developed flow, and
then they have introduced the effect of homogenous write fraction in terms of beta. So,
that has been modified to obtain the Nusselt number.
279
Another correlation is similar modification by Gratez. So, they had a simpler modification
that this is the slug Nusselt number and from that one can just calculate the thermally
developing flow. So, basically they said that the heat transfer in the slug region is thermally
developing flow and that can predict the Nusselt number in for the Taylor flow ok.
Kreutzer, they did some experiments, they wanted to do. I mean most of their work was
based around mass transfer in Taylor flow regime, and to develop a correlation for mass
transfer they did a simulation for heat transfer. And from there they look at the Nusselt
280
number in the slug region or a cavity which was that they have done a simulation, where
the cavity in which the these are two walls and the these two walls are moving
So, the fluid will eventually develop recirculation inside it. So, they calculated the Nusselt
number in these cavities and found this correlation for the slug Nusselt number.
Walsh et al, they developed another correlation in the slug region and that was a Nusselt
number into homogenous wide is equal to homogenous void fraction single phase, fully
developed Nusselt number depending on the boundary condition plus 25 LS by the raised
to the power minus half. So, again the effect of developing it
So, we will stop here. So, in summary what we have looked at in this lecture is Taylor flow
in microchannels, the heat transfer in this flow regime. Before that we looked at the basics
of heat transfer in particular, the definition of Nusselt number definition of heat transfer
coefficient and definitions of bulk mean temperature.
So, by looking at these definitions, we have also seen later on that when we looked at the
some of the correlations that has been that have been developed for Taylor flow heat
transfer. They are based on the developing flow in for liquid only flow in a channel. So, in
the next lecture we will look at the heat transfer in the Taylor flow regime at and a general
correlation that can be developed for heat transfer in the Taylor flow regime
Thank you.
281
Lecture – 14
Taylor Flow: Heat transfer 2
So, in the last class we looked at the basics of heat transfer in single phase flow, some
basic definitions and then some typical characteristics of heat transfer in the Taylor flow
regime. Based on those, based on the understanding that we developed in the last class, let
us try to develop a general correlation for the heat transfer in the Taylor flow regime. Of
course, we would need to make some assumptions to make this a closed system of
equations.
So, that correlation which can be used or a system of equations which can be used to
predict the overall heat transfer in a unit cell for fully developed Nusselt number in a
channel.
𝜕𝑣𝑧
=0
𝜕𝑧
𝜕𝜃
=0
𝜕𝑧
282
So, before we go on, let me try to say few words or sensitized you a bit about the fully
developed Taylor flow, because our general definition of fully developed is that for
velocity. We say that del or dvz over dz is equal to zero and d theta over dz is equal to
zero, my apology, because I keep exchanging this bit, either x or an z. Probably you will
understand that these are, whenever I use this, most of them, most of the time it is referring
to x axial coordinates sometimes I call it x and sometimes I call it z. So, this is for true, for
single phase fully developed flow, but when we talk about Taylor flow, it is not possible
to achieve this, because the velocity at the middle of the bubble and the velocity in the
liquid slug or even in the liquid slug at two different location, is going to be different
So, what we mean by fully developed Taylor flow or what somebody else mean by fully
developed Taylor flow is that, the flow has become fully periodic. See if you look at two
periodic locations then the flow has achieved period periodicity. So, if you compare the
velocity at two periodic locations, then this should be equal. Then it has become, flow has
become fully periodic hydrodynamically. So, if the model that we present here, is based
on professor Brian Haynes and his students and collaborators professor David Fletcher
So, what we are looking at in this plot, is along the distance, the temperature and there are
two different temperatures have been plotted; the temperature at the wall of the channel
and the temperature of the fluid or the or the bulk temperature of the fluid and this has
been done for a number of bubbles and this is a snapshot in time. So, one can look at the
different bubbles from the inlet and the wall temperature is changing. So, it has not become
steady, yet it is almost. So, what we look at that, the temperature is increasing continuously
and these are the, this is the non dimensional radial location
So, this is from zero to 1. So, that is basically small r by capital R and this is the wall of
the channel and the blue symbols represent the shape of the bubble. So, one can look at
the temperature of the liquid and the temperature of the gas, both of them decreases, but
there is a jump here. So, if we look at the corresponding location. For example, this point
and this point, there is a corresponding increase in delta T.
And similarly there is corresponding increase in delta T here, and that the same like. So,
there is a increase in temperature in both. So, when the flow becomes fully developed, then
delta T in a unit cell is a constant and it will also require that the temperature of the two
phases. So, the temperature of the continuous phase and the temperature of the dispersed
283
phase, rise in the temperature of the continuous phase, and the temperature of this
dispersed phase are same.
𝑄𝐷 = 𝑚𝐷0 𝐶𝑃 𝐷 (∆𝑇)𝐷
𝑄𝐶 = 𝑚𝐶0 𝐶𝑃 𝐶 (∆𝑇)𝐶
𝑄𝐷 𝑚𝐷0 𝐶𝑃 𝐷 𝜌𝐷 𝛽𝐶𝑃 𝐷
= 0 =𝑚=
𝑄𝐶 𝑚𝐶 𝐶𝑃 𝐶 (1 − 𝛽)𝜌𝐶 𝐶𝑃 𝐶
𝑄𝑊,𝑈𝐶 = 𝑄𝐷 + 𝑄𝐶
So, based on this we can write the heat flux Q by the dispersed phase is equal to m dot
dispersed phase CPD into delta T of dispersed phase. Similarly, the Q for slug will be
equal to m dot slug CP slug or we can say, because what we are looking at rather than slug
we are looking at, here is the continuous phase which is slug plus the film. So, m dot c is
CPC delta T continuous phase
Now, based on this we can get, because delta T of dispersed phase and delta T of
continuous phase is going to be same. So, we can basically obtain QD over QC is equal to
m dot DCPD over m dot c. CPC is equal to m and the total heat that will be coming from
the wall. So, Q wall in a unit cell will be equal to QD plus QC. So, one will have QD over
Q wall is equal to m over 1 plus m and Q continuous phase divided by Q wall is equal to
284
1 over 1 plus m. You can also write m in terms of ρ d into beta CPD divided by 1 minus
beta ρ C CPC
We will have in both the cases U here, but then the U will cancel out. So, that is the
definition of m. Basically the ratio of thermal mass mass flow rate multiplied by the heat
capacity. So, that is important relationship that how much or what fraction of the heat is
going to the dispersed phase and what fraction of the heat is going to the continuous phase,
if d is gas. So, then we can see that m, because beta and the minus beta they will be
equivalent numbers, CPC are also of equivalent and. So, because ρ gas is very small, when
you compare with that of liquid. So, m is almost negligible; so you can neglect and most
of the heat, then this shows that for gas liquid flow the heat goes in the liquid phase only
ok.
𝑚𝐷0 𝐶𝑃𝐷
𝑇 + 𝑇𝑆
∫𝑈𝐶 𝜌𝑢𝑐𝑃 𝑇𝑑𝑉 𝑚𝐷0 𝐶𝑃𝐷 𝑇𝐷 + 𝑚𝐹0 𝐶𝑃𝐶 𝑇𝐹 + 𝑚𝑆0 𝐶𝑃𝐶 𝑇𝑆 𝑚𝑆0 𝐶𝑃𝐶 𝐷
𝑇𝐵 = = =
∫𝑈𝐶 𝜌𝑢𝑐𝑃 𝑑𝑉 𝑚𝐷0 𝐶𝑃𝐷 + 𝑚𝐹0 𝐶𝑃𝐶 + 𝑚𝑆0 𝐶𝑃𝐶 𝑚𝐷0 𝐶𝑃𝐷
+1
𝑚𝑆0 𝐶𝑃𝐶
Now, we will define the bulk temperature. So, bulk mean temperature can be defined as
integral per unit cell 𝜌u Cp Tdv divided by integral for a unit cell ρ u Cp dv. So, in an
average, we can write this, because ρ u is ρ udv refers to the mass flow rate. So, we can
285
write this as m dot for the dispersed phase. Actually we can consider three phases or
consider three zones, consider droplet or bubble film and slug
M dot d CPD and temperature of the dispersed phase plus m dot film Cp of the continuous
phase, because film is in the continuous phase only, and this is temperature of the film plus
m dot slug cp, continuous phase temperature of the slug divided by m dot d CPD plus m
dot F CPC plus m dot s CPC. So, we can neglect the flow rate in the film. So, we are
neglecting the mass flow rate in the film and; that means, we say that m dot film is small
compared to m dot D, not m dot slug
𝑚𝑇𝐷 + 𝑇𝑆
𝑇𝐵 =
𝑚+1
So, we do that then we have this relationship turning as m dot D CPD divided by m dot s
CPC td plus Ts divided by m dot d CPD over m dot s CPC plus 1. Now this is
approximately equal to m, why approximately, because we have neglected the mass of the
film in both the cases for the continuous phase. So, this Tb is mTD plus Ts over m plus 1.
So, one of the approximations that we have made in this case is that we have neglected the
mass flow rate in the film ok, which is quite ok. If we are using for gas liquid flow for, but
for liquid-liquid flow or for film that surrounds the slug that is not or that may not be the
case
286
So, now we will be looking at the heat transfer in the droplet regime. So, the picture that
we have in this case, is like this. We have a continuous film in which there is a stream of
continuous phase, which is slug and dispersed phase which is drop or bubble move. Now
we can treat this, the films in the slug and drop.
So, it is a very clear the that there is a film surrounding the droplet or surrounding the
bubble, because this is of continuous phase; however, for the slug one need to think a bit,
because the liquid recirculates mainly in the middle region and then beyond this
recirculating region, there is a small film and this film, the thickness of this film may or
may not be same of the film, surrounding the droplet or bubble, but for our analysis we
will assume the two films to be equal to simplify to simplify things
𝑇𝑊𝐷 − 𝑇𝐷 = (𝑇𝑊𝐷 − 𝑇𝐹𝐷 ) + (𝑇𝐹𝐷 − 𝑇𝐷 )
So, what we are going to look at the in this case is, account for the resistances in the droplet
regions. So, we will draw it again, the channel wall, a droplet and the heat entering from
the wall, the heat flux being qw and then it comes into the film and from film, it enters in
the droplet region and so on. So, the temperature difference T wall in the droplet region.
So, the temperature of the wall is TWD minus T droplet which is temperature of the droplet
we can also write this as this is the film. So, the film temperature we write as TFD or the
temperature at the interface between the film and droplet
So, we can write this as TWD minus DFD plus TFD minus TD. So, we have just subtracted
and added the film temperature. Now we can write this T wall minus TFD is equal to q
wall; that is the heat coming in divided by the heat transfer coefficient. So, delta T is equal
to q wall over h and that will be h of wall in the droplet region. we will neglect any
difference. So, we will we are going to assume hwd is equal to hw slug regime. So, and
that is when we do this, when we write this equation we also neglect convective effect.
𝑞𝑊 𝑞𝐹𝐷
= +
ℎ𝑊𝐷 ℎ𝐹𝐷
1 𝑞𝐹𝐷 1
𝑇𝐷 = 𝑇𝑊𝐷 − 𝑞𝑊 [ + ]
ℎ𝑊𝐷 𝑞𝑊 ℎ𝐹𝐷
So, when we neglect the convective effect, basically we are writing. So, we will do that
later on anyway, but by doing this approximation what we are saying that delta f in the
287
slug region will be equal to the delta f in the droplet region. So, that is qw over hw in the
droplet region plus TFD minus T D. So, TFD minus T D will be qFD divided by h FD.
now qFD over hFD. So, we can write this expression as td is equal to TWD minus qw into
1 over hw plus q Fd over qw 1 over h FD. What we have done is? we have taken T D on
this side and wrote it like this Tw minus qw has been brought out
So, 1 over hw we have just approximated is, this hw dropped D subscript here plus qFD
over qw, because qw has come out. So, we have a qw in the denominator into 1 over hFD.
Now we need to write this q FD. So, because we knew that qd over qw is equal to m over
m plus 1. You can remind yourself from here qd over qw is equal to m over m plus 1. So,
that gives us q FD over qw is equal to q FD divided by the droplet region. So, that will be
2 pi R length of the droplet
Remember this R is, we are approximating R is equal to rRdroplet. So, we are neglecting
the film thickness in this case divided by qw over 2 pi, this is qw for the entire unit cell.
So, that is why we have 2 pi R L u c. So, 2 pi r 2 pi r will be cancelled out and what we
have is qFD over qw into Luc over LD, and we can substitute qFD over qw as
approximating that qFD is about same as qD. So, what have we just talked about in this m
in a minute m over m plus 1 L u c over LD
Now, why qFD over qD. So, the heat what is qD? qD is the heat that is being entered in
the droplet or the droplet can have heat entering from the film region and the two caps. So
by doing that we have neglected heat transfer from bubble or because we are using
droplets; so, from droplet caps at the front and rear ok.
288
1 𝑚 𝐿𝑈𝐶 1
𝑇𝐷 = 𝑇𝑊𝐷 − 𝑞𝑊 [ + ]
ℎ𝑊 𝑚 + 1 𝐿𝐷 ℎ𝐹𝐷
1 1 𝐿𝑈𝐶 1
𝑇𝑆 = 𝑇𝑊𝑆 − 𝑞𝑊 [ + ]
ℎ𝑊 𝑚 + 1 𝐿𝑆 ℎ𝐹𝑆
So, when we substitute this, then we will have temperature of droplet is equal to T wall d
minus qw 1 over hw plus m over m plus 1 Luc over LD 1 over h F droplet. Similarly, for
slug one can write T slug is equal to T wall slug minus q w 1 over hw plus 1 over m plus
1. So, we have this as because now it is slug. So, the heat will be 1 over m plus 1 and LUC
over Ls 1 over hF slug ok. So, we have a relationship between the temperature differences
and all this.
289
𝑚𝑇𝐷 + 𝑇𝑆
𝑇𝐵 =
𝑚+1
1 𝑚+1 𝑚2 𝐿𝑈𝐶 1 1 𝐿𝑈𝐶 1
= [(𝑚 + 1)𝑇𝑊 − 𝑞𝑊 { + + }]
𝑚+1 ℎ𝑊 𝑚 + 1 𝐿𝐷 ℎ𝐹𝐷 𝑚 + 1 𝐿𝑆 ℎ𝐹𝑆
𝑞𝑊 𝑚 + 1 𝑚2 𝐿𝑈𝐶 1 1 𝐿𝑈𝐶 1
𝑇𝐵 = 𝑇𝑊 − { + + }
𝑚 + 1 ℎ𝑊 𝑚 + 1 𝐿𝐷 ℎ𝐹𝐷 𝑚 + 1 𝐿𝑆 ℎ𝐹𝑆
𝑞𝑊 𝑚 + 1 𝑚2 𝐿𝑈𝐶 1 1 𝐿𝑈𝐶 1
𝑇𝑊 − 𝑇𝐵 = { + + }
𝑚 + 1 ℎ𝑊 𝑚 + 1 𝐿𝐷 ℎ𝐹𝐷 𝑚 + 1 𝐿𝑆 ℎ𝐹𝑆
𝑞𝑊
ℎ𝑜𝑣𝑒𝑟𝑎𝑙𝑙 𝑜𝑟 ℎ𝑈𝐶 =
𝑇𝑊 − 𝑇𝐵
1 1 𝑚 2 𝐿𝑈𝐶 1 1 𝐿𝑈𝐶 1
= +( ) + 2
ℎ𝑈𝐶 ℎ𝑊 𝑚 + 1 𝐿𝐷 ℎ𝐹𝐷 (𝑚 + 1) 𝐿𝑆 ℎ𝐹𝑆
Now, next that the substitute this slug and droplet temperature in the bulk temperature. So,
we had this bulk temperature Tb is equal to m TD plus Ts divided by m 1 plus 1. So, we
can substitute the two values from here, and we will get m over m plus 1. So, we can write
this is 1 over m plus 1 and we will have m plus 1 Tw
Ideally we will have, we add these two terms m TWD plus T ws. So, another
approximation we are going to make here, that was what they have made is in their
290
correlation is T wd is equal to T wall slug is equal to T wall. So; that means, saying that
the temperature at the wall is same in the slug, wall temperature is same in slug and bubble
regions bubble or this approximation is more valid in the droplet region, where the heat
capacity of the two phases will be of the same order. So, the temperature will be about
same for the two phases.
So, this is m plus 1 Tw minus qw into m over hw or m plus 1 over hw plus m square over
m plus 1 luc over ld 1 over h FD plus 1 over m plus 1 Luc over Ls 1 over hFS bracket
close. Now we can substitute, we can spend this and we see that Tb is equal to T wall
minus qw over m plus 1, and this entire thing in bracket. So, we can just represent this
entire expression in the mind metal bracket things out and their expressions in copy here.
Now when we substitute this then we will get Tw minus Tb is equal to qw over m plus 1
and we can define an h overall or huc is equal to q wall divided by T wall minus. So, this
will be T wall minus q wall into this bracket, entire thing in the brackets where uc is equal
to q wall divided by T wall minus T bulk
So, from this relationship we get 1 over huc is equal to 1 over hw. As you can you can see
this m plus 1 and m plus 1 will cancel plus m over m plus 1 squared m square and we have
one more m plus 1 Luc over Ld 1 over hFD plus 1 over m plus 1 squared Luc over Ls 1
over hFS. Bracket closed. So now we can substitute, we can spend this and we see that tb
is equal to twall minus qw over m plus one and this entire thing in bracket, so we can just
represent this entire expression in the by the middle bracket to ease things out.so the entire
expression is the copy here. Now when we substitute this then we will get tw minus tb is
equal to qw over m plus 1 and we can define an h overall or hUC is equal to qwall devided
by T wall mins so this will be the T wall minus q wall into this bracket and entire thing in
the bracket so hUC is equal to q wall devided by T wall minus T bulk so from this
relationship we get one over h UC is equal to one over hw as one can see this m plus 1 and
m plus 1 will cancele plus m over m plus one square m squared and we have one more m
plus one LUC over L D one over hFD plus one over m plus 1 square LUC over Ls one
over h FS So, this is a relationship between the overall heat transfer coefficient wall. The
heat transfer coefficient at the wall and m which is the ratio of the thermal mass of the two
phases slug lengths entire unit cell length and the heat transfer coefficients for the FD and
hFS
291
So, if you look at now, what we need is, that overall heat transfer coefficient is a function
of three heat transfer coefficient h wall or the film heat transfer coefficient h film to droplet
h film to slug and m which is ratio of the thermal mass and Luc over Ld and Luc over Ls
what else yeah. So, these are the factors. So, these things we can know from the
characteristics of the flow. Now what we need to know are these three coefficient hw, hFS
and hFD and if we know, then it becomes a closed system and we can calculate the overall
heat transfer coefficient for the Taylor flow region.
292
So, as we have already discussed that we have assumed in this case, that the wall heat
transfer coefficient for the slug and the droplet regime is assumed to be same, and we can
consider only conduction. So, hw is equal to k over delta film, where k is the thermal
conductivity of the convective phase. So, we neglect any convective effect in the film. So,
the pure conduction will give heat transfer coefficient is thermal conductivity divided by
the diffusion length.
So, this is delta F and delta F one can obtain from bretherton’s correlation or ASSILOUS
and KOAK relation which is 1.34 Ca, which is capillary number raised to the power 2 by
3 into 1 plus 3.35 Ca raised to the power 2 by 3 Ca is capillary number defined as mu u
over sigma ok
Now, for slug heat transfer coefficient for hfs, because if we look at the flow, the flow in
the slug which is moving, it has an analogy with the developing flow. So, is similar
correlation that is used for the developing flow in the liquid only flow case or this liquid
flow in the channel it is used, which is for the h boundary condition 4.364 which is Nu for
fully developed liquid only plus a 1 by Ls star plus a 2 Ls star raised to the power 1 by 3.
So, where Ls star is inverse nusselt number or Ls over d Renaults number Prandtl number
ok.
293
So, by fitting their data for the experiments. So, it has been found that the values of a 1
and a 2 are 0.171 and 0.0663, and so that gives that correlation, gives the heat transfer in
the liquid slug.
Using similar arguments, they have also obtained a relationship for the, this is not fully
developed. So, this is film to droplet. So, the Nusselt number is 4.36 for the same
correlation, but a 1 and a 2 these are same as for liquid and the flow ok. So, one can obtain
Nusselt number for the liquid only from this correlation, and then it is become system of
equation becomes closed. So, one can calculate the overall or over all Nusselt number for
a in a unit cell in the Taylor flow regime
So, basically in this lecture what we have done is, we have looked that the a simple model
based on the fundamental principle for the overall Nusselt number in a Taylor flow regime,
or in a unit cell which consists of a droplet or a bubble and the liquid slug. The model is
based on the assumption that there is a continuous thin film of liquid on the wall and the
droplets and slugs. The stream of droplets and slug move, move over it; there are a number
of assumptions that we have made along the, because to make a system of equations to be
closed one of them, is that the film thicknesses in the around the slug and around the
droplet they are same. This may not be the case every time
Another assumption that has been made that the heat transfer to the droplet phase is
happening from the film or from to the slug phase, is happening from the film only, and
294
there is no interphase in exchange at the interface between the two phases ok. So, once we
get to calculate this, we need to calculate the heat transfer at the wall, which can be done
by assuming only pure conduction in the channel wall and the correlations for heat transfer
coefficients in the droplet. And in the slug flow regime they are calculated based on the
developing flow correlations for liquid only developing a correlations for developing flow
for liquid only flow in a channel ok.
Thank you.
295
Lecture - 15
Taylor Flow: Mass Transfer
In this lecture we are going to talk about mass transfer in Taylor flow. As we have seen
that the Taylor flow, or segmented flow or what we call so called slug flow this can occur
in gas-liquid flow in micro channels or liquid-liquid flow in micro channels.
So, we will try to cover both the, mass transfer or mass transfer for gas-liquid mass transfer
and liquid-liquid mass transfer. Additionally, or the reactions which are catalytic in nature,
catalytic in nature and require a solid catalyst in many instances the channel wall is coated
with a layer of the catalyst. So, we have a gas-liquid solid or liquid-liquid solid mass
transfer also happening in such cases.
So, we will cover all such cases here. One might try to have an analogy between the gas-
liquid heat transfer where the heat transfer is happening from wall to the liquid and to the
gas phase or from wall to the liquid that is continuous liquid and the dispersed droplet
phase. Similarly, when we have solid, liquid, gas mass transfer or three phase mass
transfer.
Then there is interaction or there are mass transfer resistances in between gas and liquid
phase and liquid and solid phase as well as gas solid or phase via the thin film that
surrounds these bubbles. So, it is it will be good to have an analogy and try to have a
unified model which can be Tailored according to with the mass transfer, heat transfer
need.
296
So, because this course I am not sure how many of you are from the chemical engineering
background. And in the chemical engineering background also to understand everything
in this course one need to have a background in fluid mechanics heat transfer and mass
transfer.
So, we are going to revise some of the basics of mass transfer to start with and then we
will look at the specific details of mass transfer in Taylor flow regime. So, mass transfer
is basically transfer of a solute from one phase to another it can be in a single phase also.
But what we are talking about that the how the solute is distributing in the bulk of the
liquid so that can be the mass transfer form in a single phase. But what we are talking
about mass transfer between gas-liquid or liquid-liquid phases.
So, mass transfer is generally driven by the concentration gradients or to put it more a
specifically in thermodynamic from thermodynamic point of view mass transfer occurs
because of the chemical potential or gradient in chemical potential.
So, it can have as we can see here that it can be the transfer of solute from one fluid to
another or for example, absorption of gas in the liquid phase or dissolving of a fluid in
another at which is what I was talking about the absorption of gas in the liquid phase.
And it can accompanied it can be accompanied with chemical reaction or it can be

accompanied with chemical reaction or it can happen without chemical reactions. So,
297
without chemical reaction some of the examples are say for example, gas absorbed in a
liquid.
So, for example, CO2 absorption in water or CO2 absorption in amines or sodium
hydroxide for that matter and number of liquid-liquid extraction examples are there. Then
drying and distillation so or drying is also a form of mass transfer it is the evaporation and
then or the taking out the moisture from one phase and then distillation.
Then mass transfer with chemical reaction. So, there are number of chemical reactions
some of those are say halogenation, polymerization, and nitration to name a few.
So, in conventional chemical engineering or in the industrial practice a different a number

of contactors or reactors are used. So, there are two terms here where we as we see here
contactors and reactors.
Basically they can be used interchangeably the difference in the naming is because when
there is only contact between the phases, but there is no reaction. So, there is no point in
calling this as a reactor and we term it as contactor and then there is reaction involved then
we call this as a reactor.
So, different reactors have been used or are being used in the industry and more and more
reactors where the process can be intensified process intensification can happen a number
of different reactors are being developed with such a purpose.
298
So, for example, is stirred tank reactor in the stirred tank reactor by the stirrer or by the
motion of the stirrer it can be a different kind of stirrer can be present there. So, they induce
a mixing of the two phases one or the dispersed phase is broken down into a smaller
droplets or the smaller bubbles and that gives larger interfacial area and the reaction
between the two bubble column reactor in the bubble column reactor a number of
variations of bubble column reactor are present.
So, in a column of liquid which may be a stationary or which may be also flowing. So, it
depending on that it will be a continuous gas-liquid reactor or semi batch reactor in which
one phase is stationary or in the batch mode and the other phase gas is moving through the
column.
So, in the bubble column reactor basically in a column of liquid the gas is introduced it
can be introduced at the bottom it can be introduced from the top and that the bubbles
when they are introduced a number of bubbles are generated there may be million of
bubbles and then they have a plume, bubble plume generated large interfacial area in
addition they also cause the mixing in the liquid phase. So, large interfacial area mixing
and they are used for number of industrial applications.
Then packed bed reactors, so in the packed bed reactor as the name denotes that in a
column the void is filled with the in solid bed for a catalytic reaction this might be catalytic
or it might be a packed bed that may be a static bed it may be a fluidized bed.
So, on the surface of these catalysts the reaction and then voids they cause the reaction
between the different pages for example trickle bed reactor. Now next example is plate or
tray columns for example, distillation column where there are the plates segment the
different sections of the column and mass transfer happens at each stage in between two
plates and so on.
Then film reactors so in film reactors basically generally used for gas-liquid reaction or
gas-liquid purposes where a continuous liquid film flows on the wall of the channel and
gas flows to it. So, the reaction happen at the gas-liquid interface and there are number of
other variations of these reactors in conventionally.
299
So, the contacting pattern in these reactors may be by bubbling as we have seen in bubble
column or a film reactor filming or a spraying. So, by a spraying the small droplets of the
mass transfer can happen of say for example, further when the drying of milk formation
from milk to milk powder this kind of reactors are used and say another application can be
the coating of different pharmaceutical granules etcetera.
So, then the distributing two phase flow stream. So, devising the ways by which the two
phase flow stream can be mixed or the turbulent structure or the mix mixing structure in
the fluids can be created. So, there can be different contacting patterns and that reactors
can be classified based on this contacting patterns.
300
So, contact between the phases is important because the mass transfer between the two
phases happens by the interface so it is important to have a large interfacial area ok. So,
that is where we have a micro channels coming to picture.
Now when we talk about reactions or for reactions especially reaction in multi phase then
there can be two kind of kinetics or two extremes of kinetics in one case where the reaction
kinetics is slow in another case the reaction kinetics is very fast and rest is in between the
two limits.
So, because for the reaction to happen the gas and liquid or liquid; liquid or the phases
which are to react this will come into contact first, there are two kinds of resistances in this
case one is the resistance for the reaction. So, we can say that one over the reaction rate
that can be the reacts are the resistance by the kinetics and the resistance by the mass
transfer.
Now as we know then when there are reactions in chain then the slowest step or which
have the smallest rate will be the rate limiting step. So, if the reactions are slow or
moderately slow then mass transfer will not be a problem because the phases will mix and
then there will be enough time for the reaction to happen.
However, if the reaction is fast and the reaction is fast as soon as the two phases come into
contact reaction will happen and the rate of the reaction will be limited by the mass
301
transfer. So, in such cases where the reaction rate is limited by the mass transfer which
will happen where the reaction chemistry or reaction kinetics is very fast the overall rate
is controlled by mass transfer.
And as we know that micro reactors have very large interfacial area density. So, per unit
volume they have very high interfacial area so that is where micro channels or micro
reactors can know why intensification to the process can be used as a tool for
intensification for increasing the efficiency or the performance yield selectivity of the
reactions with fast kinetics. Additionally, in the conventional reactors most of them it is
not possible to control the bubble or droplet size precisely.
For example, take a bubble column reactor you have a sparger and the gas is coming from
it the bubbles will be generated depending on the local fluid conditions the size of the
bubbles at the center and at the corner near the wall will vary.
Now these bubbles will interact with each other and they will coalesce or break up and
there will be a size distribution of bubbles which can vary an order or more. So, it is very
difficult to get an accurate idea of the interfacial area density in such cases and there will
be non uniformities so it is very difficult to have control over the precise control over the
happenings that are happening in the reactor.
So, in such cases Shatila flow regime where we have bubbles of a are bubbles of a size or
a particular size we can generate or control the size of the bubble, or size of the droplet
and the reaction can happen continuously in these size of in these regular sized bubbles or
the uniform bubble sizes.
So, we can have a very precise control over the happenings over the activities that can
happen in the in a micro reactor so that is advantage of micro reactors.
302
So, we are here we have listed down the different advantage or issues that are relevant to
micro reactors. So, in micro reactors we have interfaced area to be large just to remind
ourselves if the channel size is one-meter area over volume is one of the order of one.
And for one meter and if this becomes 1 millimeter then the same number becomes 103, it
should becomes 0.1 mm the same number will become 10 raised to the power 4. So, as
compared to one we can get interfacial area of 10 raised to the power 4 in micro reactors.
So, of course, a lot of advantage in terms of interfacial area then avoiding heat hotspot
formation so when the reaction is exothermic and large amount of heat is being generated
during the reaction then where the reaction happening in these zones the the high
temperature may occur and the hot spot may form.
So, in micro reactors as we have discussed in the previous lectures the heat transfer one
can precisely control the temperature in the reactors. So, one can avoid the hot spot
formation and it is also easier to or to carry out these exothermic reactions safely in micro
reactors.
Then one can handle specially the hazardous materials or a specialty chemicals fine
chemicals which cost a lot. So, these coming especially in the pharmaceutical industry or
consumer goods industry or so in these cases one can work with because the micro
channels size is small.
303
So, one can work with the small quantities and mitigate the risk of handling hazardous
materials in their speciality chemicals one can work with smaller nano litre or the pico
litter size of the droplets and carry out the reactions and find out the I mean do the
analytical chemistry or whatever formulations or the reactions are to be done especially in
pharmaceutical applications one can do.
So, micro reactors are useful there then the always question will come that all this is fine,
but micro reactor is for the same region that it handles very small amount of volume. So,
how do we scale this up so scaling up micro reactors I will again give the example of same
printed circuit heat exchangers that we talked about in the heat transfer case.
So, the same concept where a sheet of metal it can have channels edged straight channels
or meandering or tortuous channels etched on this plate if the number of channels may be
1000 and then stack those plates together. So, one has 1000 micro reactors or hundreds of
micro reactors and then hundreds of these plats plates stacked up together.
So, if one can distribute the gas and liquid phases or liquid-liquid phases or whatever the
liquids one is working with uniformly in these micro channels then the scaling up can be
achieved without really changing the size of the channel.
So, the by scaling of achieved the scaling up in micro reactors is achieved by numbering
of the channels by numbering of the units that we are working with rather than scaling up
the reactor from lab scale to pilot scale to the industrial scale.
So, in with the with these promises of chemical micro processing or micro reactors a
number of applications or number of researchers has focused on understanding the mass
transfer, chemical reaction, reaction engineering in these micro reactor channels in past
two decades not only from the academic word from the industry.
And some of those has started to come up in the industrial production also. So, so for
example, in the pharmaceutical industry the impetus has been to use micro channels for
the continuous production. Because in the most of the reactors that are used in the
pharmaceutical industry they are batch reactors.
So, and batch reactors they have its own disadvantages. So, it will be advantageous to
convert these processes into continuous processes, but the volume that is handled in these
304
reactors may not be high or in these industries may not be high or they need to use the
same system for different processes. So, this can be easily achieved using micro reactors
and this has been or this has been increasingly explored by the pharmaceutical industry.
So, coming back to micro reactors specially for gas-liquid or liquid-liquid flow Kashid and
Minisker I think I have a spelling mistake here. So, Kashid and Minisker they divided the
micro mixers into or micro channels or micro reactors into three different categories based
on the contacted principles micro mixers, micro channels, and falling fall film reactors.
So, in the micro mixers some of the examples are mixer settler, or inter digital mixer, or
cyclones. So, if they can also have a static internals. So, which gives rise to chaotic mixing
or in micro channels where the mixing is basically either by diffusion or convection. So,
one can have say different kind of Y shape channels or T shaped channels or micro packed
bed reactors.
Channels with mesh or membrane contactor the other option is falling film micro reactor.
So, in this case a liquid film falls on the walls of the channel and the gas is flowing either
from the bottom or from the top. So these are three different categorizations for micro
reactors.
305
And now we will look at some generic features of the mass transfer models. Now there
can be or there are two different kind of models because one need to look at the mixing
within the phase itself once the transfer through the interface has happened then the solute
or the liquid or the gas phase will mix in the liquid phase so that comes under macro model
so, that is mixing within a phase.
And then before that when the mixing happens at the interface one need to look at mass
transfer between the phases. So, different micro models what are called what are this called
mass transfer between phases at the interface the models developed or called micro
models.
306
So, looking at the micro models which are the specific to two phase flows there are
different models that are being used. So, the first one is stagnant film model so it assumes
that there is a, at the interface there is a stagnant film which is imaginary.
So, this film are don’t really exists but the assumption is that this film and in this stagnant
film the mass transfer proceeds via molecular diffusion just next to the interface and the
remaining fluid remains well mixed. So, one has the concentration gradient here and then
it is well mixed that is what the assumption in the stagnant film is.
So, the mass transfer coefficient in the stagnant film KL in the second phase is given as D
where D is the diffusivity or molecular diffusivity or molecular diffusivity divided by delta
interface.
So, the thickness of this interface one can also look at the units the unit of D is meter square
per second and the unit of delta is meter. So, the mass transfer unit in this case is meter per
second so it because this is in imaginary layer surrounding the liquid. So, the question in
such models generally come what is the thickness of this film.
So, the stagnant film model requires to determine or it is required to determine the
thickness of this film.
307
Another approach that is used is called penetration model and this is due to Migbie and so
in this basically what it suggested that each liquid elements take the same time in contact
with the fluid or each element of surface.
So, surface means the interface. So, consider a bubble that is rising in the liquid. So, a
liquid element will come in contact with this bubble at this point and then it will move on
the interface of the bubble and then eventually we will leave the.
So, what the assumption of penetration model is that liquid at the interface is continually
being renewed. So, at one point there was one molecule of liquid, but the next time it
moves here and another liquid come here. So, one molecule of liquid or one liquid particle
or fluid particle will remain in contact with the interface for a certain time and then it will
be it will leave the liquid.
𝐷
𝑘𝐿2 = 2√
𝜋𝜃
So, this liquid is being continuously replaced by a new liquid. So, what they have
suggested that the mass transfer coefficient is proportional to the square root of diffusivity
divided by 𝜋 𝜃 where θ is the exposure time or contact time of the liquid to get the gas
phase.
308
Now this approach seems to be more realistic for dynamic system where the the bubble or
the dispersed phase or that the relative velocity between the two phases.
It has been suggested that the choice of the penetration model and film model can be
determined by a non dimensional number or Fourier number which is defined as the ratio
of diffusive or molecular diffusion divided by the storage depth.
So, molecular diffusion is given by the diffusivity or molecular diffusivity and the storage
rate is the length scale. So, for a channel it will be the channel radius or diameter and r
square over 𝜏 so, τ is a typical time scale.
So, the ratio of this is known as Fourier number and if the Fourier number is less than 1
that means the diffusion rate which is smaller than the storage rate then it is suggested to
use the penetration model. If it is other way round that if the diffusion rate is high, then
one can assume the film to be a stagnant and use a stagnant film model.
So, there are versions in which one uses a combination of film and penetration theory so
there are number of other theories that can be based around this or further modification to
these theories.
309
Now, consider the mass transfer at an interface. So, if this is the interface between two
phases and if we consider that there is considerable resistance on both sides of the interface
so we plot the films surrounding this and this is delta int 1 we can call in phase 1 and this
is delta int 2 in phase 2
The concentration in phase 1 is C 1 and partial pressure in this phase is p 1 and that is
decreases to the a concentration let us call this as C 1* p 1*.
And then in the second phase the concentration is C 2 and the interface the concentration
is C 2 and the interface. So, at equilibrium and steady state the mass transfer trough
interface from the the rate of mass transfer from phase 1 to phase 2 will be equal so the
rate of mass transfer will be equal.
∗ ∗
𝑘𝐺1 (𝑝1 − 𝑝1,𝑖𝑛𝑡 ) = 𝑘𝐺 (𝐶2,𝑖𝑛𝑡 − 𝐶2 )
So, what is given as is as k G 1 p 1 minus p 1 interface * is equal to k G 2 C 2 at the

interface minus C 2. Now we need to relate this and the p 1 into * and C 2 into * and so this
is given by Henry’s law.
∗ ∗
𝐶2,𝑖𝑛𝑡 = 𝐻𝑝1,𝑖𝑛𝑡
310
It suggests that C 2 is equal to H p 1* interface. So, a H is called Henry’s constant for this
is for gas-liquid systems for liquid-liquid H is replaced by K and K is called partition
coefficient.
So, some more basics about the mass transfer the the reactor because the mass transfer is
a function of mass transfer coefficient as well as the interfacial area so, the reactor
performance is generally given as k multiplied by interfacial area density. So, it may be a
so k gas to liquid interfacial area gas to liquid. So, a specific interfacial area that is
interfacial area per unit volume meter square per meter cube k has unit of meter per second.
So, the total unit will be meter per second divided by meter now some non dimensional
numbers. So, Sherwood number is an analogue to Nusselt number in heat transfer.
So, the Nusselt number in heat transfer if you remember it is convection over diffusion or
conduction H D over k similarly
𝑘𝑑
Sherwood number is 𝐷
or so where d is a length scale and k is the mass transfer coefficient and the D is diffusivity
Schmidt number is an analogue to Prandtl number in heat transfer.
α cp 𝜇
So, Prandtl number is 𝑘
311
𝜈
So, that is 𝛼
𝜈
Schmidt number is defined as 𝐷
where 𝜈 as you will know that μ over 𝜌 so and the and D is the molecular diffusivity both
of them have unit meter square per second. So,
𝑘
α is thermal diffusivity which is defined as ρ cp
D is molecular diffusivity or mass diffusivity and ν is momentum diffusivity.
Another important non dimensional number which is relevant during the reactions, which
gives a ratio of or
Hatta number square is a ratio of diffusion time to the reaction time.
𝜏𝐷
𝐻𝑎2 =
𝜏𝑟
So, 𝜏𝐷 is diffusion time, and 𝜏𝑟 is the reaction time. So, this is important that we have react
mass transfer coupled with the reactions.
Now, when the experiments are conducted different methods are used to measure the mass
transfer. So, mass transfer will be basically measured by measuring the concentrations
312
either locally or at the inlet and outlet. So, every time we want to do the there are two
different methods one is offline method. So, for the offline method one take a sample or
one has the known concentration at the inlet and then the concentration at the outlet has to
be determined.
So, because it is multi phase flow so we have to or we might need to determine the
concentration in the two phases and we need to separate the two phases. So, one need to
look at a phase separation method and this phase separation has to be quick because the
reaction times or the contact time in micro reactors is short so if the separation time is
longer than the accurate results will not be obtained.
So, the phase separation has to be quick one of the natural way to do it separation of
different wettability of the liquid to the channel wall. So, that will require lot of
understanding of the interface or the wettability behaviour of liquid different liquids with
the channel wall ok.
So, some of the efforts has gone in this direction the other challenges that because what
we know is the concentration at the inlet.
So, to because as a channels for example, if one uses a T junction channel then the changes
in the concentration just before they mix and when this happens so the mass transfer will
be very high in this region. So, we will not be able to get the mass transfer in the fully
developed region or where the hydrodynamics has well developed and a structured pattern
has been obtained.
So, one of the options people have looked at that they have devised two channels in one
case the study the mass transfer in the entire length of channel and in the another case the
mass transfer studied just in the inlet section so that one can know how much mass transfer
or what mass transfer has taken place.
In this inlet section and then subtract it from the total mass transfer and of course, once the
sample has been obtained at the outlet it has to be analyzed or using different analytical
techniques chromatographic or spectrometric balances.
313
In the online measurement most of these online measurement techniques which means that
the concentration is being measured as the fluids move in the channel so, these online
measurement techniques most of them are based on the optical measurements.
So, for example, one can take the pictures if for example, gas-liquid flow is happening and
CO2 is being absorbed in any wedge then the concentration of or the size of the bubbles
will change because the CO2 is being dissolved in the fluid.
So, as one move from the inlet to the outlet the size of bubble will keep reducing so, if one
can take the image of the channels at different lengths then one will have by estimating the
size of the bubbles one can have an idea about the mass transfer that has happened from
gas phase to the liquid phase.
So, the other option is to use different indicators say for example ph indicators which can
monitor the colour change so by monitoring the change of the colour at different lengths
so one can make an estimate of the change in concentration a relatively recent, but very
promising technique is LIF or liquid induced fluorescence where dyes fluorescent dyes are
available which are sensitive to the concentration.
So, these concentration sensitive dies are introduced and by the laser light the reflected
light is the intensity of the reflected light or the is read by the images and then an estimate
of the concentration field can be made.
314
So, now coming to mass transfer in the Taylor flow regime it is extensively used in most
of the mass transfer applications in micro channels for gas-liquid or liquid-liquid flow.
Taylor flow regime or the segmented flow is one of the most popular flow regime in which
work with because it provide interfacial area which is large as well as which is well defined
interfacial areas.
Another advantage of this is that the axial dispersion is reduced for example, if we have a
bubble and two slugs surrounding the bubble and there is very little interaction you almost
no interaction between slug 1 and slug 2. So, there is no dispersion we can treat these two
streams separate with each other and there is no interaction between two stream.
So, there is no axial dispersion for these two streams and this has intense mixing. So, if
number of number of bubbles are in you can think of these number of slug reactors in
series which will eventually be a plug flow reactor.
And then of course, the diffusion lengths are small and there is internal re circulations. So,
the radial mixing is also enhanced.
Another question which might come in general in all of us we can have this question that
the bubbly flow regime the size of bubbles is very small. So, why not use bubbly flow
regime where one can use or when one can get very high interfacial area.
315
And bubbly flow regime has been classified in different literature from small bubbles they
are also called bubbly flow regime or spherical bubbles or almost a spherical bubble which
occupy the entire channel they are also called bubbly floe regime.
So, this will be the case when the volume fraction is small and there is very small. So, in
this case the interfacial area will be smaller. If we have high gas holdup one may have
smaller bubbles interfacial area might be high. But again no control on the bubble or
droplet size flow might be unstable and then there is the it is catalytic reaction then the
diffusion lengths will be larger.
So, one the most attractive flow regimes in micro channels is Taylor flow regime the
annular flow regime because again if it is liquid-liquid flow. One can have parallel flow
and can have the interaction between the phases, but for the gas-liquid flow the annular
flow is generally unstable there are instabilities present on the interface so that can be a
problem for a number of cases.
So, mass transfer in Taylor flow there can be number of different type of problems that
one may want to study or different applications can be there. So, one is that there is mass
transfer between the gas and liquid phases. Another one is mass transfer between gas-
liquid and solid catalytic wall and this mass transfer can be with or without reactions.
316
So, the mass transfer mechanism in such cases will be a can be complicated. So, let us look
at the mass transfer we have the systems have reaction on the channel wall. So, if we have
a solid wall and the Taylor bubble into it so there are two paths for mass transfer from gas
to for the gas to go to the solid one is via a liquid slug.
So, the gas interacts from these gaps with the liquid slug and then slug moves to the from
the slug the gas comes in contact with the solid phase. So, the first contact is gas-liquid
mass transfer at the interface via the front and rear bubble caps.
And then transfer of dissolved gas molecules. So, the gas that has dissolved it transfer from
here to the wall the other path is via the wall. So, it goes via the film so the gas dissolves
in this liquid film and then from film it moves towards the wall. Generally, this film is
almost a stagnant. So, it is one can use the stagnant film model there.
317
This is thought to be a dominant mechanism especially when the bubble is longer than the
interfacial area of the film region will be large compared with the interfacial area and those
caps.
However, in some cases if this liquid film becomes saturated then there will be no mass
transfer via the liquid film.
318
So, one need to look into all these cases now for systems with reaction on the channel wall
so what we have looked at is gas to solid mass transfer. Now another important mass
transfer that one need to look at from liquid to solid.
So, that will basically be in the slug phase the liquid interacts with the solid. So, again
though there is only one liquid but as we know from the hydrodynamics that this have
liquid slugs on it and surrounding the slug there is a thin film of liquid.
So, one can take two resistances in series the first resistance to mass transfer when it is
circulating in this bulk and then transport through the film.
So, what we have looked at is mass transfer in case where there is a gas solid liquid system
and reaction is happening when there is no reaction on the channel wall then one just need
to look at the mass transfer.
So, the steps at the wall are no longer important, but we still have the mass transfer via
caps and the film in this case so if we have say CO2 absorption in NaOH or CO2 absorption
in water then the bubbles will be in contact with the so, the gas-liquid mass transfer from
the cap regions and the gas-liquid mass transfer from the film region.
319
1
𝑘𝐺𝐿𝑆 𝑎𝐺𝐿𝑆 = 𝑘𝐺𝑆 𝑎𝐺𝑆 + ( )
1 1
+
𝑘𝐺𝐿 𝑎𝐺𝐿 𝑘𝐿𝑆 𝑎𝐿𝑆
So, one can write that the total gas to solid mass transfer or k let us say G L S into a G L
S will be equal to there are two parallel steps that the mass transfer happens via the liquid
film and mass transfer via cap.
So, for this we can write k G S a G S plus there are two steps in series. So, for which we
can write this the steps in series that mass transfer via a k G L a G L plus 1 over k L S a L
S.
So, we can draw a schematic of it so things become clearer we have a bubble here and
what is called G S which gas-liquid via film and this is gas-liquid and this is liquid solid.
So, the total mass transfer gas to solid mass transfer is the combination of it can happen
via the liquid film and via caps. So, via liquid film we can have k G S a G S directly plus
it can have two steps in series at it has to a steps in series mass transfer via the caps and
mass transfer from the slug to the wall.
So, these two terms in series one can remember the expressions that we developed for heat
transfer and you can see the analogy between the two here.
320
𝜋(𝑑 − 𝛿𝐹 )2 4(𝑑 − 𝛿𝐹 )2
𝑎𝐺𝐿 = 𝜋 =
(𝐿𝐵 + 𝐿𝑆 )𝑑 2 (𝐿𝐵 + 𝐿𝑆 )𝑑 2
4
𝜋𝑑(𝐿𝑆 + 𝑑 − 2𝛿𝐹 ) 4(𝐿𝑆 + 𝑑 − 2𝛿𝐹 )

𝑎𝐿𝑆 = 𝜋 =
(𝐿 + 𝐿 )𝑑 2 (𝐿𝐵 + 𝐿𝑆 )𝑑
4 𝐵 𝑆
4(𝐿𝐵 − (𝑑 − 2𝛿𝐹 ))
𝑎𝐺𝑆 =
(𝐿𝐵 + 𝐿𝑆 )𝑑
Now, these because the bubbles can be approximated so one can easily get an idea of the
geometries or the interfacial area densities in these cases. For example, for the a G L case
where G L is the interfacial area density of the caps.
So, if we say that this film thickness is delta F and length of the slug is L S and length of
the bubble is L B. So, one can write G L, a G L is the interfacial area of the caps and
assuming that they are spherical cap so, hemispherical caps.
So, one can have π d cap squared. So, let us say this channel diameter is d and the cap
diameter will be d minus delta F square per unit cell volume. So, unit cell volume will be
π by 4 L B plus L S d square. So, π will cancel out and we will have 4 d minus delta F
squared divided by L B plus L S d squared.
321
In the same manner one can also calculate the area for the liquid solid. So, the volume of
the unit cell will remain same π by 4 L B plus L S d square now this length is the area is 2
π d into L of this slug length.
So, L S plus if we consider the entire length in this is the where the mass transfer is
happening via liquid slug then we will have L S plus d minus delta F and it will be 2 delta
F because there are films on the top and bottom so that will this will be π dl.
So, it will not be there π will cancel out and we will have 4 L S plus d minus 2 delta F over
L B plus L F into d. Similarly, one can find out the area for this film region which is called
that a G S which will be equal to liquid zone fluid get rid of this 4 L B plus L S d L B
minus d minus 2 delta F.
So, these are the difference interfacial areas that we can now we can use approximations
for example, if the film is very thin then we can neglect the film with respect with respect
to d and so on and so forth.
In summary what we have looked at is obtained the we reviewed some of the basics of
mass transfer look at the some of the non dimensional numbers Sherwood number, and
Schmidt number.
The definition of the mass transfer at the interface the film the micro models we for
example strain and fluid model and penetration theory and then we have looked at the
mechanism qualitatively the mechanism at the interface.
Now in the next lecture we will try to look at different models that have been developed
by researchers based on their experiments or from the fundamentals to develop the to give
these mass transfer for the film region, mass transfer from cap at the bubble caps from gas
to liquid slug and then mass transfer from slug to the wall ok.
Thank you.
322
Lecture - 16
Taylor Flow: Mass Transfer
Hello. So, in this lecture we are going to continue our discussion of Mass Transfer in the
Taylor flow regime. In the previous lecture we discussed about the basics of mass transfer
some non-dimensional numbers and the general features of mass transfer in the Taylor
flow regime. In this lecture we will try to use that knowledge to develop some of the
models or we will discuss some of the models that have been developed in the literature in
last 15-20 years.
𝜋𝑑𝑏 𝜋𝑑
≈
2 2
𝜋𝑑
𝑐𝑜𝑛𝑡𝑎𝑐𝑡 𝑡𝑖𝑚𝑒 𝜃 =
2𝑈𝐵
𝜋𝑑𝑏2 4
𝑎𝑐𝑎𝑝 = =
𝜋/4𝑑2 𝐿𝑈𝐶 𝐿𝑈𝐶
323
So, we will start with the mass transfer for gas liquid transfer. So, for a Taylor bubble as
we have discussed several times that one can look at a unit cell, and the bubble the best
approximation in terms of simplified geometry for the bubble can be the spherical caps
and the cylindrical middle portion.
So, for gas liquid mass transfer Van Baten and Krishna from university of Amsterdam,
they considered the mass transfer from gas bubble to the liquid phase and considered the
contributions, from the bubble caps at the front and rear and the cylindrical middle the film
region separately.
𝑘𝐿 𝑎 = 𝑘𝐿𝑐𝑎𝑝 𝑎𝑐𝑎𝑝 + 𝑘𝐿𝑓𝑖𝑙𝑚 𝑎𝑓𝑖𝑙𝑚
So, one can write 𝐾𝐿 𝑎 which is the overall gas liquid mass transfer coefficient into the
interfacial area, that is equal to 𝐾𝐿 cap into a cap plus 𝐾𝐿 film into a film. So, first we will
consider the contribution from the caps and for simplification for geometrical
simplification caps are considered to be hemispherical. The diameter of channel is 𝑑 and
let us say the diameter of the bubble is 𝑑𝑏 .
Now, in our picture of the derivation that they did for was for a rising Taylor bubble in a
capillary and then it was validated with extends the CFD models, CFD modelling. So, in
the near the Taylor bubble as we have discussed that there is recirculation in the liquid
zone.
So, the a if the flow is in this direction, then the a liquid particle comes on this side and
move and then it goes. So, the recirculation or a liquid particle when it recirculates inside
the liquid slug it comes into the contact with the this much length of the bubble.
Similarly, on the other side so, let us just consider the other side of the bubble on, the other
side when the so, this the bubble is moving in this direction. And so, then the liquid comes
in contact with the bubble interface the cap and then again.
So, in a liquid molecule the entire distance or the it remains in contact with bubble caps
along a distance, which is half of bubble cap or the circumference.
So, the average distance travelled is equal to 𝜋 d b the entire circumference this side and
this side and the half of this will be π d b by 2 and we can assume this that d b is equal to
324
dc neglecting the small film thickness. So, for thin film thickness one can have this d b is
equal to the channel diameter, which is by this. So, you write is π d by 2.
Now, contact time in this case will be so, when we have the length is π d b by 2 the contact
time theta will be length over distance. So, π d by 2 and the velocity of the bubble let us
say is U B. So, the contact time will be π d by 2 divided by U B.
Now, from the penetration theory K L cap is equal to D over π theta into 2. So, that will
be equal to 2 root of D by π pi d by 2 U B this will be equal to 2 root 2 over π square root
of D Capital D U B divided by D 2 root 2 by π d U B by d. So, remember this D is molecular
diffusivity or mass diffusivity and small diameter a small is small d channel diameter U B
is velocity of the bubble and if it is not known a priori we can take this almost to be equal
to U D B.
𝐷 𝐷 2√2 𝐷𝑈𝐵
𝐾𝐿,𝑐𝑎𝑝 = 2√ = 2√ = √
𝜋𝜃 𝜋𝑑 𝜋 𝑑
𝜋 2𝑈
𝐵
Now, the other thing we need to know is a cap. So, a cap is equal to this will be divided
by the cross sectional is divided by the volume of the unit cell. So, that will be π by 4 d
square LUC and the surface area of 2 hemispheres together put together will be one
hemisphere. So, that will be π d b squared ok. So, that will eventually be 4 by LUC with
the assumption that d is almost equal to d b. So, we have calculated the mass transfer
coefficient and the interfacial area density for the bubble caps ok.
325
𝐷 𝐷𝑈𝐵
𝐾𝐿,𝑓𝑖𝑙𝑚 = 2√ = 2√
𝜋𝜃𝑓𝑖𝑙𝑚 𝜋𝜀𝐺 𝐿𝑈𝐶
Now, coming to the contribution from the film region so, for the film region again one can
define KL film is equal to 2 square root of D over π theta in contact with the film. Now
theta film can be defined as the length of the bubble theta is contact time. So, contact time
for the film can be defined as the length of the bubble the film region divided by UB
.
And if we define epsilon G, which is void fraction or gas hold up, it can be defined roughly
equal to L B over LUC which will hold good for long bubbles and long unit cell because
the volume of the spherical caps can be neglected in that case. So, we can have LB replaced
by epsilon G LUC divided by U v. So, that gives us K L film is equal to 2 square root of
D over π and theta film epsilon G L U C and U B.
Now, we can write the total contribution. So, that is KL a is equal to KL cap plus KL cap
a cap. So, before we do that we also need to find a film. So, it is like a film now a film is
equal to the volume of the unit cell which will be π by 4 d squared L U C and the surface
area will be π dl for this cylindrical region.
326
So, that will be π d b into L B π and π will cancel out and that will be d b LB 4 divided by
d LUC and for the case and d b is approximately same as d we can write this as 4 epsilon
G LB by LUC can be written as epsilon G. So, 4 epsilon G and this is D square. So, 4
epsilon G over D. So, that is the KL and al film.
2√2 𝐷𝑈𝐵 4 𝐷𝑈𝐵 4𝜀𝐺

𝐾𝐿 𝑎 = √ + 2√
𝜋 𝛼 𝐿𝑈𝐶 𝜋𝜀𝐺 𝐿𝑈𝐶 𝑑
8√2 1 1 8√𝜀𝐺 1
𝐾𝐿 𝑎~√𝐷𝑈𝐵 [ √ + ]
𝜋 𝑑 𝐿𝑈𝐶 √𝜋𝐿𝑈𝐶 𝑑
So, we can now write the total contribution which is K L a is equal to K L cap into a L
cap. So, KL cap is let us go back to 2 root 2 over π into root D U B over d a cap is 4 over
L U C plus the contribution from the film is 2 square root D U B divided by epsilon G
LUC into 4 epsilon G over d. So, let us just check 2 root d U B by π epsilon G LUC. So,
we have epsilon G L U C π then to 4 epsilon over D.
So, we can see from here that in both the cases K L a is proportional to root D U B. So,
the overall mass transfer coefficient it is proportional to the square root of diffusivity and
it is proportional to the square root of the velocity of the bubble. The constants we have
here inside is 8 root 2 over π into 1 over root d, 1 over L U C plus 2, then that will again
327
the 8 root epsilon G divided by root pi. So, that is 4 into 8 epsilon G DUB already out. So,
you will have root π L U C into 1 over d.
So, this is basically function of channel diameter and unit cell length. So, that suggests that
the mass transfer coefficient will be higher for larger channel diameter and it will be also
higher when the units cell length is shorter will also depend on the volume fraction.
Now when they compared this model with the CFD simulations for mass transfer they had
good agreement with CFD simulations and also found, that mass transfer is dominant in
the film region.
Now, when it comes to the rectangular channels; because in micro-fluidics; people not
only use circular capillaries, but more and more rectangular or square capillaries. So, in
the rectangular channels film thickness is not a constant delta F is not uniform everywhere
and bubble caps are not necessarily hemispherical.
𝐷𝑈𝐵 1 𝐷𝑈𝐺 1
𝑘𝐿 𝑎 = 𝐶1 √ = 𝐶1 √
𝜀𝐺 𝐿𝑈𝐶 𝑑 𝐿𝑈𝐶 𝑑
𝑈𝐵
= 𝑈𝐺
𝜀𝐺
328
So, in their work they further considered a simplified model, where they consider that cap
mass transfer can be neglected and then considered the film contribution only. So, K L a
in this case will be proportional to a constant C 1. So, on the constants they were brought
together under C 1 and then that is equal to D U B by epsilon G L U C 1 over d and U B
by epsilon G it can be written as the superficial velocity of gas.
So, this is replaced by a quantity which is known, while performing experiments it is easier
to use in the models of C 1 square root of D U G over L U C into 1 over d. And what they
found is that C 1 from their experiments the comparison with the experiments in circular
as well as rectangular capillaries that C 1 is 1 about 4.5 the value from the model from the
earlier model it was about 8 divided by root π which is very close to this value.
They suggested that applicability of this model was good during their experiments and
given a criteria that this will be good when U G over U L divided by L B square root of it
is greater than 3 or the contact time is less than 0.1 or the Fourier number is less than 0.1.
Now, while using this model one need to take into account the fact that is the length of the
bubbles and slugs are very long then in that case the film might get saturated and there will
be no mass transfer to the film. So, in this case the film will remain in inactive player
during the mass transfer so, one need to use this model with a caution; looking at the length
of the bubbles and slugs that are generated for a particular configuration.
329
So, now what we have looked at is mass transfer for gas liquid interface looking at the
contributions from the bubble cap and contributions from the film region and we said that
in the limit of short contact times it is the mass transfer through the interface in the film
region, which is dominant and we gave the relationships for the 2 cases based on
penetration theory.
Now, the other region which we need to consider is mass transfer in the liquid solid where
the liquid in the slug as well as liquid in the film region may interact with the wall. So, this
will depend on the basically on the flow behaviour in the liquid slug. So, as we know that
the flow in the liquid slug has internal circulation. So, which has vortices? So, it will
depend on the intensity of these vortices and then we have this picture of or this
configuration of Taylor bubble that on a thin film the bubbles and slug move. So, it will
also depend on this thickness of the liquid film on the wall and the of course, the surface
area between the liquid and solid phases.
So, in the parameters that we study this will translate to the 2 phase velocity or U T P,
which is some of the superficial velocities of gas and liquid the dimension of the channel
d length of the slug and diffusivity D ok.
So, one can develop a correlation based on the mass transfer considered the 2 region, where
in the liquid slug region there is internal recirculations in the liquid slug and the film region
and put them put the 2 resistances in series. So, while the mass transfer from the vortex to
330
the film can be considered using the stagnant film model what is what we need to
concentrate on or what is to be looked upon is mass transfer from the film to wall.
So, as we have seen in heat transfer during the heat transfer in Taylor flow regime that
there is an analogy for the flow in the liquid slug with the single phase developing flow in
a channel. So, same will come into play here and we know that in such case the Nusselt
number or Sherwood number for mass transfer this is a function of Graetz number where
the Graetz number is a function of L over d Re Sc, Sc is Schmidt number ratio of mass
diffusivity a momentum diffusivity by mass diffusivity and also number as usual D is the
channel length.
So, there have been several correlations for defining for modelling the behaviour in the
liquid solid region some of them 2 of notable of them, we present here one is by Kreutzer
he gave that the Sherwood number is a function of Graetz number where the Graetz
number is defined as l over d Re Sc as we seen in the previous slide.
Now, it also have 2 parameters alpha and beta and this alpha this alpha and beta they are
weak functions of the non-dimensional slug length. So, we can see that the Sherwood
number and in this the L is the length of the channel so, if the slugs are long. So, this can
be L slug or L channel depending on what one is considering. So, if the slug is long then
this term will go to 0 and what we will have is as such is equal to root of alpha square and
331
where alpha because this term will again be very small compared to one. So, we will have
Sh is equal to alpha which is 40.
So, the single phase for the same boundary condition the Sherwood number for single
phase flow is 3.63. And so, one can see the one order of magnitude difference the for this
developing flow in the liquid slug the Sherwood number is about 10 times this is for the
only liquid slug.
So, if we consider for the entire unit cell with 50 percent void fractions, then also this
number will be about 20. So, which is about 5 times the Sherwood number in single phase
liquid; single phase liquid flow.
So, this is again further high then the single phase liquid flow. So, a reason for this has
been suggested by Kreutzer that in the liquid slugs what happens that there is a
recirculation and the Sherwood number is basically a non-dimensional gradient at the wall.
So, that will depend on the concentration difference divided by a length scale from the
channel wall to the minimum concentration.
So, this is channel wall to minimum concentration. In case of single phase flow the
minimum concentration will be at the centre and so, this delta will be radius of the channel.
Whereas, in this case the minimum concentration region will be located at the centre of
this vortex and if you remember from the hydrodynamics that we discussed one can
calculate the, this distance this is about r over root 2. So, if you put that will be about 0.7
r and this is 0.3 r. So, the distance of the minimum concentration for this delta is about 0.3
r.
So, this distance has decreased for the Taylor flow plus the another factor that contributes
here is that the in the single phase flow the liquid at the centre it moves forward in it has a
high velocity whereas, in the Taylor flow the liquid at the centre is also brought towards
the wall. So, the residence time of all the fluid elements in these slugs are equal whereas,
the residence time in this case is there is quite significant distribution of the residence time
in single phase flow ok.
332
(1 − 𝜀𝐺 )𝐿𝑈𝐶
𝐺𝑧 =
𝑑𝑅𝑒𝑆𝑐
So, another correlation that has been given by Van Baten and Krishna and both of these
correlations are one by Kreutzer and by Krishna they have been developed with extensive
validation with CFD modelling. So, Van Baten Krishna they have given 2 correlations one
for the wall slug region and one for the entire channel. So, this gives the Sherwood number
2.4 plus 1.5 d over L S and in this case divided by L tube 1 minus epsilon G. So, where
epsilon G is void fraction d over d square U B and for the channel this is Sherwood number
0.5 epsilon G over G z where the Graetz number in this case is slightly modified.
So, they have considered a Graetz number one minus epsilon G L channel or you can
consider L U C divided by d Reynolds number and Schmidt number. So, the difference in
this is in terms of and they have multiplied the Graetz number with 1 minus epsilon G. So,
considering effectively the length of the slug and they had quite a good agreement with
the CFD results and some of the experimental results.
So, these are the correlations for mass transfer and the liquid solid wall. So, another
interesting a region, where one would need to have models is for see liquid liquid mass
transfer, but to the best of my knowledge there is no not very well developed model
especially from the fundamental principles there are several models, which based on the
fitting to the experimental data or from the dimensional analysis. So, we are not going to
333
discuss that and I request you to look at the literature for those things. In this lecture what
we have essentially looked at 2 models for the one for the gas liquid mass transfer for
Taylor bubble at the interface where we have considered the contributions from the cap
and contributions from the film region.
Now, if the bubbles are long and the contact times are large then the film might get
saturated and it may become inactive; however, when the contact time is sought it has been
observed and the model has been developed for such case that the mass transfer from the
film region is the dominant one it will depend on the length of the bubble and the length
of the slug. So, a picture one can imagine that when the bubble is in contact with the film
the film that gets saturated with the gas and when the slug comes by this mixing happens
between the film and the slug.
Now, the other model that we have looked at is mass transfer between the liquid and solid
wall. So, one can have a series model there that there are two different regions one is mass
transfer from the re-circulating slug to the film in contact with the wall and other is mass
transfer from wall film between the film and the wall.
So, for the mass transfer between the film and wall one can use a stagnant film model,
where K L B will be equal to capital D over delta the film thickness and for the slug region
there are 2 models that have been proposed by Kreutzer and Professor Krishna Raj group
and both these model are based on the developing a mass transfer or in single phase flow
regime with some modifications to those correlations. So, that is all for this lecture.
Thank you.
334
Lecture – 17
Flow Boiling in Microchannels
Hello in this lecture we are going to discuss Flow Boiling in Microchannels. Flow boiling
in microchannels is probably one of the earliest applications of flow in microchannels
because it has applications in refrigeration and compact heat exchangers. So, in the earlier
lectures we have looked at the heat transfer without phase change, specifically in the
Taylor flow or slug flow regime.
Now, we will try to look at how the flow boiling in microchannels is different from that in
conventional channels and what are the different effects, what is the mechanism of heat
transfer. So, in the last 20-30 years there have been lot of research specifically using
experiments or experimental research, where people have looked at the flow boiling in
microchannels.
Now, there are two different viewpoints: one group of researchers say that nucleat boiling
is the dominant mechanism of heat transfer during flow boiling in microchannels. On the
other hand, another group of researchers they say that the evaporative heat transfer or
forced convective heat transfer is the dominant mechanism of flow boiling in
microchannels. So, in this and the subsequent lecture we will try to address some of those
questions and look at the state of the art of flow boiling in microchannels briefly. If you
Google the review articles on flow boiling in microchannels, you will find at least about
10 or more articles on review of flow boiling in microchannels they might differ in aspects.
So, this is a very active area of research recently.
Now before we start discussing a flow boiling in microchannels, it is a good idea to remind
ourselves or to get familiarized with different terminology or some basic fundamentals of
boiling in channels or boiling in general. So, boiling generally refers to the heat transfer
by evaporation over a heated surface. Now there can be the boiling is categorized in two
different categories: one is Pool Boiling and another is Forced Convective Boiling.
335
So, the pool boiling refers or this categorization is based on the if the liquid is static, if the
liquid is not flowing then it is pool boiling, a general example of boiling in the cattle or
boiling in the pan that we do every day in our kitchens. The other example or the other and
the other hand we have forced convective boiling; which is boiling of liquid flowing in a
heated channel. So, looking at the first at pool boiling what we have here is a boiling curve
for pool boiling and these curves have been obtained by Nukiyama and then researchers
later on and a the results have been presented in form say in the form of a boiling curve.
So, what the experiments generally involved that in a pool of liquid, a heated wire is
inserted and then either the heat flux is maintained constant and the wall temperature and
the or the superheated temperature is measured or the wall temperature is increase slowly
and the heat wall heat flux is measured, one of those can be done. So, this result they have
been presented in terms of wall heat flux versus superheat which is in terms of T wall
minus saturation temperature at a constant pressure.
So, as you can see that this graph is a log-log plot. So, it is between and the log of wall
heat flux and log of the superheat. Now initially when the surface is heated and the wall
heat flux is increased, there is no two phase flow, natural convection adds up in the liquid
and the hot liquid which is closer to the wall goes up and then it cools down and come
back again, so these natural convection is set up. And as the wall heat flux is increase, this
keep increasing; at its certain point which is called onset of nucleate boiling; let us call this
point as point A. There as you can see there is a small decrease in T wall minus T saturation
and then so in this case the boiling starts from here and you can see that this is called partial
336
nucleate boiling what happens that bubbles start forming. So, the phase change starts
happening in the cavities that might exist or that generally exist cavities or grooves or
crevices in the wall. So, there some air might be trapped and those cavity is the air trapped
in those cavities grow with the liquid phase changing to the vapor phase and as those
bubbles grow at a particular diameter the bubbles leave the wall and then again another set
of bubbles are released.
So, there are two mechanisms that play here; the natural convection which was occurring
and there was only liquid phase present plus the boiling caused by the nucleus of bubble.
So, that is why it is called partial nucleate boiling and you might see here that the slope of
this curve increases because the contribution of nucleus and boiling keep increasing with
the increase in wall heat flux. At point B onwards we see that the curve becomes straight
or a linear curve.
So, and this is called fully developed nucleate boiling. So, this region is natural convection
as you can see from here. The second region which has been marked here refers to partial
nucleate boiling, this region and the third region which is marked by fully developed
nucleate boiling. So, in this case the nucleate boiling is the dominant mechanism of heat
transfer and the natural convection is almost negligible and we compare it with the nucleate
boiling.
Close to the maxima we see that there is a decrease in the heat flux. So, the slope of this
line keeps decreasing and then we get a maxima at point C which is called critical heat
flux. So, in this region the nucleate boiling start decreasing because the number of bubbles
that are there they keep increasing and the continual supply of bubbles becomes difficult.
Now so, as we have seen that the pool boiling curve there are three different region. The
first region is referred to after the onset of boiling nucleate boiling and in the nucleate
boiling we have two regions; the partial nucleate boiling and fully developed nucleate
boiling.
337
Now, moving on after the nucleate boiling at this maxima point C is called where what we
have is critical heat flux. So, at this heat flux the surface is intermittently covered by vapor.
So, most of the surface is covered by the vapor phase and the conductivity of vapor is
smaller than that of liquid and then so we have very little heat transfer when you compare
with the nucleate boiling region.
So, that is why there is a certain decrease in the wall heat flux and the when this heat flux
decreases then this is disastrous for a number of applications because the temperature of
the surface might shoot up or it will shoot up. So, then there is a problem. Ah So, in this
case in the transition boiling the solid surface is partially covered by the vapor and the then
the vapor bubbles they keep going up and then some boiling happens, but this is decreased
significantly when they compare with the nucleate boiling.
338
Now as the heat flux is increased further the film boiling happens. So, in this case a
continuous film of vapor separates the solid wall. So, this is a vapor film; this separates
the solid wall from the liquid region. So, the heat transfer is decreased further and then
with the increase further increase in the wall heat flux this increases because there is always
a thin film of vapor here and then the heat transfer happens accordingly ok. So in the pool
boiling there are three different regions; nucleate boiling, transition boiling and film
boiling.
So, while the transition boiling is the transition from the nucleate boiling to film boiling.
In the nucleate boiling what we have? The bubbles nucleate or the bubbles form at the
solid surface, if it is heterogeneous nucleation and these bubbles will grow and release
from the surface. Whereas in the film boiling the entire solid surface is covered by the by
a layer of vapor and the heat transfer is significantly reduced.
We might see some examples of film boiling in our day today life. If you have a hot pan
if you and throw some droplets of water on it, then we see that these droplets start dancing
over it, with really almost no contact between the solid surface and the droplets because
there is a thin film of vapor over it and this effect is also known as Leiden frost effect.
339
So, nucleate boiling is the preferred mode of boiling heat transfer as large heat fluxes can
be sustained while maintaining low heated surface temperatures. So, we can see when we
go to transition boiling regime that the heat flux decreases, but delta T or T wall minus T
saturation keep increasing. So, the wall temperatures keep increasing in this case. Now the
bubble nucleation can be by two mechanisms; one is homogeneous bubble nucleation
which refers that the nucleation can happen anywhere in the liquid.
So, this is defined as the appearance of the bubble in the liquid pool and away from the
wall straight in the liquid. Now these bubbles are created when the liquid is superheated
and this is initiated by a cluster of molecules of liquid which have energy higher than the
average energy of the molecules in the pool of liquid. But in this case the super heat or
super heating the amount of superheating of liquid can be very high. On the other hand,
the heterogeneous bubble nucleation happens at the pits, scratches or grooves.
So, basically no surface is perfectly smooth. There are always micron size or smaller
cavities or scratches or grooves present on the surface and when the liquid is dropped on
it, there might be some air trapped on it and then when the nucleation happens or when the
heats heat is given to this these the air trap will grow with the liquid convert into vapor
and the bubble bubbles grow in it. We are not going to the details of this bubble, bubble
nucleation, but in any standard boiling book you can find the different relations for bubble
340
nucleation and further details of heterogeneous bubble nucleation as well as homogeneous
bubble nucleation.
So, when there is one isolated bubble what happens that in the, in during the heterogeneous
bubble nucleation after the boiling inception the bubble start growing and then it grows to
a critical size; which is generally called the bubble departure diameter. So, this critical size
is known as. Then when the bubble has grown to a sufficient size the bubble departs from
the surface. So, it leads the solid surface with so if we have a picture of a nucleation site.
So, this might be a cavity or groove on which the bubble which starts growing then this
might grow further and then it might lead at a particular size and then once it leaves, the
bubble leaves the surface, a small amount of liquid, a small amount of vapor it is still
present and when the bubble leaves the thermal boundary layer that was growing around
the bubble that is disrupted and the bubble leaves and then there is fresh liquid or the
cooled liquid that comes in and replenish the superheated boundary layer or again.
And then again a new thermal boundary layer grows, the bubble grows and then gas pocket
starts to grow and so on and further again the same process is repeated almost periodically
if you keep the heat flux constant. You can observe this if you keep the heat flux constant
in a pan or water boiling in a pan open to atmosphere. So, the diameter at which this bubble
starts leaving surface is called bubble departure diameter
341
So, that was about the pool boiling and in the pool boiling we have seen two mechanisms
of heat transfer at low heat fluxes what we have is nucleate boiling which is caused by the
nucleation of bubbles and at high heat fluxes the film boiling takes place. Now the other
form of boiling which is known as flow or convective boiling it refers to the boiling in a
channel. Now, in an environment when gravity is present, the orientation of the channel is
going to matter because in the presence of gravity buoyancy forced will be there as the
liquid and vapor phase has different have a very different densities.
So, the boiling in the vertically upward channel, vertically downward channel and
horizontal channel are going to be different. Generally, the orientation of the channel is
preferred to be vertically upward; nevertheless, in a number of applications the channel
orientation might be vertically downward or horizontal or where the boiling is not the
intended application there when the boiling occurs; it may occur in a vertically downward
or horizontal orientation.
So, this image shows a lot of information about flow boiling in a in a conventional channel,
in the vertically upward where channel orientation is vertically upward or the flow is
happening in the vertically upward direction. So, as you can see in this case up to point A,
this is only single phase liquid. So, the line on the left hand side represents the wall
temperature and line on the right hand side represents the temperature of the fluid.
342
So, for the single phase liquid as the wall heat flux is increased, the wall temperature keep
increasing, as well as the liquid phase temperature also keep increasing. Further on as near
the wall, the temperature reaches a particular point the wall temperature is above slightly
above a saturation slightly above the saturation temperature, then the wall temperature
becomes constant and bubbles start growing near the wall or bubbles start being generated
near the wall. However, the bulk of the liquid is still sub cooled. So, in this case what we
have the liquid is sub cooled, but the liquid near the wall may be superheated because of a
temperature profile in the channel.
So, when if you remember the single phase flow heat transfer in a channel the liquid that
is near the hot wall that will be at higher temperature than the liquid at the center. So, when
the liquid near the wall is at higher temperature it starts growing bubbles and then these
bubbles will leave the wall and move towards the center. They might condense depending
on the flow conditions and so, overall slightly above this one start looking at or the quality
becomes 0 or the quality starts growing more than 0, slightly after when this sub cool
boiling starts.
And once so, just at the sub cool boiling liquid quotient. So, at this place you can see that
there is the liquid, if you look at the liquid in an average sense then the temperature of the
liquid is sub cooled. So, that is why we have and the sub cool boiling the quality to be 0
here, but there are small amount of bubbles present plus there is some amount of liquid
which is sub cooled.
So, that is why the quality is shown to be 0 here. As we increase the heat transfer, now we
can see that the wall temperature and the fluid temperature both of them have become
constant and so, the difference between the wall temperature and the fluid temperature is
a constant and at a particular point until a certain length what we observe is what is called
saturated nucleate boiling.
So, in this case the mechanism of boiling is still nucleate boiling. So, bubble nucleation
happens and then the quality is smaller so that means, the vapor fraction it is smaller. Now
you can see that in the sub cooled boiling and saturated nucleate boiling the bubbles are
smaller. So, if you refer to the discussion of flow regimes that we had for adiabatic flow
where there was no heat transfer then at low gas flow rates the flow regime was bubbly
flow regime. So, here we see the appearance of the bubbly flow regime at low at the
343
entrance of the channel and because the quality is low here, as we move further down the
channel these bubbles coalesce and big grow and the occurrence of slug flow or the large
bubbles happen.
Further down we see the occurrence of annular flow. So, because the quality has increased
sufficiently and the as the quality increases you know the volume fraction of gas becomes
very high because the density of gas is very small when compared with that of the liquid.
So, at a certain distance annular flow starts setting up; so that means, there is a continuous
film of liquid and gas is there in the core. So, no more nucleate boiling is the dominant
mechanism that the bubbles are the heat that is being coming in it is being transferred from
the film and then the evaporation happens and the nucleate boiling does not happen in this
case. So, this is called forced convective heat transfer through the liquid film and in this
case the convective heat transfer through the liquid film is happening.
And after that you might see some of the, that the annular flow has some drops of liquid
dispersed in the gas core. Eventually as we move further down the film thickness keep
decreasing and at a certain point the dry out will happen. So that means there is no liquid
film in contact with the wall and vapor is directly in contact with the wall. So, as you can
see here the wall temperature suits up and then that as heat transfer is very inefficient in
the in this region and then finally, there is only vapor phase present and one will have only
the single phase vapor present in the channel.
So, we have different flow regimes in terms of heat transfer or boiling heat transfer the
first being the convective heat transfer for liquid just single phase. Then the bubbly flow
is start during the sub cool boiling and as the boiling increases further we have saturated
nucleate boiling and this happens of the nucleate boiling happens generally for slug flow
and bubbly flow regime. In the annular flow regime what is observed is force convective
heat transfer to the liquid film and then when the dispersed droplet flow or mist flow
happens and then and the finally, only vapor phase is present in the channel ok.
So, as you can see the heat transfer coefficient will be a constant in the single phase liquid
flow. Further up here we can see that the delta T is constant and so, the heat transfer will
be a constant in the or heat transfer will be dependent on Q wall over delta T and this delta
T is a constant in this case. Now this delta T even though the temperatures are increasing,
but delta T is constant so heat transfer is constant here. In this case delta T is a constant
344
and over temperature and this bulk flow temperature is also constant. Remember that the
heat transfer coefficient is defined as Q wall divided by T wall minus T bulk where the T
bulk its same at the saturation temperature in the saturated nucleate boiling region or in
the forced convective region, but in the single phase liquid region the liquid will be sub
cooled. So, the saturate saturation temperature and bulk flow temperature will be different.
So, in the flow or convective boiling there are two different mechanisms of heat transfer
that are in play; one is nucleate boiling heat transfer, another is convective boiling heat
transfer as we have just seen. So, the nucleate boiling heat transfer it is similar to the
mechanism is similar to nucleate pool boiling. The only difference is that the bubble
departure will be affected by the flow patterns or the departure of the bubble or the release
of the bubble is going to affect the flow behavior in the channel. So, the interaction
between the flow and the bubble departure is an additional factor when we consider
nucleate boiling in the flow or convective boiling as a mechanism.
So, in this case one should remember that the bulk flow might affect the growth of the
bubble and departure of the bubble and bubble induced convection process. So, bubble
might induce some convection which can add to the already existing the convection
process.
What is observed that in the nucleate boiling heat transfer, the heat transfer coefficient is
a function of strong function of heat flux. But it is independent of the mass flux or the
345
vapor quality. So, vapor quality you might remember is the mass fraction of the vapor
present in the channel. Now for the case of convective boiling heat transfer or the for
convective boiling heat transfer mechanism the dominant mechanism is the convection
process between a heated wall and the liquid phase and in this case the heat transfer
coefficient is almost independent of the wall heat flux and it is a strong function of mass
flux and the quality.
Now, what we have looked at is two different mechanisms of boiling in conventional

channels; pool boiling and two different modes of boiling, pool boiling and forced or flow
boiling. In the pool boiling where the heat transfer is happening in a stagnant liquid over
a surface two mechanisms can exist depending on the heat flux, at low heat flux, at low
wall heat flux and at low super heat what we have is nucleation boiling and at high heat
fluxes above a critical heat flux, transition boiling and then finally, film boiling happens.
Whereas, in the forced convective boiling, nucleate boiling happens predominantly in the
bubbly and slug flow regimes and the forced convective boiling happens in the annular
flow regime.
So, there is a difference when we talk about the flow regimes in during the phase change
process and flow regimes during the adiabatic flow, there is an important difference
between the two. When we talk about the adiabatic flow, the flow of gas and liquid the
346
flow rates of gas and liquid phases are constant. So that means, the flow after some distance
the flow becomes fully developed or a flow achieve a steady state.
Whereas, during the boiling as we move along the channel the liquid flow rate or the
fraction of liquid decreases whereas, the fraction of the vapor phase keep increasing. So,
the void fraction or fraction of the gas or volume fraction of the gas, keep increasing and
as a result we have different flow regime along the length of the channel. This will also
depend on the wall heat flux.
For example, the case that we have discussed is at low or moderate heat flux in a
sufficiently long channel. If we have a very high heat flux applied on the channel wall,
then one will have a very high degree of superheat near the wall. And the bubbles start,
bubbles will start growing and one can see that there might be film boiling, starting near
the wall which is detrimental for a number of applications because the heat transfer in this
case will be very poor and the wall temperature will shoot up to very high values.
Now, with this information or let us think about the boiling in microchannels. So, first
thing to think about is that what is so different that the boiling will be different in
microchannels? Then the second question or a related question comes. So, what is the
channel size at which these micro effects if any start to come into play? So, there are
different criteria a number of criteria have been suggested by researchers.
One of them by professor Kandlikar and he suggested that for channel size between 50 to
600 microns they can be termed as microchannels and between 600 microns to 3000
micron or 3mm, they can be termed as mini channels and channels above 3mm, they can
be termed as conventional channels. However, this classification seems to a bit arbitrary
because it does not take into account any physical mechanisms and the effect of the
properties of the fluid that are flowing in the channel.
347
So, another criteria or a number of criteria have been suggested based on the bubble
departure diameter. Because what can happen that when the pool boiling or the when the
nucleate boiling occurs in microchannels the bubble departure diameter probably one of
the criteria can be; when the bubble departure diameter is equal to the size of the channel
or is compeller comparable to the size of the channel, then the flow mechanism is not going
to remain same as one will see in as one will see in the conventional channels. Because in
the conventional channels there is sufficient space for the bubble to move towards the
center, whereas in a micro channel or in a mini channel or in a millimeter size channel the
bubble departure diameter may be of the same size as the size of the channel or even bigger
than the channel size.
So, in this case the bubble will not grow in the radial direction it will have to grow in the
axial direction when the bubble departure diameter is larger than the channel size and one
will have Taylor flow kind of flow regime in this case. Moreover, even if the size of the,
bubbles are smaller they can easily coalesce as and make a bigger bubble and then so some
of these micron size effects. So, bubble departure diameter seems to be an important
parameter or important criteria which can determine that when the channel size should be
considered to be micro or micron size channel or or smaller channel.
So, this correlation, given by Fritz for pool, nucleates boiling. During pool boiling he has
this, the bubble departure diameter as a function of wall contact angle. So, 𝛽 is wall contact
348
angle, σ is surface tension and this wall contact angle is in degrees. σ is surface tension g
is gravity and 𝜌𝐿 − 𝜌𝑉 . So, 𝜌𝐿 is the density of liquid phase and 𝜌𝑉 is the density of the
vapor phase. However, when it comes to the flow boiling as one can see this is only the
ratio of buoyancy forced and surface tension. So, it does not take into account the shear
force that may act or that acts on the bubble steering bubble departure. So, it remain on be
very suitable for flow boiling conditions.
So, using this bubble departure diameter a criteria. Kew and Cornwell they give a
correlation based on a non-dimensional number which they termed as confinement
number, which is a ratio of capillary length and hydraulic diameter and one can relate this
with say bond number. So, this is bond number one can see is
∆𝜌 g 𝑑2
.
𝜎
So, if one look at this then this becomes
1 𝜎 1 0.5 𝜎
= . So, Co =𝐵𝑜 = ∆𝜌 g 𝑑2
𝐵𝑜 ∆𝜌 g 𝑑2
So, this confinement number is nothing, but inverse of this square root of bond number.
So, Kew and Cornwell this suggested that when the confinement number is greater than
0.5, then the flow can be term to be Microscale. So, if you look this in bond terms of bond
number so 1 of bond number is greater than 0.5 or sorry 1 over bond number square root.
349
So that means, 1 over bond number is greater than 0.25 or when bond number is less than
4, it one over point two five is four. So, a bond number is less than four then the flow can
be termed as microscale. So, this means and the bond number is less than 4; that means,
the surface tension starts becoming important when compared with the buoyancy forced.
So, this criteria is based on if you look at the bubble departure diameter is comparison in
this case is a comparison of surface tension force and the buoyancy force. So, this is one
criteria that the suggested by Kew and Cornwell.
Then another criteria based on the bond number was suggested by Cheng and Wu. So, that
was based upon the bond number itself and the bond number if it is less than 0.05 then the
termed it has microchannel and between 0.05 to 3 it was termed as mesochannel and bond
number greater than 3, this was termed as microchannel. So if we look at the Kew and
Cornwell they had bond number less than 4 for the microscale effect to assume importance.
350
Now, another criteria that has been suggested by Harirchian and Garimella; They
suggested that not only the effect of bond number, but the effect of mass flux or the flow
rate of the liquid is also an important criteria to determine if the microchannel effects are
going to be important. And they suggested a that the
𝐵𝑜0.5 × 𝑅𝑒𝐿𝑂 < 160
root of bond number multiplied by the Reynolds number when it is equal to 60, below 60
if bond number raise to the half into liquid only Reynolds number if it is less than 160 then
they termed is as flow to be microchannel or when the bond number is less than 160 over
Reynolds number is square then, then the channel can be considered to be microchannel.
So, these are different criteria, that has been suggested. The latest one in which they have
also shown they have also considered the effect of not only the properties of the liquid, but
also the flow behavior, so which can take into account probably the shear forces.
351
So, now coming to the dominant mechanism flow boiling in microchannels, there are two
different school of thought and in the recent years there has been a lot of research to
identify that which of the mechanism of what is the dominant mechanism boiling. In one
school of thought it is suggested that nucleate boiling is the dominant heat transfer
mechanism. And this assumption or this comes from the observation that the heat transfer
coefficient is a function of heat flux, but not a function of mass flux and vapor quality.
And based on this it has been suggested that nucleate boiling is the dominant mechanism
of heat transfer.
On the other hand, researchers have suggested that it is not the nucleate boiling, but the
evaporation process or the convective flow boiling or convective evaporation is the
dominant mechanism of boiling and it has been shown by in developing some models,
some mechanistic models that the heat flux dependency can be explained by this thin film
evaporation process. So, we will further look at into this the models that has been
developed till date and a bit more into detail about the discussion on dominant mechanism
in flow boiling in microchannel in the next lecture.
So, to sum up in this lecture we have looked at some fundamentals of boiling in

microchannels in general; the pool boiling and forced convective boiling and in the pool
boiling there are two different mechanisms nucleate boiling and film boiling and the
transition between the two. So, there are three different regimes that are generally shown
352
on a pool boiling curve. Then for the forced convective boiling there are two different
regimes or two different mechanisms; a nucleate boiling which is dominant in the slug
flow and the bubbly flow regime and the annular flow boiling or the or the convective
forced boiling which is dominant in the annular flow regime. Then we have looked at some
of the criteria’s that have been suggested, some of the criteria in the literature that have
been suggested for from macro scale or from large channel to micro scale transition. And
it appears that most of these criteria are based on bond number which is responsible to
determine the bubble departure diameter. One of the recent criteria includes the bond
number as well as the mass flux in terms of the Reynolds number of the liquid.
So, the discussion on the mechanism of heat transfer in microchannels; there are two
different schools of thought one suggests that the nucleate boiling is the dominant
mechanism and the other suggests that the forced convective boiling is the dominant
mechanism. So, we will look at further on these both, the discussion or both the point in
the next lecture.
Thank you.
353
Lecture - 18
Flow Boiling in Microchannels Contd.
Hello. So, in this lecture we will continue our discussion on Flow Boiling in Micro Channels.
So, in the previous lecture we looked at the fundamentals of boiling nucleate boiling and
convective boiling and their contribution in general in conventional channels. Then we looked
at the different criteria that have been proposed especially with respect to the boiling in small
diameter channels where the physics of boiling is going to change because of the confinement
of the flow.
So, in this lecture we will look at some of the flow regimes or flow regime maps that are there
in the literature for boiling in micro channels and then we will discuss a bit about the
mechanisms of boiling in micro channel and some model. So, as we have been doing in the
entire course to get some prior information or took in order to start with we have looked at
what is happening in the conventional channels.
So, same goes for the flow regime maps in micro channels for boiling. So, people have tried to
compare the flow regime maps for micro channels with those in the boiling in conventional
channels and there does not seem to be the maps or the criteria proposed in the conventional
channel they does not fit well for the data obtained for micro channels.
Then we have also seen two phase air water flow or two phase gas liquid flow in micro channels
for adiabatic conditions where there is no heat transferred and in such cases we have obtained
different flow regimes predominantly slug bubbly churn annular and slug annular.
So, people have also try to compare the two things, but the difference in the two is that the flow
rates of the gas and liquid are constant in the adiabatic case whereas, in the case of boiling the
evaporation is happening. So, the gas flow rate or the vapour quality increase continuously and
the liquid flow rate keep decreasing.
So, along the channel length if the channel is sufficiently long one see different flow regimes
in the channel length itself, but one do not see the fully developed flow. So, for example, one
may not be able to see the regular periodic Taylor flow; that is observed in case of adiabatic
354
flow where as in a boiling one might say that these bubbles there are a number of long bubbles,
but they will have different sizes and they will grow in size and they move along in the channel
they will even coiners with each other.
So, the flow regimes will depend on the channel size on the gas and liquid superficial velocities.
So, here the gas and liquid superficial velocities we can have as say vapour quality with will
which can represent or which is a measure of gas flow rate and the total mass flux that can
represent the liquid or total mass flux multiplied by 1 minus x, where x is the quality can
represent the liquid flow rate.
Apart from this the wall heat flux can also play a major role in the occurrence of the flow
regimes.
What we have in this slide is; the experiments that have been performed in two different
channel sizes: one is 4.26 mm and the other one is 1.1 mm and this is for boiling of R 134 at
10 bar pressure and they observed that while the boiling regimes in 4.26 in mm channel they
looked like conventional channel whereas, there was significant difference in the flow regime
maps that were observed in the millimeter or 1, 1 millimeter size channel.
So, the first thing they suggested that they have not observed any mist flow. So, you can see
the mist flow is determined or is characterized as the droplets dispersed as mist in the almost
dry wall. So, where there is the when almost all the liquid has dried up vapour has become
355
saturated or maybe superheated in this vapour there are small liquid droplets there which are
dispersed.
So, they did not see any such thing and flow regime was annular and then probably dry out
regime whereas, the confined bubbles. So, here they have seen bubbles which are confined in
the channel and there it was a liquid slugs kind of thing and even the image in the bubbly flow
when there are bubbles which are confined. So, this bubbles, when after nucleus and they did
not departed; before departure the bubbles elongated.
And then they also suggested that the liquid film surrounding the slug or the liquid film thinning
liquid film. So, the liquid film surrounding the liquid slug that started thinning or liquid. So,
that is another characteristic and then they saw that the churn flow, whereas, in the large
channels it was very chaotic the chaotic nature of the churn flow decreased at the channel size
decreased and it has been observed in other experiments that in smaller channels the churn flow
is not even observed ok.
So, these are some of the differences that has been noticed we had presence of similar flow
regimes as we discussed in gas liquid adiabatic flow in micro channels we have bubbly flow
confined bubbles slug annular flow and then this is churn or depending on the condition slug
annular condition.
356
So, a flow map has been proposed by Thome’s group and they looked at the flow regime on
the coordinates where mass flux is on the y axis and the vapour quality is on the x axis and then
they identify it or found that, because the flow pattern it depends strongly on the coalescence
rate on the bubbles are generated they grow and then they leave the wall and these bubbles may
coalesce. So, the bubble generation rate and coalescence rate that is a determining factor that
this flow regime may occur. And it is strongly dependent on the heat flux and this flow regime
was further map was further extended by further experiments by their group.
So looking at the 4 flow regimes that they identified IB which is isolated bubble regime, so this
is characterized by that the generation bubble generation rate is larger than the coalescence rate
of the bubbles and in this case they saw both bubbly and the slug flow.
The next flow regime is coalescence coalescing bubble regime. So, in this case in this regime
the bubble coalescence rate is significantly larger than the bubble generation rate. So, isolated
bubble regime and then coalescing bubble regime when the bubbles have coalesced of course,
we will get an annular flow regime. So, the next provision was the annular flow regime and
then it was observed until a dry out occurred on the channel wall. So, the next regime was post
dry out regime. they have also given expressions for the different transitions.
357
So, the transition from bubble to slug regime it depends on liquid only Reynolds number and
this is gas only weber number boiling number. So, boiling number which we have not seen
𝑄𝑤𝑎𝑙𝑙
until now is defined as Q . So, Q wall is wall heat flux G is mass flux and hLV is latent heat
𝐺ℎ𝐿𝑉
of vaporization.
So, from this one can observe that x from boiling to slug regime it depends on q0.41. So, it is
dependent on q and then slug to annular transition it depends on liquid only Reynolds number
and liquid only weber number and then annular to post dry out transition is dependent on the
ρLσ 𝐿ℎ
ratio of the density of the two phases and and
𝑔 𝑑
358
So, another flow regime that has been given by Harirchian and Garimella they showed it on
two different non dimensional numbers as coordinates on the x axis is bond number raised to
the power 0.6 × Re on and the y axis boiling number × Re. So, boiling number is same that we
have defined in the previous slide and you might remember from the previous class that this is
called confinement number convective confinement number here Bo is the bond number and
Re is the liquid only Reynolds number.
So, they identified a two different transitions here one is the vertical line as we see here this
vertical line corresponds to Bo 0.5 × Re. So, the x axis is equal to 160 this point is 160 and that
suggests that bellow this one get the two flow regimes that we get here is this is confined
annular and the other one here is confined slug. So, the flow regimes below this are confined
above this one get either bubbly flow regime or churn wispy annular or churn annular flow
regime.
So, the vertical line separates the confined annular or slug flow regime with the bubbly or churn
annular flow regimes. Now the other line that separates is the slanted line which is given by
this expression and that the boiling number is 0.017 bond number raised to the power 0.4 and
Reynolds number raised to the power minus 0.3.
So, in this case we see that below the line the flow regime is either bubbly or slug and above it
the confined annular. So, this represents the that the bubbles have coalesced and they have
made a continuous vapour stream or they have depending on the surface tension and they have
359
the circular regimes or bubbly regime in the bubbly regime there will be bubbles will be circular
or will have a curvature where as in the annular flow the bubbles have coalesced on may and
have made a continuous gas stream ok.
So, these are the two flow regimes that has been suggested and then Harirchian Garimella they
improved upon their flow regime maps and represented it based on Npch or phase change
number. So, this is called phase change number while the x axis is remain same and this phase
change number is equal to boiling number × Lheated divided by dh × (ρL- ρV) divided by ρV. So,
this takes into account the effect of the heated length of the micro channel. So, such a flow
regime this also enables to predict or determine the distance from the inlet where different flow
transition can occur.
So, in this case also the same data has been presented, but it has a different ordinate ok. So, as
we discussed.
360
In the last class that there are two different views on the dominant mechanism of flow boiling
in micro channels on the one hand people suggest that nucleate boiling is the dominant
mechanism of flow boiling in micro channel others suggests that it is not the nucleate boiling
a, but the evaporation or the convective boiling or the thin film evaporation that occurred in
the slug flow that is the dominant mechanism of boiling.
So, let us look at where does this these belief comes from.
361
Now, in conventional channels the nucleate boiling it dominates in bubbly and slug flow
patterns and the enhancement in heat transfer occur from different physical phenomena such
as evaporation of the liquid film that is near the bubble and the bubble growth and then this
causes strong mixing in the bulk flow because of the bubble agitation when it leaves the
nucleation site. So, that is the enhancements or those are the factors which causes the
enhancement of heat flux during nucleate boiling.
So, the nucleate boiling it depends on the bubble generation frequency the bubble departure
diameter and diameter range of active nucleation site and all this increase with heat flux and
pressure. So, as a result the heat transfer in the nucleate boiling in conventional channels it is
a function of heat flux and system pressure, but it is independent of mass flux and vapour
quality.
So, this particular observation that if the heat transfer coefficient is proportional to or is a
function of wall heat flux and system pressure and it is independent of mass flux and vapour
quality then it has been concluded for the experiments in flow boiling in micro channel that the
nucleate boiling is the dominant mechanism. So, that is the reason so; that means, the number
of experiments suggest that the heat transfer coefficient in flow boiling in micro channel is
dependent on wall heat flux and that is the reason that people have concluded that nucleate
boiling is the dominant flow mechanism
362
On the other hand, in the convection boiling of which occur at high vapour quality and the flow
regime is annular flow regime nucleate boiling is almost suppressed there is less number of
bubbles and the mechanism of heat transfer in this case is similar to single phase forced
convection and the enhancement occurs, because of evaporation of liquid film and the
increased gas velocity.
So, in this case; when the heat transfer coefficient increases with the mass flux; so when the
convective boiling is dominant then the mass flux the heat transfer coefficient is dependent on
mass flux and vapour quality, but it is independent of heat flux. So, if this is, so then one infers
that the flow boiling in micro channel is dependent on the convective boiling.
So, now this was the principal reason for belief that flow boiling in micro channel is nucleate
boiling as most of the experiments showed a strong dependency of the wall heat flux or the
strong dependency of the heat transfer coefficient on the wall heat flux and it can be a for a
number of reasons; because depending on the flow conditions depending on the heat flux
conditions one can have, one might be operating in the regime where nucleate boiling is
dominated or when might be operating where other mechanisms such as convective boiling or
thin film evaporations are dominating.
So, based on their experiments and analysis Thome’s group suggested that at low vapour
quality the bubbly flow regime occurs and in this case the thermal transport is dominated by
nucleate boiling; however, at the intermediate vapour qualities the bubbles elongate and
363
elongated bubbles and slugs of liquid. So, Taylor flow kind of flow regime is observed and
then in such case the heat is transferred by the single phase convection in the all liquid and
vapour zones and in the film a conduction or convection occurs through the thin films.
So, in this case the heat transfer coefficient is highly dependent on the bubble frequency.
Now, when the vapour quality is high then the flow regime is annular flow regime and in such
case the heat transfer is expected to govern by the convection and conduction in the evaporation
of thin film. So, in the annular flow there is a thin film that will reduce in the thickness because
of the evaporation of the thin film. So, the conduction through the film and the evaporation will
be the; dominant mechanism.
Now, they suggested that the dependence of boiling heat transfer coefficient on the heat flux it
can be explained by thin film evaporation process in the elongated bubbly flow and elongated
bubble frequency and transient heat conduction processes both are strong functions of heat
flux. So, they developed a model for this.
364
This is a visual picture of such that at low when the vapour quality is low, then bubbly flow
occurs and the nucleate boiling is observed at higher vapour quality nucleate boiling as well as
liquid film evaporation in slug flow both occurs even bubbles also nucleate in the liquid film.
Further, on the evaporation in the thin film occurs and most of this contribution comes at the
tip of the bubble and as the quality is increase the annular flow is observed.
So, now based on his under physical understanding develop from the experiments and CFD
simulations Thome he proposed a three zone model to explain the boiling in the slug flow
365
regime and he had the following assumptions that the wall the boundary condition on the wall
was a constant wall heat flux boundary condition and slug was considered as a cyclic path
passage.
So, slug flow was modelled as a cyclic passage with the constant frequency independent of the
axial position and the flow was considered as a homogeneous flow; that means, the gas and
liquid phases had the same velocity there was no slip between the gas and liquid phases.
So, it was homogeneous flow both the phases were considered to be liquid and vapour they
were considered to be at the saturation temperature the film that surround the slug it is
considered to be stagnant and they also assumed that the bubbles which are generated they
continue to grow downstream and do not coalescence do not coalesce. So, the coalescence has
not been taken into account in their model.
So, this is a visual picture of their model they considered three zone the liquid slug and the
bubble which evaporates. Now depending on the length of the bubble and the velocity and wall
heat flux there may be 2 of 1 region. So, as we see in the upper image there are two regions
that the bubble the film surrounding the bubble has dried out; where as in the image below the
film remains always surrounding the bubble. So, one can have two zone or three zones
depending on if they dry out of the film occurs in the bubble or does not it does not occur.
366
So, they considered the heat transfer for such a case and calculated that the heat transfer or the
local heat transfer at is equal to sum on the heat transfer coefficients in each of these three
1
zones. So, 𝜏 of liquid slug h liquid slug + t of film h of film z + t of dry zone × h of dry
𝑡
So, basically one can also write this in terms of the fractions of time; so τ . So, they consider
τ as the period this period is τ ; so when one bubble is generated and after that another bubble
is generated. So, this bubble generation process has been assumed to be periodic and τ is the
time period or between that two bubble generation. So, that will be equal to one over f or bubble
generation frequency.
Now, as we can see that this model all of these are function of z and this is time averaged heat
transfer coefficient. So, now, we need to get the heat transfer coefficient in the liquid slug we
need 𝜏 and we need three times t L s t film t dry zone if it occurs and the expressions for heat
transfer coefficient in the liquid slug heat transfer coefficient in the film and heat transfer
coefficient in the dry zone
So, the flow velocity we can calculate the times based on the length divided by the velocity.
367
So, we need the flow velocity as we discussed earlier that the flow velocity is obtained by
assuming flow to be homogeneous. So, the flow velocity is assume that
𝐺 𝐺𝑥 (1−𝑥) 𝑥 1−𝑥
Ugas = Uliquid = 𝜌𝑥 + = (𝜌 + )
𝑣 𝜌𝐿 𝑣 𝜌𝐿
So, gas flow rate divided by the or gas flux divided by the gas density plus G 1 minus x divided
by ρ L or we can write this as G x over ρ v plus 1 minus x over ρ l.
𝐺(1−𝑥)
Ls = ×𝜏
𝜌𝐿
𝐺
Lv = 𝜌 × 𝜏
𝑣
Now, the liquid slug length or l slug can be obtained as g × 1 minus x divided by ρ L × τ . So,
that is the liquid slug length and similarly the liquid vapour length is g × x τ by ρ v
𝐿𝑠 𝐺(1−𝑥)𝜏 𝜏
𝑡𝐿 = = 𝑥 1−𝑥
= 𝑥 𝜌𝐿
𝑈 𝜌𝐿 ( + )𝐺 +1
𝜌𝑣 𝜌𝐿 1−𝑥 𝜌𝑉
368
Now, from this we can find out note that this includes film and dry regions. So, this is l film
plus l dry. So, from this we can find out t in the liquid that is equal to L s over u let us say this
is U. So, this will be equal to G 1 minus x τ over ρ L × x over ρ v plus 1 minus x ρ L × G. So,
G and G cancel out now what we have is or we can just write G here and now we can divide
by x. So, we will have τ divided by x over 1 minus x ρ L over ρ v plus one
𝐺𝜏 𝜏
𝑡𝑉 = 𝑥 1−𝑥
= 1−𝑥 𝜌𝑉
𝜌𝑉 ( + )𝐺 +1
𝜌𝑣 𝜌𝐿 𝑥 𝜌𝐿
Similarly, t vapour can be obtained as τ g x τ divided by ρ v × g x over ρ v plus one minus x

over ρ L G and G cancel out here and we have τ divided by x ρ v over x over 1 plus 1 minus
x over x × ρ v over ρ L. So, these are the expressions for t L and tv.
Now, we are not going to discuss the criteria for the film dry out here I mean we are not going
to develop the criteria for the film dry out here, but we will look at briefly that. So, what we
have here is we have been able to obtain t L and t v which is t film plus t dry out ok.
So, now the next question is that we need to find out the film thickness and from that we can
obtain a relationship for the rate of evaporation of the film or the decrease of the thickness of
the film with time using energy balance. So, first thing is that we need an expression for the
liquid film thickness at the no jump the bubble. So, if this is our slug zone then what is the
thickness of the liquid film at this point.
369
So, based on the experiments and the analytical model or the semi empirical model proposed
by Moriyama and Inoue for a bubble which is confined between two parallel heated plates and
it is growth they developed a relationship for the liquid film and the nose. So, let us call that
𝛿𝑓𝑁 , which is the thickness of the liquid film at start
Now, to utilize this they introduce a correction factor of which is called let us say 𝐶𝛿𝑁 . So, this
was a model parameter and this depends on the. So, they obtained a expression for the film
thickness at the nose which we are not going to discuss in detail.
Now, based on the energy balance and expression was obtained for the evaporation in the thin
film. So, for the bubble region we have the film thickness here as delta n and when it passes
through the thickness of the film keep decreasing. So, if we consider a delta Z length of the
film then we can apply a heat balance that q 1 pi r where r is the radius of the channel delta z.
𝑑𝛿
𝑞(2𝜋𝑅∆𝑧) = 𝜌𝐿 2𝜋(𝑅 − 𝛿) ∆𝑧∆ℎ𝐿𝑉
𝑑𝑡
So, this is the heat incoming from the wall in this delta Z thickness 2 pi R delta Z this is equal
to minus ρ L 2 pi R minus delta, where delta is the thickness of the film d delta over dt delta Z
× delta h of vaporisation.
So, this is the heat coming from the wall and this is the heat required for the evaporation of the
d delta thickness of the film. So, we can cancel out these two terms and delta Z can also be
370
cancelled out and we assume that R is equal to R minus delta as delta is small negligible with
respect to R.
𝑅 = 𝑅 − 𝛿 𝑎𝑠 𝛿 ≪ 𝑅
𝑡
𝑞𝑡
∫ 𝑞𝑑𝑡 = ∫ −𝜌𝐿 ∆ℎ𝐿𝑉 𝑑𝛿 = = 𝛿𝑓𝑁 − 𝛿(𝑧, 𝑡)
0 𝜌𝐿 ∆ℎ𝐿𝑉
So, we get from here q is equal to this R and R minus delta can be cancelled out. So, we get q
is equal to minus ρ L q dt is equal to minus ρ L delta h L v × d delta. So, if we integrate this
from 0 to t and at t is equal to 0. We will have delta n is the thickness of the film and delta is
the thickness of the film at an arbitrary time.
𝑞𝑡
𝛿(𝑧, 𝑡) = 𝛿𝑓𝑁 −
𝜌𝐿 ∆ℎ𝐿𝑉
So, we will have this will give us q t is equal to delta qt divided by ρ L delta h L v is equal to
delta n minus delta Z t or we can get the film thickness delta Z t is equal to delta n or delta f N
minus q over ρ L delta h L v × t. So, this is an expression for the thickness of the film.
Now when the film thickness becomes less than a delta minimum; then a dry out region occurs.
So, in the first model that term developed one of the parameters was or one of the fitting
parameters was this minimum film thickness. However, later on with the experiments and
comparison with the experiment others experimental data they suggested that the actual surface
roughness can be taken as the minimum film thickness that is where the dry out start occurring.
So, this is not a fitting parameter anymore and it can be taken as an input in the model
depending on the or based on the roughness of the surface
Now, heat transfer in the liquid film, because the film is considered to be stagnant. So, and for
a thin film one can consider as film is equal to conductivity of the liquid divided by the delta.
So, that will give the heat transfer in the liquid film.
371
Now if we go back to our original model, then what we required is the bubble frequency or τ
which is the period the time periods and heat transfer coefficients. So, we have already looked
at the time periods for liquid slug and vapour we have also looked at the expression which
gives us that when the dry out will occur. So, we can find out t film and t drive. So, we can get
the three times we have also discussed the heat transfer coefficient on the film.
Now, in the liquid slug; because, it is a single phase liquid; so one can use the expression for
single phase developing flow in the single phase developing flow in a channel depending on or
one can calculate the Reynolds number and find out if it is a laminar flow or turbulent flow and
calculate the expression accordingly.
Similarly, for the dry region one can calculate the heat transfer coefficient based on the gas
properties again for the single phase vapour flow. So, one can take into account the laminar
and turbulent regime and to find out if the flow is turbulent or laminar and get the expressions.
So, we have this model developed except that the frequency or the τ is not there.
So, what they have shown that the frequency is a function of
𝑞
(𝑞 )n
𝑟𝑒𝑓
q over q reference raised to the power n. Now this q reference is again a function of 𝛼 or a 𝛼𝑞
and reduced pressure. So, P over P critical and this is saturation pressure raised to the power n.
372
So, that suggests that f is proportional to q raised to the power a into this reduced pressure or P
saturation divided by P critical raised to the power say b. So, f is or the frequency of the bubble
generation is proportional to q raised to power a. So, from this one can see that the bubble
generation frequency and therefore, the heat transfer coefficient is overall heat transfer
coefficient is a function of the wall heat flux.
In the model they had three different adjustable parameters one is bubble frequency and we
have seen to give bubble frequency they need to provide one constant and two parameters here
and the nose film thickness correction factor 𝐶𝛿𝑁 .
And the minimum film thickness. So, they were able to eliminate one of the fitting parameters
channel roughness or based on the channel roughness. So, this is not arbitrary anymore.
However, the parameters for the bubble frequency and the film correction factor. So, there are
four parameters that needed to be determined from the experimental data.
So, based on the parameters that they have determined they have obtained the fitness of the
model. So, this model is based on the assumption that the evaporation in the liquid film is the
dominant mechanism of flow boiling in micro channels; however, several experiments have
shown or in recent times that there might be the presence of small bubbles in the liquid film
region in the liquid slug or in the liquid film. So, if this is the case the nucleate boiling can also
be important.
373
So, in my opinion it has not been yet the final verdict is not out yet what is the dominant
mechanism of flow boiling in micro channels there is still needs to be a lot of experiment to be
done specially look at the flow pattern for different channel roughness for different liquids for
different channel cross sections at different mass flux and identify the flow patterns and
corresponding a flow regime flow regimes and related to the related to at the mechanism of
flow boiling in micro channels ok.
So, in summary what we have discussed today is that flow regime maps the two flow regime
maps one is by professor Thome’s group and another one is professor Garimellas group
professor Thome’s groups model was based on the mass flux and vapour quality as the
coordinates whereas, professor Garimella he had the two non dimensional numbers as the
coordinates on the y axis he had a bl × re boiling number × liquid only Reynolds number and
on the x axis the convective confinement number which is bond number raised to be power 0.5
× Reynolds number and based on that they suggested the occurrence of different flow regimes.
Then, we discussed the possible regions why people think that nucleate boiling is dominant
and why people think that convective boiling may be dominant and then we have discussed
briefly the model by Professor Thome’s groups where he has considered the slug boiling in
slug flow regime in micro channels. So, he basically in the model identified the three zones the
liquid slug zone the film zone and the drive vapour zone in the liquid slug zone and in this
model the heat transfer coefficient is the time averaged heat transfer coefficient. So, one look
at the heat transfer coefficient at one particular location and some it for the entire time period.
Now, in the liquid slug zone the heat transfer coefficient can be dif determined from the single
phase correlations for developing or fully developed regions depending on the length and the
developing length etcetera same is true for dry vapour zones for the dry regions, because there
is only vapour present in this region for the film region the because the film is stagnant. So, the
heat transfer coefficient is k over delta and because the now we had to obtain three times t film
t dry zone and t slug zone. So, for the t slug we can calculate the length of the slug divided by
the two phase velocity or divided by the homogeneous velocity.
Then similarly we could calculate the t film plus t dry the length of the vapour zone divided by
the velocity and then we calculated an expression for the evaporation in the film based on the
energy balance in the film the heat coming in from the wall and heat being utilized for the
374
evaporation. So, from that we could calculate the evaporation rate in the film and so, the
thickness of the film or rate of decrease of the thickness of the film.
So, from that we can also calculate the time at which the dry out will occur the initial thickness
of the film can be obtained they have developed a correlation based on the literature data
available in the literature for a bubble growing between two parallel plates which are being
heated. So, then the last parameter was the bubble frequency or the bubble period which is a
function of q over q reference and it is there is a power law.
So, the bubble frequency is a function of q. So, that is where the dependence of the heat transfer
coefficient or the overall heat transfer coefficient in the slug flow regime is observed which
explains the dependency of the heat transfer coefficient on wall heat flux even in the thin film
evaporation models ok. So, that is all for this lecture.
Thank you.
375
Lecture - 19
Flow Measurement Techniques
Hello, in this lecture we are going to discuss about the Measurement Techniques; in
particular, which are useful for micro channels and specifically which are relevant for
multiphase flow in micro channels.
So, because for the flow measurement first we need to think about what are the parameters
that we are going to measure. In micro channels it is often difficult to do point based
measurement. For example, what I mean by point based measurement say in a flow, one
wants to measure the velocity; then the velocity measurement for say for to measure the
speed of aircraft is done by say pitot tube and pitot tube measures the velocity at one
particular point only.
Such measurement devices affect the local flow field and in micro channels it is almost
impossible to have such a sensor to measure the velocity at all. Even though these sensors
will affect the velocity field, but if we neglect the errors introduced in the velocity field,
because of such probes even then it that the size of the probe itself will be of the micron
size. So, it is not possible to do such velocity measurements in the micro channels.
376
So, thinking about what are the parameters that one would be interested in measuring in
micro channels. And most of these parameters except say a pressure measurement, one
would be interested to measure these parameters as a whole field measurement; that means,
one would like to have information of the parameters as detailed as possible.
So, for example, when we plot the information that we get from computational fluid
dynamics, if we are able to get all that information from the measurements, then we can
validate our CFD simulations as we as we can use this data to develop accurate models
and design efficient devices.
So, thinking about the equations or the Navier-Stokes equations we have 4 equations the
continuity equation as well as the conservation equations, momentum conservation
equations. And these 4 equations are solved to obtain the flow field in a isothermal flow
for incompressible fluids. So, the unknowns in this are velocity field and pressure field.
So, those unknowns if we can measure while it is not possible to a measure or where, while
there are very few techniques available which can measure the pressure locally, but
recently there are techniques being developed which can measure the pressure locally, but
if you look at the literature there are very few articles available where the local field local
pressure field measurements have been performed.
So, what we are interested in to measuring the velocity field or the three components of
velocity and pressure field if possible, if not possible pressure field then the one would
like to definitely measure pressure draw or differential pressure between the two points.
So, at the entire channel cross section, because the flow that we are focused about is a
multiphase flow and in a multiphase flow as we have seen throughout the course the heat
transfer the pressure distribution, the velocity distribution is strongerly coupled with the
phase distribution, where the gas phase is present and where the liquid phase is present.
So, based on this we need an accurate prediction of the phase distribution. So, this phase
distribution can be in form of flow regimes, which is the first cut information that we
would like to have and then we would also like to know the void fraction or the volume
fraction of the phases, then the temperature field.
So, if you look at the now we discussed about the velocity and pressure fields and then for
multiphase flow we also need to know the volume fraction field now if that temperature is
377
an important parameter; that means, if the flow is non-isothermal there is heat transfer
involved and we also need to know the temperature field.
But as you have seen and as we have also discussed during heat transfer in the Taylor flow
regime and during boiling that measurement of temperature in the liquid is rarely done.
Generally, it is the measurement of the temperature of the wall is performed it may be by
using thermocouples or more recently by using IR cameras; the measurement of the wall
temperature is performed.
So, the wall temperature is obtained easily, but obtaining temperature of the fluid is
challenging there are some experiments have been performed and they are increasingly
being performed using LIF laser induced fluorescence. So, where a dye is introduced in
the liquid, which emits the fluorescence light or which emits the light whose intensity or
the intensity of the light is a function of temperature. So, one can measure the temperature
of the liquid in such a case.
So, one also need to know the temperature field and for mass transfer applications one
need to know the concentration field. So, again most of the measurements of concentration
are based on the point measurement that one measure change in concentration, that is
concentration at the inlet and outlet the distribution of the solute in the channel or local
distribution of the species is not known really in such a case.
So, again increasingly the use of laser induced fluorescence with suitable dyes is being
used or it is to measure the local concentration field as well as.
So, the temperature field and concentration field they can also be used by using they can
also be measured using laser induced fluorescence. So, as most of this discussion or the
velocity field pressure field they are coupled with phase distribution in multiphase flows.
So, let us discuss about phase distribution first before we do that.
378
Here is a typical experimental setup of which is used in most of the experiments where
there is only adiabatic measurement of the flow. So, this is the test section and in this test
section the air if it is gas liquid flow the air is supplied from a source or it might be either
from a compressor or from a cylinder and then the pressure is measured at this point the
flow is measured and this comes into a mixing device, the water comes through a mass
flow meter the temperature of water is measured and the two phases is mix in this mixer.
Now, in some cases the mixer can be in the test section itself which is good because then
the channel dimensions are not going to change. So, then these mixers they the flow from
here comes in and it is passed through the channel cross section. And then the pressure
differential using the two pressure terms as a measure the temperature at the inlet and outlet
in measure and the images of the pick test section or the flow in to test section is taken
using a CCD camera with either a zoom lens or a microscope, because it is not possible to
get good enough resolution using just using the camera. So, one either need a zoom lens
or a microscope to do the task
379
So, for the phase distribution of course if one can image the system then one can get the
phase distribution. So, for imaging one needs optical access to the test section, by optical
access I mean the test section should be transparent.
So, the transparency is required that the channel should be of a transparent material, as
well as the flow should also be or the fluids they should also be transparent. For example,
if there is there are a large number of bubbles present in the channel, then the visibility
inside the channel will be affected and it will not be possible to obtain the images at of
what we will get is the images at the wall of the channel and not at a particular cross section
in the channel.
So, of course, this task in micro channels we require zoom lenses or microscope, then we
perform heat transfer experiments often it is required to supply the heat through the
channel wall. So, when we supply the heat through the channel wall it has to be covered
by a metallic or electrical heater. So, the optical access to channel in such cases is restricted
and often the measurements or the optical access is obtained either making a small section
visible, in such test section or looking at the flow field looking at the imaging the system
just downstream the heated section.
So, these systems are based on the imaging or optical imaging then one where optical
imaging is not possible one can also measure the phase distribution, based on the properties
of the two fluids which have good enough contrast the properties, which are significantly
different in the two phases.
380
For example, one can measure the electric impedance of the 2 phases, that can be the
measurement of resistance, the measurement of capacitance, or the measurement of
inductance the most common ones are the measurement of electrical resistance and it is
easier to cheap and build one just need to put in the wires and then one can get the cross
sectional averaged resistance of the phases it is or the other technique is based on the
refractive index based measurement.
So, the refractive index of the phases if they are different then with the development in the
techniques, which can capture very sensitively the variation in the refractive index then
those techniques can be used again to get an estimate of the phase distribution.
So, when we have optical access to capture the image we generally need to use cameras
and these camera need to be very high resolution and high speed camera also. So, there are
in presence of specifically two types of cameras, which are used one is called CCD and
another is CMOS and these names of the cameras are based on the sensors that we use.
So, CCD camera is based on a charged coupled device which is in short is called CCD.
So, the CCD camera converts the light into electric charge, based on the photoelectric
effect and the CMOS sensor is also based on the same principle.
381
So, in the CCD camera an array of many individual CCDS or many individual the sensor
elements they are arranged either in a line or in a rectangular array and then and these each
element is called a pixel these pixels are charged by the incoming photons.
So, the charge this charge is transferred through a limited number of output nodes which
is converted to the voltage and this voltage is sent off as an analog signal to the computer.
So, this the speed of the CCD camera is limited by the number of outputs; and through
which the signal can be sent. So, the frame rates that can be achieved in a CCD camera is
limited. However, they have because all the pixels capture the light and the uniformity of
the output is high. So, they have low signal to noise ratio and the picture quality that one
get using CCD camera is better, especially in the low light CCD cameras are known to
perform better than coms also or CMOS counterparts.
But for their manufacturing a special technique is need to be used so, they are costly they
are costlier, than the CMOS counterparts CMOS cameras which is complementary metal
Oxide Semiconductors.
So, they have the semiconductor sensors and again these semiconductor sensors they has
they have it is own charge to voltage conversion devices. And so, each pixel or each sensor
has charge to voltage conversion device and then this sensor itself has a digitization circuit.
So, there itself the voltage is converted to digital bits. So, it is it the rate of transfer can be
382
high. However, the sensor size in CMOS cameras is larger than the CCD ones. So, a typical
CMOS camera will have 20 micron whereas, a CCD camera sensor will be 9 micron.
So, the sensitivity of to low illumination levels is high. So, they might perform poorly or
they perform poorly when you compare with CCD cameras in lower light. However, they
can have very high speed thousands of frames per second or at and at low pixel or at low
resolution they can even go to up to half a million frame rates per second.
So, most of the high speed applications are based on CMOS cameras and with the new
development in the camera technologies these days, the sensors are being developed which
combined the good of both of words, which can use CCD camera technology for the better
resolution and the CMOS technology for the better or the high speeds.
So, these cameras they need to be used along with zoom lenses or microscopes. So, the
simplest microscope that is used is based on bright field microscopy.
So, the bright field microscopy is that a light source emits the light from the bottom. So,
the test section is here and the light is emitted from the bottom and it is visualized either
through eyes or a camera. So, of course, when we want to do the measurements and
recorded we need a camera.
So, the recordings were made will be done from the other side and the illumination will be
done from the below and observed from above. So, this results in the contrast by the
383
attenuation of the transmitted light that is transmitting being transmitted through the test
section in the dense areas of the sample. So, you see a good and contrast image of the
sample.
Now, this will be dependent on because these are not static images we have flow happening
in this. So, we need to look at that the camera should have a sufficiently short camera’s at
a time.
So, for those who do not know what the camera shutter time means, it is the time for which
a camera sensor is exposed to the light when the picture is being taken. And this determines
the exposure time it will determine the light that will reach the camera.
So, at high flow velocities and this exposure time so, when we look at the exposure time
that will be determined by the resolution that we want and the flow velocities that we want
to capture. So, say for a flow velocity of 0.1 centimetre per second and the spatial
resolution of one micron a shutter time of around 0.1 microsecond is needed.
Now, the illumination that happens is generally using a continuous light sources. However,
at low shutter times or smaller shutter times less than 10 microseconds it is preferred if
one have a stroboscopic light source so, that the illumination can be matter and enough
light is provided to the camera.
384
So, the simplest method for the illumination or and the visualization is bright field
microscopy. However, that met that is not always good enough to take the pictures or to
get the better resolution. So, the fluorescence microscopy is often used and it has a number
of applications specifically with respect to diagnostics in micro channels one can measure
the void fraction distribution, one can measure the flow velocity, one can measure the
temperature field, one can measure the concentration field, using fluorescence microscopy.
So, the fluorescence microscopy in this the sample is illuminated with a light, which is
often of a specific wavelength. So, this light is monochromatic light this can be done using
a sub say xenon arc lamp or mercury vapour lamp LEDs or lasers and this light, which is
incident on the sample and this sample seeded sample means the flow. One of the phases
of the flow often the liquid phase or the continuous phase, its seeded with fluorescent
particles or which are called Fluorophores. So, these fluorophores they absorb the incident
light and then emit the light.
Now, when they emit the light the emit the wavelength of the emitted light and the energy
of the emitted light is different from the energy the light of the light, that is incident of
course, the energy of the light that is emitted will be lower than the light that is incident.
As a result the wavelength of the emitted light will be higher than the light that was
incident.
385
So, the fluorophores they emit light of a longer wavelength, when the light is incident on
it. And this light is collected and goes to the detector, which may be a camera using a
dichroic mirror and emission filters the illumination light the light which was being used
for illumination is separated from the fluorescent light.
And in epifluorescence microscope this is both the lights which is used for the excitation
of fluorophores, and detection of fluorescence both are done in the same microscope
epifluorescence microscope is where the excitation of fluorophores and the detection of
fluorescence both are done in the same instrument, it is called Epifluorescence microscope.
So, it is basically that the images are taken of the fluorescent particles. So, based on the
intensity of that fluorescent particle one can capture a field. And if this intensity of the
emitted wavelength or the emitted light is temperature dependent then one can measure
the temperature field. If it is concentration dependent then one can measure the
concentration field and if it is not dependent on such parameters then one can measure the
distribution of one phase.
So, it is often used and this technique is also used or epifluorescence microscopes are also
used for micro PIV applications, because the illumination in micro PIV is the volumetric
illu, illumination, unlike a general PIV, or 2D PIV, where the illumination is through a
light sheet. So, we have discussed the we will be discussing the details in a separate lecture
about the PIV.
Now, this liquid phase can be labelled with either a organic dye or semiconductor nano
crystals for example, CdSe dots and then one can obtain phase distribution shape of the
interface and concentration fields etc.
386
For temperature measurement a dye named as Rhodamine B is used which can be used to
measure the temperature field and this technique is known as laser induced fluorescence
or LIF to measure the temperature for concentration measurement in the liquids the dye
that is used is called Rhodamine 6G, for now in micro channels the intensity of the light is
very small. So, one often does not use a continuous light source, but rather a pulsed light
source is required. For example, a bubble velocity of about 0.1 meter per second and
exposure time of less than 100 nanoseconds is required.
So, that the resolution is below 10 microns. So, when such high as such high temporal
resolution is required, then one need to have pulsed light source, which are often also used
in micro PIV and PIV application. So, the two common light sources are Nd: YAG and
Nd: YLF solid state lasers, which can have very short pulse duration up to 10 nanoseconds
and they are useful and the trigger of the pulse is synchronized with the camera.
So, we need a synchronizer to synchronize the pulses that when the camera shutter opens
and when the light is triggered from the laser and the same technique is used in micro PIV
as we have just discussed.
387
So, now other technique till now we have discussed about 2 microscopes and cameras. So,
these systems can use the microscope that can image the flow and from this one can obtain
the volume fraction field in the channel, using a temperature sensitive dye one can obtain
the temperature field, using a concentration sensitive dye one can obtain concentration
field.
Where the optical access is not there then there are different other techniques one can use
so, one is say based on refractive index based sensor. So, as we know that as the light
enters from one medium to another the part of the light is reflected and the part of it is
refracted. So, what is total internal reflection?
So, we have let us say 2 mediums medium 1, where the refractive index is n1 medium 2
where the refractive index is n2 and the light is incident at an angle 𝜃1. And it goes out at
angle θ2 and we know from Snell’s law that n1 sinθ1 where θ1 is the incident angle and 2
sin θ2 θ2 is the angle of refraction.
Now, if we keep increasing θ1 then at one particular angle let us say θc the reflected light
that light will not refract anymore say if we have this n1sinθc = n2.
So, of course, this is true when this is only pass possible when n1 is greater than n2. So,
this at this critical angle the light will not refract anymore and for a higher angle the
reflection will be there.
388
So, that is what is meant by total internal reflection that their light does not reflect any
more it reflects. So, this is a powerful tool based on which the light can be confined and
the optical fibber’s and optical fibre probe probes are also based on this technology or
based on this principle total internal reflection.
So, one of the techniques that have been developed is that the reflection of laser light, at
the fluid interface it can provide the information about the flow characteristics.
So, in this case a laser beam enters the channel normal to the flow direction, at such an
angle that the total ref reflection, total internal reflection takes place when a gas or liquid
which is dispersed in a continuous liquid. So, it might be either a gas phase or liquid phase
which is dispersed, that when a gas or liquid phase dispersed phase it passes through and
this is dispersed in the continuous liquid phase; which has a higher refractive index is
present at the illuminated positions of the.
It is done in such a manner that when the dispersed phase passes through and the dispersed
phase should have a lower refractive index than the continuous phase. And it should come
at such a angle light should come at such an angle that the total internal reflection can take
place.
389
So, by detecting the emergent beam on the other side of the channel; one can obtain the
phase distribution. So, when only the continuous phase is present, the beam is almost
completely absorbed there is no nothing happens.
So, that the light intensity of merging beam image very low in such a case where as it will
it may not be the case in case of when the liquid passes. So, based on ref changes in the
refractive index one can obtain the phase distribution.
390
Now, another technique, which is known as MIBD or Micro Interferometric Backscatter
Detection so, in this case a laser light is used to create high contrast interference fringes.
And the shift in fringes is based upon the change in refractive index of the fluids. So,
because the two fluids will have different refractive indices so, based on the fringe shift
one can measure the change in the refractive indices and change in the properties of the
fluid.
So, the technique it has been reported the sensitivity it can the technique can detect the
changes up to 10-7 of refractive index units. And this technique can be used to measure
temperature distribution concentration and pressure distribution as a function of refractive
index. So, in this work by Fultz and Allen recently, they have developed a technique to
measure the local pressure distribution in a channel by this technique by MIBD technique.
So, the other technique that can be used apart from imaging and refractive index change is
based on the change in the electrical properties of the two fluids that when one fluid say
gas and liquid they have a very different electrical conductivity or resistivity or electrical
capacitance.
So, if one can have a probe which or one can have wires two wires through, which the
which connect the 2 ends of the channel. And if this voltage or the resistance is measured
and from this measurement of resistance one can know the change in the phase distribution,
or one can know the phase change at one particular location with respect to time.
391
These probes are easy to construct, they are low cost and if a number of electrodes are
used then with reconstruction algorithms, one can use these techniques as tomography
techniques, where one can determine the entire 2D field at a cross section using
tomography. So, one can reconstruct the phase distribution at the entire section using the
using a number of probes which is called tomography technique.
So, based on this a number of velocity measurement techniques are also developed for
velocity measurements in micro channels. Now most of the techniques which are used in
micro channels for velocity measurement or otherwise are non-intrusive, because as we
have discussed and the start of this lecture it is not possible to do the point measurements
in micro channels.
So, some of the popular techniques are particle image velocimetry and the related
technique particle tracking velocimetry, then laser Doppler velocimetry, and MRI or
Magnetic Resonance Imaging. We will discuss micro PIV or Particle Image Velocimetry
in a separate lecture.
392
So, it has a brief discussion of laser Doppler velocimetry, in this technique like micro PIV;
it is also non-intrusive and it measures the instantaneous velocity of the flow fields. So,
one can measure the velocity in the space as well as in time and in one can measure all
three components. So, full 3D velocity with temporal resolution.
So, it is again a powerful technique it is based on the as one can guess from the name it is
based on the Doppler principle and the wave that is used for it is a monochromatic laser
beam, it is sent towards the target and the light that is reflected is collected and the change
in wavelength of this reflected light is of course, a function of the relative velocity between
the target object the relative target objects relative velocity with respect to the source.
So, based on the Doppler Effect this relative velocity can be measured and then velocity
is obtained by measuring. The change in wavelength of the reflected laser light this change
in wavelength is often measured by interference fringe patterns by superimposing the
original and reflected light. So, from that one obtain the ∆𝜆 and from ∆𝜆 one obtain the.
So, what is ∆v or the relative velocity? So, that is the principle may have laser Doppler
velocimetry.
393
Now, coming to the pressure measurements the pressure measurements is a difficult bit,
but it is very important bit because it is required to determine the pumping power to run a
system. So, it is always required to design a system efficiently, one need to have an idea
about the pressure drop that happens in channel. Often pressure transducers pressure
transducers, which can capture the low changes or small changes in pressure are used and
depending on a particular system one can estimate the range of pressure changes, that are
that should happen in the channel and accordingly one should select a pressure transducer
for a particular application.
It is particularly difficult in to measure pressure changes in micro channels, because one

is the channel geometry and wall roughness. So, specifically as we have discussed earlier
also that the roughness of the channels can be an important factor in micro channels,
∈
because the roughness which is let us say 𝑑 . The channel characteristic might be negligible
in a small in large diameter channels, but this is something which can be important in large
channels.
So, the pressure drop is of course, being will be a function of roughness so, that is one
factor. Then that experimental test tricks that are developed for micro channel applications
if not developed carefully they are often susceptible to flow oscillations and flow
instabilities and the pressures that are being measured they are small pressures.
394
So, a even a small change or small oscillations in the flow or small instabilities they can
cause errors large errors in the pre measurements and then specifically during the phase
changes. The accelerational pressure drop which is a large component of the pressure drop
when the phase change happened, because the because the areas occupied by different
phases keeps changing along the length the accelerational pressure drop is a important
component of the pressure drop.
So, that the different magnitudes of pressure drop component is difficult to obtain in
specifically in micro channels, because no particular correlations are available which can
relate the void fraction will the change in the pressure drops. Apart from that there are
entrance and exit pressure losses, which happen in micro channels also, in a large diameter
or micro channels also, they can be because of expansion or contraction or the flow
distribution where the two phases mix.
So, pressure drop measurements are generally done using pressure transducers, but one
need to take into account all these factors so, that the accurate pressure drop measurement
can be obtained ok. So, with that we can summarize that one of the or the techniques in
micro channels are based on where one wants to have the good view of the flow field and
the phase distribution and that is possible using either bright field microscopy or
fluorescence microscopy.
For void fraction measurements the techniques are being developed based on the electrical
impedance changes, total internal reflects will reflection or based on the changes refractive
indices and these techniques have potential to measure the local temperature and pressure
distribution also.
So, with that we have an overview of flow measurement techniques in micro channels, in
another lecture we will discuss briefly the micro particle image velocimetry technique and
it’s basic principle.
Thank you.
395
Lecture – 20
Particle Image Velocimetry
Hello. In this lecture we are going to discuss Particle Image Velocimetry. So, this is one
of the most powerful techniques and very frequently used in recent years to measure the
velocity in different configurations different systems including in microfluidics. For
microfluidic applications there are some issues that need to be resolved and some changes
need to be made in the experimental setup and the corresponding experimental setup is
known as Micron Particle image velocimetry or Micro Particle image velocimetry.
So, as the name suggests that in particle image velocimetry veloci metry means; the
velocity is measured and this velocity is measured by the images of the particle. So, where
do the particles come, because what we are talking about is the flow of fluid? So, this flow
or any flow that we are measuring the velocity in this flow is seeded with the particles.
Seeded means particles are introduced in the fluid the concentration of the particles is
small, the size of the particle is small, and one of the major requirement of a the technique
is that the particles should follow the flow faithfully.
So, to get that once we have the particles such that these particles follow the flow faithfully,
then the velocity by taking the images of these particles of the velocity of the particles is
measured by post processing techniques or analysis or image analysis. And then the
velocity, because the particles follow the flow faithfully the velocity of the flow is obtained
as the velocity of the particles. So, the technique is called non-intrusive flow measurement
techniques.
396
So, the non-intrusive means that in this technique no intrusion needs to be made in the
flow. For example, in the pitot tube we need to insert the pitot tube in the flow or so that
the measurement of the flow can be made at that particular point unlike that the particles
are introduced in the flow; however, we have the requirement that the particles should not
affect the flow. So, the particle does not affect the flow field and we do not introduce any
sensor or any probe as such in the flow. So, the flow field is not disturbed and the flow can
be measured accurately.
So, it is called non-intrusive flow measurement technique, then the velocity measurement
is indirect as we are not measuring in the technique the velocity of the fluid particles or
the fluid molecules directly rather, the velocity of the of the solid particles or the liquid
particles or the tracer that is introduced in the fluid is measured and from that the velocity
of the fluid is measured. So, that is why it is called an indirect velocity measurement
technique.
The whole field velocity measurement so, if you have looked at some of the images in
computational fluid dynamics; that means, you can get the velocity field at a particular
cross section or a in the in the computational fluid dynamics, what one does is solves the
navier stokes equations in the entire computational domain this computational domain
might be 2 dimensional or 3 dimensional.
397
So, in this computational domain one obtains the velocity field at different points or in the
entire computational domain for example, in a 2D plane.
Similarly, in this technique one obtains the velocity field at a 2D plane for the entire plane
in which the measurements has been made, unlike the point probe method where the
velocity at a particular point are measured apart from this there are very few other
techniques, which can do the whole field velocity measurements.
Say for example, other technique is laser Doppler velocimetry which is based on the
Doppler Effect and radioactive particle tracking, which can measure the velocity based on
the attenuation of the radioactive material or radioactive particle that is introduced in the
flow.
The special resolution of this technique is very powerful and is limited by the special
resolution of the camera that is used to take the images of the flow. So, the more powerful
camera one has the more number of pixels one have on the sensor and the more or the
better special resolution one can get for the flow. The optical resolution will again be
limited by the frame speed of the camera or the speed at which the camera can capture the
images as well as the speed of the laser.
One of the limitations of the technique is that it requires optical access to the test section.
So, the optical access means that the light should be able to enter the test section and it
should be a transparent so, that the images of the test section can be captured. And this is
not possible say for example, if one wants to take velocity measurements in an
experimental setup or in an industrial conditions, most of the pipes or the industrial
equipment will be opaque.
So, there is no optical access as such and one might either need to have an optical window
to measure the flow or need to mimic that is in the industrial conditions in a separate
experimental setup.
This can also be a problem the optical access can be a problem even in a transparent setup,
when one have a multiphase flow and the visibility or the or the transparency of the flow
is affected by the presence of the large number of dispersed, particles, or dispersed bubbles
or droplets. For example, in a bubble column where a number of bubbles are present in the
fluid it might be challenging to take the images of the fluid in a in a bubble column.
398
So, in this technique the steps are that the flow is first seeded with particles the particles
they should be following the flow faithfully they should be moving with the fluid. So, the
velocity of the particle should be the velocity of the fluid at that particular point, then this
flow is illuminated this is often done using a laser light, but sometimes for a cheaper
equipment one can also have white light or nowadays with the development of LED lights
LED lights are being increasingly used for PIV applications.
So, the flow is illuminated with a light generally a lasers are used because the light from
the laser is monochromatic light so, that is why the laser light is preferred and then the
images of the flow are captured. So, generally the lasers that are used for PIV applications
they are double pulse laser. So, the one light wave or light pulse is introduced or is sent to
the test section and image is captured and within a small time duration another pulse is
sent. So, the 2 pulses or 2 light waves are sent at a time instant or time difference of ∆T
and with this known time difference the images of the particles are taken.
Now, these images may be captured in a single frame. So, the images of the particles are
may be captured for the 2 pulses in a single frame or they may be captured on 2 different
frames. And then these images are sectioned into or divided into smaller interrogation
areas and then the corresponding interrogation areas from these 2 frames are analyzed, in
case of a single frame using autocorrelation and in case of 2 frames using cross correlation.
399
So, cross correlation where we have 2 images and when it is captured in a single frame
one uses autocorrelation technique.
So, this is a typical schematic of a PIV setup, you can see that this is the test section in
which the flow is happening from in this direction and this flow is seeded with the tracer
particles. Now the laser light is introduced on it and you can see this that the laser light in
a 2D PIV system for a conventional channels or large channels the laser light is converted
using light sheet optics it is converted into a light sheet.
So, it illuminates a 2 dimensional plane of a small thickness of a small order of small width
in the test section and the velocity in this test section is measured this will be measured by
imaging optics.
So, it is the velocity is measured at first light pulse is introduced at time T and another
light pulse is introduced at time T´ and so, 2 images at time T and at time T´ are captured
these are the particles which are illuminated. So, the light scattered from the particle is
taken into the camera.
400
Now, these 2 images which are taken at time T and time T´ which is equal to time T + ∆T
they are cross correlated and from that the average displacement of the particle is obtained
for each interrogation window. So, the local velocity field in the each interrogation
window is obtained by cross correlation technique.
So, by looking at this experimental setup we can identify the main components of the PIV
setup. It is important as the technique is dependent on the tracer particles following the
fluid. So, it is important to have the tracer particles on a case to case basis depending on
401
the fluid phase. For example, the fluid is gas phase or liquid phase the density of the fluid
or the size of the test section all those considerations needs to be taken into account when
selecting tracer particle for a particular PIV application, then the laser and optics.
So, the type of laser and the light sheet and other optical accessories they need to be taken
into account or they need to be they are the main components of the setup. And the camera
in which the images are to be captured the resolution of the camera as well as the frame
rate of the camera will determine the spatial and temporal resolution. Of course, for the
temporal resolution one also need to have a high speed lasers only then the time resolved
PIV or very high flows can be captured, because until unless one have a high speed lasers
high speed cameras will not be useful for capturing the flow at large velocities or high
velocities.
A synchronizer will be required, which can synchronize the camera and laser. So, the pulse
the synchronization between the light pulses, which are sent to the test section and the
images that are captured they can be synchronized. And then finally, once the images has
been captured lot of work needs to be done into analyzing the flow a number of a softwares,
open source softwares are available on the web as well as all the OEMS all the a
manufacturers of PIV systems. For example, Dantec Dynamics Lavision as well as TSI,
they have developed their own post processing tools to analyze a flow in these systems.
402
Now, we let us look at these components one by one. So, the tracer particles if they do not
have the same density as that of the fluid, then the because of the buoyancy the particle
will have a motion of their own in the fluid. So, it is often advisable that the particles are
non-buoyant or so, they have the same density of as that of the fluid fast response to change
in the fluid velocity.
So, the particles if they are in stokes regime then the particles can follow the flow
faithfully. Now for a stokes flow one can derive the characteristic relaxation timescale that
is the time, which is required for the particle to respond to the change in the velocity of the
fluid.
So, the TS is the characteristic relaxation time which is given as dp2 𝜌 p over 18 𝜇 μ is
fluid viscosity ρ p is particle density and dp is particle diameter. So, this relaxation time
is a measure of the tendency of the particle to attain velocity equilibrium within the fluid.
Now a non-dimension number is defined, which is called stokes number ratio of the
relaxation time to the timescale of the flow, which is the ratio of a flow length scale and
the flow velocity scale.
So, the characteristic flow time is tau and characteristic relaxation time we write as tau S
which is given by this. So, flow length scale for example, in a channel it will be the
hydraulic diameter of the channel V is the typical velocity or the average velocity in the
channel.
So, that gives us a measure of the tendency of the particles to attain the velocity of the
fluid, low stokes number means the particle can adjust to the flow fast, because the
characteristic that the relaxation time is smaller than the flow time where as high strokes
number suggests that the fluid inertia will be important and it will not be good for PIV.
So, this will not be preferred for and because it will take a particle will take some time or
the relaxation time will be comparable to the flow time.
So, it will affect the measurement and particle may not represent the fluid velocity.
403
Now, so the considerations we just now discussed based on the stokes number, that is for
from the fluid dynamics considerations which require that as we see from here that the
relaxation time or the stokes number is proportional to dp2. So, one will prefer from these
considerations that low particle size or small particles will be preferred so, that they can
follow the flow faithfully.
However, there is another requirement which comes from the optical requirement. Now it
is known that the particle image intensity the intensity of the image that is captured it is
proportional. So, the better the image intensity the better contrast and better image one will
get for post processing or for analysis post processing generally refers to process the image
∆x
to obtain, the velocity field or to obtain the displacement and from their one gets the
∆T
velocity field.
So, the particle image intensity is proportional to the scattered light that is of light power.
So, the light that is scattered from the particles, now the light is scattered by small particle
that depend on a number of factors which are the ratio of refractive index of the particle
and the surrounding medium, which is the fluid of which the velocity is being measured it
also depends on the size of the particle, the shape of the particle, and orientation of the
particle, while if the particle is spherical then shape and orientation will not matter. So,
much, but the size of particle will do and it also depends on the polarization and the
observation angles.
404
Now, there are 2 different criteria, which according with to which the light scattering by
the particle can be they can depend upon. So, one is that the if the size of the particle let
us say the size of the particle is dp is greater than 𝜆, which is wavelength of the incident
light remember. We talked about when we talked about lasers we said the lasers are
preferred, because in a PIV we would like to have monochromatic light. So, there is one
particular wavelength of the light that is incident on the test section or on the sample.
So, if the size of the particle is larger than the wave of the wavelength of the incident light,
then one uses Mie’s scattering theory or Mie’s scattering criteria or when the particle is
smaller than the wavelength of the incident light which will be the case in a micro PIV,
then one need to use a Rayleigh scattering criteria and this criteria have a different
dependence of the scattered light on the particle diameter.
So, looking at the Mie’s scattering criteria which is when dp is greater than lambda, when
the particle size is more than the wavelength of the incident light, it has been observed that
the scattered light intensity increases with increasing particle diameter it is probably
proportional to dp2.
So, one would prefer larger particle diameter would be better for, now one need to satisfy
2 opposing criteria. For small particles the condition of that the particles should follow the
fluid one should have small particles. And for the case when the particle size is larger than
the wavelength of the incident light so, this criteria is when dp is greater than 𝜆. Then from
405
the fluid dynamic considerations one would prefer a small dp and from the optics
considerations one would prefer a larger particle diameters the one would need to have a
balance between the 2 conditions and decide the size of the particles.
So, here are typical particles that are used for liquid flows in general or in conventional
channels. So, polystyrene particles, which are of the size between hundred or 10 to 100
microns, aluminum flakes, size between 2 to 7 microns or the hollow glass spheres, which
is where the size between 10 to 100 microns oil droplet us between 50 to 500 microns
oxygen bubbles and the size can vary between 50 microns to 1 mm. So, these are some of
the tracers that are used for liquid flow in conventional channels.
406
Now, when we come to the other criteria where the tracer particle size is less than the
wavelength of the channel, which will occur when the observation field size is decreased
and increasing optical resolution is required. So, in this case for example, the best example
for this is micro PIV applications or so, when one needs to measure the flow velocities
very near to the wall. So, in such cases the size of the particles will be required for example,
if we are looking at a flow in a 100 micron channel then of course, one will need to have
smaller particles one cannot afford to use the particles which are of the size of say 50
micron.
So, in such case the particle diameter is smaller than the wavelength of the light. And the
amount of light that is scattered the scattered light varies as dp-6. So, the scattered light is
1
. So, the Rayleigh scattering criteria requires the particle diameter to be as small as
𝑑𝑝6
possible, which is opposite to Mie’s scattering criteria, when we were considering the
particles sizes greater than the wavelength of the incident light.
So, for such cases the diameter of flow particles must be small enough. So, as not to affect
the flow of course, smaller the particle it will be good, but one also need to take into
account that it does not affect the Brownian motion.
So, we will briefly discuss that in a later slide. So, the typical particle diameter that are
used is between 50 to 100 nanometers and the typical generally Nd-YAG laser are used in
407
many PIV applications, which has and the visible light in which the wavelength of the
light is 532 nanometers. So, the 532 nanometer light or the other laser that is used is Nd-
YLF again the wavelength is 527 nanometer so, these are the 2 often used lasers.
Now, this the wavelength of light in this case is 5 to 10 times the particle diameter, if we
are using 50 to 100 nanometer sized particles. So, this places is significant constraint on
the image recording optics.
So, generally what is done in such cases or specially for micro PIV that a in place of the
direct imaging one uses epi-fluorescence imaging to record light from the fluorescently
labeled particle. So, the particles they are labeled flow by fluorescence. So, the particles
are fluorescent particles and an optical wavelength specific long pass filter is used to
remove the background light.
So, what happen that in micro PIV the particles which are introduced they are fluorescent
particles. So, the fluorescent particles when the light is introduced over fluorescent
ℎ𝑐
particles they will emit a light with a smaller energy. So, we know that energy E = . So,
𝜆
if a light of say 532 nanometer is incident on the particles the light emitted by the
fluorescent particle will be of a larger wavelength. So, a long pass filter is introduced which
allows only the larger wavelength particle say. For example, if 532 nanometer is the
incident light and if let us say the light that is being reflected is 600 nanometer.
408
So, the filter will be such that it will not allow the incident light or. So, in this case say if
we have a 560 nanometer lens. So, this allows the light having wavelength larger than 560
nanometer to pass through it, but it does not allow the 532 nanometer large of 532
nanometer wavelength light. So, this a long pass filter removes the background light and
the fluorescent light is sent to the image. So, the light fluoresced by the particles is
captured.
So, first we have discussed about the tracer particles and their sizes for a large channel
applications as well as small channel applications and the 2 criteria, when the particle size
is smaller than the wavelength and when the particle sizes is larger than the wavelength.
Now the light source that is used is often the lasers sometimes white light source is also
used, but it is not monochromatic. So, there will be problems in terms of wavelength, but
it is cheaper. So, for depending on the application and the accuracy required one can
choose the light source, with the development in the light emitting diodes or the LED lights
in recent years then use of LEDs for PIV application is increasing and every day one have
a new technique in which the LED light can be used for a PIV applications is being
developed.
Nonetheless the established PIV techniques generally use lasers more. So, because they
emit monochromatic light and have high energy densities. So, for the lasers the wavelength
of the light that is introduced and the and the energy density of the light is very important.
409
Now this for conventional channels this light is converted using the light sheet optics it is
converted into a light sheet to eliminate the tracer particles on a particular plane.
A laser, typically have 3 main components the laser material, which is which can be atomic
or molecular gaze gas so, which will be the gas lasers or semiconductor or solid materials.
The pump source which excites the laser material by electromagnetic or chemical energy
more and more lasers, which are used in PIV applications are diode pump lasers and the
third component is mirror arrangement, which is a resonator which allows an oscillation
within the laser material.
Now, so, the lasers this would have high power beam and sort pulse duration. So, the
power requirement that is proportional to the size of the sample larger size more power is
required. Now the 2 most often used lasers are Neodym-YAG laser and Neodym-YLF
laser these lasers emit lights at the Nd-YAG laser they emit light at 1064 nanometers and
it is harmonics of 532 nanometer and 266 nanometers generally it is 532 nanometers that
is used in the application.
Similarly Nd-YLF the ultimate light at 1053 nanometer and 526 or 527 nanometer and it
is the 527 nanometer, which is used in the PIV applications ok.
410
So, this laser is converted into a light sheet optics using a combination of a spherical and
a cylindrical lens. The cylindrical lens expands the laser into a plane. So, that converts a
laser into a plane and this a spherical lens it compresses the plane into a thin sheet. So, that
is required for PIV applications in conventional channels whereas, the illumination is not
a sheet illumination in micro PIV or to analyze the flow micro channels it is volume
illumination as we have seen before.
411
Now, the light sheet illumination is not used for micro flows because that light, which is
generated it may be few nanometers thick. So, it is not practical to use a light sheet there
and there will be significant deflection in the light sheet form optics. So, it is a volume
illuminated flow is used in micro PIV.
Now the depth of field and the depth of flow the 2 things are compared when discussing
the volume illumination. So, depth of field refers to the distance a point source of light
may be displaced from the focal plane and still produced and accept we acceptably focused
image.
So, the distance on the 2 sides of the focal plane where the and acceptably focused image
is obtained is known as the depth of field.
So, now the 2 cases can be there that when the depth of field of the system exceeds the
depth of the flow. So, in this case what will happen that all the particles that are there in
the field of view they will be illuminated, because all the particles are well focused. So,
one will not have an idea for example, in a when one has a light sheet optics then a
particular plane is illuminated.
But in case of volume illumination such is not the case and the entire volume is illuminated.
And if the depth of field exceeds the depth of flow, which is the size of that the flow all
412
particles will be well focused and so, the images will be captured of all the particles that
are there.
So, one will not have an idea that one will be able to capture all the particles. So, if say
one wants to take the velocity measurement, then one will know not know at what depth
the information is from. So, what will be obtained is the depth average velocity field and
the depth of the flow will not be taken into consideration.
So, depth is the third dimension of the flow will not be there; so to take into account this
when the depth of the field exceeds the depth of the flow one need to have advanced
processing techniques. So, this problem can be addressed.
The other option is that one has an optical system whose depth of field is smaller, when
compared with depth of the flow. So, only the particles that are there within the depth of
field will be focused. So, in a volume illumination rather than the illumination the depth
of the field from the optics are from the objective lens is such that that if particular region
in the channel is focused and of that particular region, the images will be obtained or they
will be having a high intensity.
413
So, apart from the 2 considerations what we have discussed one is size of the particle
especially the Rayleigh scattering criteria, which is applicable for micro PIV and the
volume illumination. The third criteria that needs to be considered in micro PIV is the
Brownian motion, because in the micro PIV the size of the particle is of the order of 10 to
100 of nanometers. Now Brownian motion is the motion of the fluid molecules of, because
of the oscillations or the from the molecular mass motion the motion of the fluid molecules.
Now, if the size of the particles is comparable to that of the fluid molecules, then when
they have collision with the fluid molecules, then the Brownian motion or the molecular
motion of the fluid may affect the flow behavior or it may also prevent the particles from
following the fluid carefully.
So, one need to take into account those considerations in micro PIV specially because the
size of the particles is small. So, they can have 2 effects one is that they can cause an error
in the measurement of the flow velocity. And it has been estimated that the error of the
flow velocity can be estimated as
1 2𝐷
𝜖= √
𝑢 ∆𝑡
where D is the diffusion coefficient and u is the flow velocity then epsilon is the error that
is introduced in the velocity.
414
And ∆T is the temporal resolution on that or the time interval between the 2 images. So,
this introduces a limit on the lower limit on the measurement time interval. So, ∆T should
be larger enough that this error is minimized in the measurement of the flow velocity. The,
another error that may occur is because of a uncertainty in the location of the flow tracing
particles, because the a Brownian motion it can especially for long exposure times, it can
cause an uncertainty in the location of the particles for a small particles it can cause an
uncertainty in the location.
So, this also needs to be taken into account the correlations have been developed for this
case also. So, one can take into account these effects to avoid the significant on the
Brownian motion of the flow.
Now, the 2 things that we have discussed at the case of particles and the light source and
the optical seat, if required or the volume illumination of the flow, now these images
nowadays in most of the PIV applications these images are captured using either a CCD
camera or a CMOS camera; both of these cameras convert light into electrical signals. So,
CCD camera is charged coupled device, which goes and gives very good and accurate a
special resolution.
The CMOS camera which is called complementary metal oxide semiconductor they are at
high speed applications because the CCD cameras, they have a limitation with the frame
rates. So, CCD cameras are preferred at low temporal or a resolution or at low frame rates,
415
CCD cameras are preferred. However, for high speed imaging applications only CMOS
cameras or a complementary metal oxide semiconductor cameras are preferred and used.
So, when considering camera one need to look at the a light intensity, the imaging speed
which is generally given as frames per second or fps and imaging resolution. So, it will be
in terms of number of pixels in the horizontal and vertical direction. For example, a camera
might have a resolution of 1280 × 1056 pixels or something like that.
So, this these numbers represents the pixels or the divisions in the horizontal direction and
in the vertical direction on a camera or on a camera sensor. So, the numbers represent the
number or division in the 2 directions.
So, once the signals have been obtained and the images has been captured then one need
to post process these images using cross correlation. So, cross correlation have to say for
2 functions f and g in the cross correlation Cfg is defined as a
∞
∫ 𝑓 ∗ (𝑥)𝑔(𝑥 + ∆𝑥)𝑑𝑥
−∞
where f*(x) is the complex conjugate of f(x), but because we have real signals here. So, f
*
(x) is same as f(x).
416
So, the cross correlation is given by this formula and using the cross correlation one can
get the peaks, in the signal and estimate ∆x from there.
So, let us say for f(x) = δ(x-xf) the first function, which is a delta function. And another
delta function g(x), which is given as x -xg the multiplication of 2 is given by a x –(xg -
xs). So, one can get the displacement as (xg - xf) So, one gets a peak here.
417
So, using this technique one can obtain the ∆x and we already have we already know ∆p
is the time gap between 2 laser pulses and ∆x is estimated from the or cross correlation of
the 2 images.
∆x ∆x
So, one can estimate the velocity as or x component of the velocity let us say vx is ∆T
∆T
∆y
. And vy as ∆T now the technique has also been extended for a number of other applications
for example, another technique, which is known as a stereo stereoscopic PIV where 2
cameras are used and the planar imaging are images are taken and one can get the 3
velocity components, which is known as stereo scopic PIV or the volumetric PIV again
here the entire 3 entire field measurements are taken or the time resolved PIV their high
speed camera, and high speed lasers will be required and the and using high speed camera
and high speed lasers one can also measure the velocity at now the high velocities.
Now, furthermore one can also use temperature sensitive or the concentration sensitive
fluorescent dies to measure the velocity at the temperature field or the concentration field
using the same principle as I used in micro PIV that one introduces a fluorescent dye,
which is temperature sensitive. And then using a long pass filter, the light only which a
light above a certain wavelength is allowed to pass through the filter and in go into the
camera. So, the images of this is taken and from there one can have an estimate of the
temperature field or the concentration field for heat and mass transfer application.
So, in summary what we have discussed today is the basic principle of particle image
velocimetry technique and the considerations that one need to take into account for micron
particle image velocimetry. We have a main components of the experimental setup is a
laser, which should be a dual cavity or double pulsed laser, which introduces 2 pulses at a
time interval of ∆T and illuminates the flow at 2 different time instants, with a synchronizer
the camera triggers at these 2 time instants and take 2 pictures of the flow, which is seeded
by the tracer particles.
These particles are supposed to follow the flow faithfully and should be able to scattered
enough light that the particle images or good images of the particle can be captured. And
by analyzing or by cross correlating, the intensity of the particles in the 2 images captured
at time T and T + ∆T one can have estimate of ∆x in different regions of the images. And
from that one can get the local fluid velocity of the flow or local velocity of the fluid in
the test section under investigation ok.
418
So, that is all for this lecture.
Thank you.
419
Lecture – 21
Inertial Microfluidics
Hello, the topic for today is inertial micro fluidics. So, micro fluidics is often associated
with small dimensions as the name suggests that the channel size is small. And the
velocities that are present in the micro channels will also be small as a result the Reynolds
number will be small in micro channels.
So, the flow is often laminar the laminar flow suggests that the flow will happen in
laminars, with very little or no interaction between the lines or no mixing as it generally
obtain in the turbulent flows or the eddies that are present in the turbulent flows.
So, this gives us a lot of control over the flow and because no heuristic models are required
in laminar flow, heuristic means empirical or everything can be derived from the
fundamental principles at least for single phase fluid micro channels.
So, the predictive power becomes very good and one can use this predictive power to
control the flow of cells, bubbles, droplet, us particle for a number of applications.
So, in this lecture we are going to look at or examine the fact, that is inertial effect are
important. So, when we say that the Reynolds number is low in micro channels, then
immediately or the next thing is done that the Reynolds number is we generally try to
neglect the Reynolds number or inertial effects.
420
So, in the Navier Stokes equation, where we have different terms this is unsteady term,
convective acceleration term, pressure gradient and viscous term we have not written any
other body forces in this.
So, in this case when the Reynolds number is low we often tend to neglect the convective
acceleration term and the unsteady term can be neglected if the flow is steady. So, this
suggests the flow to be in this stokes regime and why we tend to do this, because this gives
us a linear differential equation which can we solve analytically.
Now this may not be always the case, because we know that in a channel often the flow is
laminar below Reynolds number 2300 or so, flow is laminar. But so, for example, if we
have a Reynolds number 100, then the inertia is not negligible we cannot neglect the inertia
term, but the flow will still be laminar.
421
So, in the stokes regime what we have there are some beautiful results that one can see in
the stokes regime. For example, in the Mirror-Symmetry Time Reversal Theorem:
Bretherton suggested that when, moving in a steady unidirectional shear flow. So, the shear
flow can be represented where we have a gradient in the velocity.
So, in a unidirectional shear flow at small Reynolds number under the action of viscous
forces alone; that means, at small Reynolds number only viscous forces are important. So,
the inertial forces has been neglected the flow is stokes; to every orbit of a given finite
rigid body there corresponds one of the body of opposite mirror-symmetry. The
corresponding orbits are mirror images obtained by the reflection in a plane perpendicular
to the streamline, but are traversed in opposite senses.
So, this suggests that if a rigid spherical particle or a sufficiently symmetric particle, which
has which is symmetric about it is rotational axis no lift force will act on a symmetric
particle for a simple channel flow. So, the particles will not experience any lift force and
by lift force here, I mean the force on the particle in the transverse direction the direction
normal to the streamlines ok.
422
So, the flow is symmetric in stokes flow; however, in micro channels also we can
experience non-linear effects due to a number of facts. For example, if the flow is non-
Newtonian it may be following a power law model or it can be say a viscoelastic fluid or
and so on.
So, the non-linear effects can come into picture when the fluid is non-linear effect can
come into picture when the fluid is non-Newtonian, or when the channel in which the flow
is happening the structure of the channel is asymmetric that can also give rise to non-linear
effects in the flow. And if the shape of the particle as we have just seen that if the shape of
the particle is symmetric about it is rotational axis, then it does not experience any
transverse lift, but if the shape of the particle is asymmetric then we can see non-linear
effects. If the particles are deformable then deformability of the particle may also induce
a non-linear effect and say a lift on the particles. And finally, what we are concerned with
in this lecture is if the inertial effects are non-negligible.
423
So, inertial effects can be important in microfluidics for as we said just the few slides back
that the flow, in the micro channels is still laminar, but non-negligible say at about
Reynolds number 100. And this have the properties of the laminar flow in the sense that
the flow is deterministic in nature, you can a still solve the equations in the sense that you
do not need heuristic models, you may not be able to solve the equations always
analytically, but you can get a numerical solution without the need for any empirical
models or heuristic models for the turbulent flow.
And the fluids and particles can still be controlled and manipulated like we can do in the
stokes regime, plus the additional effects that come into picture that the non the flow
becomes non-linear and the motion of the fluid is irreversible unlike in the stokes flow.
And these effects can be quite important. Now these effects are often, because of say for
example, on a particle there is velocity difference across the particle or across a obstacle
length. So, these velocity differences or the velocity gradients are responsible for the
inertial effects.
Now, in the micro channel these velocity gradients, because the velocity gradient is
𝜕𝑣
𝜕𝑟
or they can be typically scaled by u over typical dimension of the channel. So, these
gradients become large in the micro channels and they can affect or the non-linear forces
424
or the non-linear effects can become important, when you compare with those effects in
the conventional channel.
So, let us look some numbers here if you compare a channel of diameter 100 micron and
a large channel say of diameter one centimeter. And the flow rate in the micro channel is
𝑄
about 150 micro liters per minute. So, the velocity in this case will be the Uavg = 𝜋 . So,
𝐷2
4
and the maximum velocity will be 2 Uavg. So, from that we can calculate the maximum
velocity. So, this maximum velocity comes out to be about 0.38 meters per second and the
velocity gradient the velocity gradient across the channel will be the maximum velocity
divided by the radius of the channel. So, this velocity gradient is about 0.37 meters per
second divided by 50 micron.
Now, if we want to achieve the same velocity gradient in a large channel. So, in the large
channel where the channel size is one centimeter, which is about 10 mm or 10000 microns;
so the radius will be 5000 micron. So, across a 5000 micron radius one will need to have
37.5 meter per second of velocity to achieve the same velocity gradient. So, that is why
we have this maximum velocity of 37.5 meters per second.
Now if we want to have the 37.5 meters per second velocity then the Reynolds number in
this case become 2.5 × 105 or 2.5 lakh, which means the flow is turbulent; whereas for
the same velocity gradient the Reynolds number is only 25 in the microns (Refer Time:
425
12:33) channel. So, the fluid is still laminar we can achieve such flows probably in a large
channel, but the flow will be turbulent and highly chaotic. So, we will not be able to
manipulate the particle or we cannot control the flow of the particle in such a case.
So, that is why this effect becomes quite important in micro channels and it has been
utilized or it has been used to good effect to develop various microfluidic devices for a
number of applications especially in biological applications in cell sorting and so forth.
So, this inertial micro fluidics the term refers to use of inertia of the fluid in microfluidic
systems to do different things in channels especially to manipulate the particles to focus
particles to concentrate them or to bring an order to the flow of particles, separate the
particle transfer and mixed particles and fluids. And this requires a clever use of the
particle properties, fluid properties, fluid flow rates, Reynolds number, channel structure,
or the general orientation and so on.
So, inertial focusing is one of the related topic of inertial micro fluidics or one of the effects
that has been utilized in recent years for microfluidic applications very extensively. The
other effect which is combined with inertial focusing is the secondary flow. So, secondary
flow refers to the flow in a direction normal to the main stream line flow. So, if any channel
the flow is happening in the streamline direction then at a cross section normal to this the
flow velocities normal to the streamline are generally termed as secondary flows, which is
in addition to the primary flow which is in the stream line direction.
426
So, this secondary flow can also often be combined with the inertial focusing and this.
Secondary flow can be achieved by having some changes in the geometry of the channel
for example, in the curved channel one can have secondary flow, or if one can have
grooves from the channel, or obstacles in the channel, or some pillars used as obstacle in
the channel. So, these structured channels can alone or coupled with the inertial focusing
can have different effects in the microfluidic applications.
So, coming to inertial focusing in of 1962 Segre and Silberberg they observe they did an
experiment that in a circular channel of one centimeter diameter, they introduced some
particles randomly and different locations. So, node preference for any location and what
they observed that after some distance the particles focused at a particular position. And
this position was about 0.6 times of this if this is R then the radius of this annular location
was 0.6 or about 0.63 times of the channel radius.
And this effect they termed as tubular pinch effect, because this happens at the time when
this they observe this there was no explanation for this phenomena. And after that there
was lot of effort to explain the phenomenon.
Now, because the particles which are randomly introduced in the channel they settle at a
particular equilibrium position. So, there is radial motion of the particles of course, this
radial motion will be driven by a force on the particles. So, that means, there has to be a
427
lateral force or what we term as lift force on the particle, which drives the particles in the
radial or the transverse direction.
Now, because there is an equilibrium position; so there has to be or there have to be 2

opposing forces, because if the force acts away from the center of the channel then the
particle should go and stick to the wall or if the force is directed towards the center of the
channel, then all the particles should be in the middle of the channel.
However, this equilibrium position was observed near the or it was observed between the
center and the wall of the channel. It must be pointed out that the concentration of the
particles was small. So, they are not particles are not very highly concentrated in the
experiments.
So, there has to be at least 2 forces which are acting in the opposite the opposite direction
one force that acts towards the center and another force that acts towards the wall. And by
the compare by the competition or the equilibrium of these forces the equilibrium positions
will be attained. So, there has been a number of analytical studies and computational
studies to understand these forces.
And now, it is established that there are 2 forces on the particle that are responsible for this
tubular pinch effect. So, one force is caused by the shear gradient lift force. So, the shear
flow is the force, because of the gradient in the velocity or the shear flow where we have
428
gradients in the velocity. And the shear gradient lift force is caused by the gradients in the
shear so; that means, the velocity, gradient, are not uniform, but they are varying along the
channel.
So, this example for example, in a parabolic profile which is quadratic you can see this the
𝑟2
velocity profile is 2 Uavg (1 − 𝑅2 ) in a channel.
So, this force in a channel and the force on the particle acts away from the wall this was
seen by Ho and leal and cox separately. So, they observed that the particle will have when
it is in a shear gradient force then it will have 2 different velocities. And these velocities
will not be equal and to compensate for that the fluid will the flow field will change and a
net force will act on the particle. And this net force will be in general be directed towards
the side with a higher relative velocity and it. So, happens that in a channel this higher
relative velocity will be always towards the wall. So, the force will be directed towards the
wall.
The scaling for this force has been given as a lift coefficient which is called CSG or constant
2
𝜌 𝑈𝑚𝑎𝑥 𝑎3
for shear gradient Dh Dh the hydraulic diameter of the channel and this CSG is not
𝐷ℎ
necessarily constant as a function of the Reynolds number of the channel and the position
of the particle; so the distance of the particle from the center of the channel.
429
Another force that is required that should be acting away from the wall and that comes
from the wall lubrication force that when particle moves near the wall, then 2 things
happen; one because of the wall effect the particle lags behind, the particle lags behind the
fluid. And the streamlines the bulk of the fluid moves away and a pressure builds up in the
constriction in the constriction between particle and wall.
So, when the fluid that moves bulk of the fluid moves away from the wall the steam lines
become dense at the pressure is high compared to the other side of the particle of course,
there will be a force directed away from the wall. And this force is scaled as a constant,
2
which is constant for lift coefficient of wall lubrication ; so there while this
𝐷ℎ4
velocity scaling is same, but the dependence of the 2 forces on the size of the particle and
the size of the channel is different.
And of course, the CWI is also a function of the Reynolds number and the position of the
wall. So, the dependence of the velocity is not necessarily, what we see this equation
because the Reynolds number also have the velocity.
So, by these 2 opposing forces the effect of the or the equilibrium position of the particles
due to the inertial focusing was explain or this tubular pinch effect was explained.
430
Now, one might also think of some more lift forces for example, a lift force acts on a
particle, which is rotating in a uniform flow that there might be a relative velocity between
the particle and the fluid.
So, if the particle is rotating with a velocity ω and there is a relative velocity between the
particle and the fluid. So, the fluid which on the side where it supports the rotation of the
particle supports, the velocity then one will have a higher velocity. Let us say U 𝜔 R on
the other side the velocity will be reduced U - 𝜔 R not necessarily the same magnitude,
but.
So, that will create velocity difference and by considering the Bernoulli theorem we know
that when the velocity difference is so, at the lower side we will have high pressure. So,
the force will be directed from this side to the other side. So, there is a lift force. And this
lift has been defined here the magnitude of this lift can be given by this formula has given
by Rubino and Keller, but the magnitude of this lift force on a rotating particle was
significantly low, when you compare with the shear gradient lift force or the wall
lubrication forces on the body ok.
Then the other force is very closely related to the shear gradient lift force. So, in this case
there was a not a gradient, but a simple quite Couette flow. So, those do not know Coutte
flow Couette flow is the flow in which there the velocity profile is linear. And that is often
achieved by flow between 2 parallel plates in which the upper plate moves with a velocity
431
and lower plate is fixed. There is no pressure gradient in the flow, because if there is a
pressure gradient and one can superimpose, the parabolic velocity profile and the linear
velocity profile and the resulting velocity profile we will look something like this
depending on the pressure gradient and the velocity so anyway.
So, the Couette flow is the simple shear flow. So, Saffman studied that lift force in a simple
unbounded shear flow can be given by this formula. Where V is the relative velocity
between the particle and the fluid γ is the slope of this; that means, it is the velocity gradient
𝜕𝑣
and a is the radius of the particle ρ μ under properties of the fluids. So, he introduced a
𝜕𝑦
force which acts on the particle.
Now, this force was one magnitude one order of magnitude larger then Rubino Keller or
FRK, which is because of the rotation of the particle, but this and it will act on the sphere
because of this which the spare will migrate towards the side where the fluid velocity is
larger, but the magnitude of this force was also not able to explain that it was also
significantly small or negligible when you compare with the shear gradient lift force.
So, now having understood the forces or the physical effects behind that tubular pinch
effect, if you want to develop a device in which the fluid velocity we want to calculate the
focusing positions, then it is important to know that; what is the channel length that is
required for the focusing of the particles? So, the length let us say the particles are
432
introduced at this position and the length that will be required in which the particles can
focus from the random positions to their equilibrium positions.
Now, so, this can be calculated Lfocusing will be equal to Umigration into time. So, the
migration is the velocity lateral velocity of the particle and time is the time required for
the fluid or time required for the particle to reach and for design purposes we should
consider this tmaximum. So, which will be tmaximum will be equal to the let us say the maximum
distance that will be slightly less than R; if the particle which is near the center and want
to reach the equilibrium position. So, the maximum distance that we might need to travel
a fraction of R so, because you do not know that fraction. So, let us say that this distance
is R and the velocity the lateral velocity is Umigration and sorry this is not.
So, these are the 2 factors that will be required for calculating the Lfocusing, but that will not
be equal to like this. So, what Lfocusing will be equal to the streamline velocity? So, let us
say we take U or Umaximum into tmaximum. So, that will be the focusing distance the time in
which the particle will be traveling in a path like this, if we decompose these 2 paths 1
distance is this and the other distance is the radial distance.
So, what we want to find focusing or distance and this distance we has we have
approximated as R. So, the time to reach this position in both to cover both the distances
will be equal, but the velocities will be different the velocity in the axial direction we scale
it as Um and velocity in the radial direction is Umigration.
Now, to calculate Umigration. we can consider the balance between shear gradient lift force
2 𝑎3
CSG 𝜌 𝑈𝑚 2
2
and drag force so shear gradient lift force is CSG 𝜌 𝑈𝑚 𝑎3 , on the
𝐷ℎ 𝐷ℎ
particle.
That will be equal to the drag force on the particle, which will be because the drag forces.
So, will be acting in the radial direction and this will be because of the motion in the radial
direction. So, the velocity that we considered is 3 𝜋𝜇Um and because the velocity
magnitude will be small, so the stokes flow we can expect the flow to be stokes. So, that
is why we can use by stokes drag here and we can use the definition or the balance of the
forces to calculate the migration velocity and from that we can substitute it into t max and
get the length for focusing lens.
433
So, that is equal to Um into t max. So, R is Dh by 2 divided by Umigration which is CSG ρ U
m square a cube D h will go there we have also 3 pi mu. So, one Um will cancel out and
we will have 3πby 2μD h square and we will this will all 3 πμU into a. So, we also have a
here. So,μD h square 1 of a will cancel out and we will have ρ U m a square as the into
CSG. So, by this formula we can make and estimate of the focusing length.
Now, looking at the effect of the cross section of the channel so, in microfluidics it is often
convenient to have the channels, which are rectangular in shape or square in shape. So,
what we have seen till now that there are focusing positions or there is because the
geometry of a cylindrical or circular cylinder is symmetric about the axis. So, the particles
will focus on a annular the location and, but in a rectangular channel it is not so, obvious.
And so, number of experiments have been done especially a by Decarlo and others. So, it
has been observed that there are 4 focusing positions in a square channel and this position
is about 0.6 at a distance of 0.6 dh by 2 from the center of the channel.
Now if the particle aspect ratios increased; that means, the not the particle aspect ratio,
but the channel aspect ratio is increased; that means, the channel becomes rectangle
rectangular, then at moderate Reynolds number it has been observed that the focusing
positions changed from 4 to 2.
434
This is explained by Decarlo and others by the magnitude of the forces that act on the
particle. So, in a rectangular channel, what it shown here is the force field in a quarter of
a channel. The particle experiences similar forces from both the sides of the wall, because
the velocity gradient are same. So, the particle focusing position it is somewhere in one of
the there will be one focusing position here and another focusing position somewhere here.
However, when it is rectangular channel the velocity gradient in this direction so, this is a
part of a rectangular channel and that is what is shown here. And while the velocity this
distance and this distance is different. So, the velocity gradient in this direction, because
the distance it is small. So, gradient is large is this gradient is small. So, the force on the
particles is larger here then in this direction. So, this position seems to become unstable
and the particles focus on these 2 locations. And this has been utilized in a number of
applications to change the equilibrium position of the particles to reduce the equilibrium
positions of the particles or 2 only 2 positions; so that the particles can be separated easily.
435
However, the Reynolds number can also affect the equilibrium positions though only
slightly, by increase it has been observed that the shear gradient lift force is a stronger
function of Reynolds number. So, the increase in shear gradient lift force is relatively large,
then the wall lubrication force. So, what will happen that the force on the particle the shear
gradient lift force will be larger than the wall lubrication force? So, the equilibrium
positions are they will shift towards the wall.
In rectangular channels this might sometime also lead to the focusing position increasing.
So, the number of focusing position can become again 4.
436
Now, what we have been discussing till now is the motion of rigid a spherical particles or
the focusing of rigid spherical particles, what if the particles are deformable? So, the
deformable particle can be classified into 3 different categories or where they are they can
be elastic solid particle, where the particle outer wall behaves elastically and the inside of
the particle also behaves as an elastic solid.
The other category is droplet us where you have a fluid. So, the behavior at the interface
is governed by the interfacial phenomena or the surface tension. So, that elastic effect is
coming from surface tension and there is fluid inside the particle, because of the shearing
motion at the interface the fluid inside will also have a motion and that will reduce the
drag on the droplet.
So, the deformable droplet is another category and the deformable capsules. So, in the
capsules the capsule might have filled with might be filled with a fluid, but the outer wall
is a solid elastic or a solid wall. So, that will have a that will act as a solid wall and the
fluid will inside will have will the filling inside the particle will be fluid.
So, broadly it has been observed that there is a lift force even in the Stokes flow and the
particles are deformable. So, when you look at the inertial effects on the deformable
particle and there is superposition of the or there are 2 different forces, which are
interacting the inertial lift force and the lift force induced by the deformability of the
particle. And the by comparison of the equilibrium positions for deformable and rigid
437
particles, it has been observed that the equilibrium position shift towards the center of the
channel for deformable particles when compared with the rigid particle.
So, that means, that the deformility deformability induce lift it acts towards the center of
the channel, utilizing this fact tanaka et al. developed a microfluidic device to separate the
cancer cells, which are deformable from the rigid beads they did experiments with other
cells also, but what we are discussing or what is concerned here that the cancer cells, which
are deformable and it their motion was compared with rigid particles and they found that
the focusing length that was required for rigid particles was more.
So, as we have just seen that Lfocusing will be proportional to Umigration. So, this migration
velocity or Lfocusing will be inversely proportional to Umigration.
So, the migration velocity higher the migration velocity lower the focusing length required
or the lower the time required for the particle to reach there. And this migration velocity
comes from the shear gradient lift force. So, higher the shear gradient lift force, higher the
migration velocity and smaller the focusing length.
438
Now, we look at some of the applications of these inertial fluidic effects so, the first
application is in sheathless flow Cytometry. So, the flow cytometry refers to the focusing
of the cells for cells sorting or for cell counting in a number of biological applications.
And this is done by hydrodynamic focusing in a standard cytometer, where the processing
rate or the throughput is about 10000 cells per second. They have shown it has been shown
by Hur et al. that the processing rate can be increased up to 1 million cells per second by
using inertial micro fluidics. One of the there are 2 factors that account for the higher
processing rate; one is the higher flow velocity and the other is parallelization is possible
in the inertial microfluidic channel.
So, in the inertial micro fluidics application or the inertial focusing applications so, number
of they had probably 256 channels in parallel. And those particles they achieve the
equilibrium position there was no extra fields force fields or device was required. So, this
method of focusing is Paseo focusing, where the structure of the channel and flow velocity
are utilized to focus the particles and then the counting of the particles can be done.
439
Another application is in the separation of the particles, where the blood cells they are
separated from the bacteria. So, the size of the blood cells are larger than the bacteria and
due to the inertial force the particles the blood cells, which are significantly larger in size
they focused to the equilibrium positions.
And then by using a expansion channel expansion the focusing position was shifted
towards the sidewalls and then these particles were separated collected. So, the particles
were separated from the pathogen bacteria in the fluid. So, that is another application.
440
Then the third application is in fluid exchange around cells and particles. So, here the
particles are present in one fluid and the particles are exchanged or migrated towards
another fluid by so, in this application the rectangular channel was used. So, we have there
are 2 focusing positions and by using the different flow rates of the fluid the focusing
positions were manipulated.
So, that the focusing positions are in the second fluid. So, the particles when they arrive
from the 2 separate channels into this channel, they migrate to a position where the
focusing position is in the second fluid. So, a particle migrate from the first fluid to the
second fluid and then these fluids are these particles are separated.
Now, so, till now we have looked at the flow of particles in a structured channel, in a
straight channel without any structured in the flow or the channel without any modification
in the channel geometry except the cross section of the channel, where the cross section
we have looked at the cross section of the channel to be circular or rectangular or square
shaped. Some people have also worked with trapezoidal channel and have got some
interesting results there.
In the structured channel what we are looking at that by changing the channel path or by
creating some structures in the channel some secondary flows can be generated. So, we
have listed here 3 different structured channels the channel shape which is which can be
curved. So, we will look into it a bit later.
441
The other structured channel is grooves or herringbones on the channel walls. So, there are
grooves created on the one wall of the channel. And what this does that this causes a
secondary flow in this plane in this plane is shown here. So, you can say that in the x y z
and this is an xz plane. So, there are there is motion in the secondary direction or the
secondary flow in this case.
And this causes good mixing and this because there is secondary flow. So, the focusing
position this will also change the focusing position of the particles. So, one can utilize such
structures to good effect to change the equilibrium position say from 2 equilibrium position
to one equilibrium position or can use different applications using the structured channel
coupled with inertial focusing.
The other structured channel is creating the pillars in the channel. So, if one creates the
simple geometry in which the pillars are kept at different locations regular at the regular
interval in the space and once is secondary flow in the particle. And then combined with
energy and focusing one can again have different equilibrium positions change the number
of positions to develop applications.
So, now coming to curved channels, in the curved channels there is an interesting
phenomenon.
442
When the flow is the flow happens from say into the page in this direction. So, the inner
wall which is the convex wall and the outer wall, which is concave wall there will be a
force, which is centrifugal force on the fluid and that force will shift the velocity profile
towards the inner wall.
So, the main velocity profile velocity profile in the main direction will be shift it in the
straight channel it is supposed to be parabolic whereas, this maxima will shift towards the
concave wall. And this will cause velocity gradients to be sharp here and that will set up a
pressure gradient in the transverse direction. As a result the fluid will move from the center
of the channel towards the wall and then again come back. So, the recirculation will set up
in the secondary direction or in the transverse direction.
So, at low Reynolds number this is that the stoke regime such effect is not observed, but
when it is inertial regime.
Then there are vertices that are observed at very high Reynolds number the number of
vertices are observed to increase. And this effect is characterized by Dean number, which
is Reynolds number into d by capital D and this Dean number is ratio of viscous forces
and centrifugal force and inertial force small d is channel dia and capital D is 2 into radius
of curvature of the curve channel. So, this has been utilized this effect has been utilized in
a number of applications in microfluidics.
443
For example, the flow is that, because the flow is laminar in micro channels. So, if one had
single phase flow one would like to enhance the heat transfer or mass transfer. Then
utilizing this effect one can have intense mixing of the fluid. So, this has been used as a
means of mixing in channels and it has been observed the heat transfer coefficient or the
mass transfer coefficients can increase to 2 to 3 fold when you compare with single phase
fully developed flow in a channel. It has also been used for a hydrodynamic focusing or
creating an inertially modulated variable focal length lens or it has been used for inertial
focusing which we are going to discuss.
So, in the inertial focusing this lateral motion of the particle, because particles are moving
in the lateral direction because of the inertial lift, and the secondary flow that further
enhances the lateral motion of the particle; so this helps in particle reaching the equilibrium
position faster and that will result in the at distance, that is required for focusing become
this is smaller. One can also play around with the number of focusing positions and it has
been shown that the focusing length remains unaffected at low Reynolds number or dean
number, but at higher dean or Reynolds number the focusing length can become
significantly shorter.
Equilibrium positions can be modified. So, these equilibrium positions can be modified,
because when the particle is moving it will experience more drag and that drag additional
drag on the particle, because of the secondary flow. So, it will experience more drag and
444
the particle equilibrium position is going to change. However, if the drag caused by the
secondary flow by of or because of the dean flow is more than this can lead to mixing.
So, there are 2 different kind of curved channels has been used; one in which the channel
is spiral. So, the curvature is in one direction as you can see here and in this case the dean
flow will be steady and this is often employed for differential focusing and separation of
particles. So, you can see here that the particles separate and moved to a particular position
here. So, this is a spiral channels.
445
The other one is the curved channel where it is alternating curvature, but notice that this
curvature is not symmetric it is not same here and in this. So, one might need to make sure
that these opposing channel segment segments they are not symmetric. So, that the
curvature effects are not nullified by the 2 symmetric portions, which are alternating. And
in this it has been shown that the focusing length can be reduced and number of focused
streams can also be reduced in the in the channel. So, this has also been used for a for
applications of cell separation or cell focusing.
So, in summary we need to take into account the inertial effects in microfluidics and this
has led to a the inertial effects or the non-linear effects, in microfluidics has led to a number
of interesting applications, in particular in the inertial focusing. So, in this lecture we have
first look at the tubular pinch effect, which suggests that in a channel at high Reynolds
number in laminar flow the particles achieve a equilibrium position. And this equilibrium
position is a function of channel Reynolds number, channel cross section and the structure
of the channel.
And these have led to a number of applications, where microfluidic devices can be
developed by playing around with the channel structure, the particle shape, the particle
deformability and so on and so forth.
446
Lecture – 22
Microfluidics: Applications
Hello. So, we are now about the last stage of this course and in this lecture and the
following lecture, we will be discussing about the different applications of Microfluidics
and these applications may include the applications which are at the research stage, which
have been already realized in the industrial applications or some of the potential
application that they have potential application, but have not been explored fully yet.
These applications may be of the particular focus that I have tried to keep is on multiphase
microfluidics, but some of these applications can be op of microfluidics in general as well.
So, the microfluidics has the potential to impact in a number of a spheres of our day to day
life and I have listed some of these applications in this slide. This list is by no means and
of adjustable list.
So, the first application I have listed is Inkjet printing and we will be talking about it in a
bit detail later on. So, inkjet printing is the printing technology that we all of us have heard
447
of and it is widely used for office and desktop printing apart from the newer application
that the technology is finding every day.
The Another Application Electronics cooling; so a lot of research on boiling and heat
transfer in microfluidics or 2 phase fluids in a flow of 2 phases in micro channels and
boiling in micro channels is driven by its application in the electronics cooling industry.
In the electronics cooling, we need to remove very high heat fluxes and the space
constraint. Due to the space constraint, we need to have the compact heat transfer
mechanisms.
So, the Electronics cooling has lot of applications of microfluidics or micro channel based
exchanges. Then in the Oil and gas industry though the potential of it is not fully explored,
but as the oil reserve oils which contains the oil the systems that are used in the oil an
industry for the drilling and exploration, they are of course, gigantic if not and they are of
considerable size and then, they cannot be by any means micro channels.
But the reservoirs in which the flow takes place the pores they are of the micron size and
lot of research in the oil and gas industry. A lot of research in the flow of about flow in
micro channels has been driven before the advent of microfluidics or before the
considerable effort in microfluidics took place. Lot of research say for example, on Taylor
flow in micro channels or Annular flow in micro channels or Capillarity in micro channels
barrage driven by its application or to develop models for reservoirs.
In the Aerospace Industry; so, the goal in the aerospace industry is to achieve high
performance having while having the system to be very compact system because one need
to reduce the payload or one need to reduce the load that it want to carry for any system.
So, the aerospace industry always looks for the solutions which can provide compact
system designs and one of that is say Micro systems and apart from that say combat heat
exchangers micro thrusters etc. is also in the aerospace industry.
Automotive Industries; in the automotive industry one of the most common applications
is monolith reactors which are used for capturing the exhaust gases from the engine. Then,
in the Chemical Process Industry, as we have discussed during the course that there are a
number of applications in the chemical processing industry and when it comes to
multiphase flows in general, there is the drive towards micro processing where one want
to achieve high interfacial area density.
448
So, example microwave operators, micro contactors, micro distillation column or gas
liquid reactions where the reaction is limited by mass transfer or the diffusion limited
reaction. So, the diffusion paths are small and the interfacial area density is high, so
chemical processing industry. Actually, in the chemical process industry the drive is
towards developing the entire plant based on the micro reactor systems or a micro channel
system.
So, one need to develop while there has been lot of work on developing micro reactors,
now the drive is towards a scaling up or better to say numbering up of these reactors and
developing additional and ancillaries is to integrate the plant and the system into a
chemical plant then, in Membranes. So, we have a number of different membranes where
fouling is an issue and the gas sparging has the sparging of gas by having bubbles which
create the fluctuations in the flow and as a result the wall shear stress on the membranes.
So, that also have an application in membranes. Apart from that the flow in the membrane
pores, the pores are of micron sites if we want to understand what is happening at the pore
level. Then, one need to understand this using microfluidics or microfluidic systems.
Then, the 2 most common applications from which the microfluidics has grown in the past
few decades are in the Analytical chemistry and in biology and biomedical and health care
industries.
So, in analytical chemistry, where were want to deal with a small amount of the agents
which may be carcinogenic or which may be which not we are not be so safe or risk is
high or the cost is high.
So, in such cases one can deal with Pico liter, Nano liter volume of the droplets and do the
reactions one can also have flow chemistry solutions has come up for from microfluidics
to do flow chemistry in micro reactors.
So, that is another application, in biology, one can have because of the predictive power
and the control that one can have in microfluidics, a number of applications are there and
people are looking to use it for separation techniques, cell separation or a number of a
DNA separation or a number of other applications.
In Biomedical a number of diagnostic devices are being developed for especially the
remote area applications where one do not need to use the laboratory high end machines
449
in the in the laboratories. But one can use the diagnostic devices, microfluidics base or lab
on a chip based devices for the different diagnosis.
Then, when we come to the our Physiological systems, the Cardiovascular system or
Lungs, they all have micro channels, the arteries apart from the larger arteries; for example,
Aorta, the Capillaries and Venules, Arterioles.
They are all of micron size flexible channels and to understand the flow and understanding
the rheology of blood in such systems, understanding the air flow and the mucus air
interaction and so on.
One need to develop micro fluidic systems which can mimic such physiological system
and one can understand the flow behavior and different flow and biological reactions
interactions etc. in those systems.
Apart from that in the tissue engineering when one wants to develop tissues or the then,
one need to have an understanding and that also is part of microfluidics. In the
pharmaceutical industry, the continuous processing most of the things that in happen in
the pharmaceutical industry most of the reactors are batch reactors because of the small
amount of volumes that are generally handled and the use of the reactor for a different for
a number of different reactions.
So, the industry prefers batch reactors, but the there are certain disadvantages in the batch
reactors. For example, one may not have very uniform uniformity in different batches that
are there and continuous or the cleaning time etc.
So, there has been in recent years or drive from some of the pharmaceutical industries also
to look for the continuous processing method for pharmaceutical applications. For
example, Novartis collaborated with MIT to develop continuous processing solutions for
some of their applications.
450
So, let us look at a bit in an Inkjet printing which is an interesting problem and interesting
application of some of the things that we have learned. And it also gives us a number of
problems that we can further study on based on the physics that we have learned in this
course. So, Inkjet printing is different from the conventional printing in that; when we have
conventional printing generally for printing a newspaper or printing a books, a master
pattern is prepared and then, from that master pattern it is transferred a number of times to
the paper or the book or the or the cloth on which it is to be printed.
The inkjet printing is different in this manner that in this the jet of ink from the jet of ink
or the droplets of ink are deposited on the surface and this deposition takes place without
a need of the physical contact between the print head from where the ink is being ejected
and the substrate or the surface on which the printing is to be done.
So, there is no master pattern here and one can do the same thing or one can print the same
thing again and again or one can change the printing pattern every now and then. So, that
is one of the biggest advantage that inkjet printing has. It decouples the print head and the
substrate elements and allow machines comprising hundreds of print head with multiple
jets. So, one can scale up these print heads and do and using multiple jets do the required
printing.
Now,. So, as one can think of that in the inkjet printing, the jet of ink is generated or the
droplets of ink are generated and then these droplets are deposited on the paper on the
451
cloth or on the tile on the surface on which the printing is to be done. So, we need to
understand because the printing requires very fine length and time scales to be dissolved.
The accuracy or the or the refinement of the printing would come when the exact amount
of droplet volume, a very small amount of droplet volume can be deposited on a surface
accurately and then, this droplet can be evaporated in the minimum possible or the in the
required time in any time which can be estimated beforehand.
So, the entire process depends on the accurate knowledge of the physics that is involved
there and there is lot of microfluidic knowledge a lot of multiphase flows bubbles not
bubbles; but, droplets and the flow of droplets and the interaction of the droplets which
involve. So, we can divide these problems into three categories. First, the generation of
jets and droplets. So, in the continuous inkjet printing for example, one will have jet and
from these jets the droplets.
So, the formation of jets and droplets from the fluids and then, this droplet will travel to
the surface. So, the travel time, the travel path and if the droplet is to be discarded or is to
be collected or is to be deposited on the surface and the time between the 2 droplets all
those will come in the journey. So, from the jet to the or from the print head to the substrate
the journey of the droplets and then, spreading of the droplet on the surface and its
interaction with the droplets adjacent to it and finally, the drawing of the substrate.
Now, the technology has been developed in 3 phases or what we can say that the
development of inkjet till date can be divided into 3 periods.
452
The early developments for the inkjet were about 70s and 80s where, it was developed to
mark and code to develop or to have barcodes for the food products or to do the print the
dates and prices on the packagings.
So, these were the applications where one needed high speed the same job has to be done
again and again and it has to be done at high speed. But, high resolution is not mandatory
and the technology matured slowly then, mid 90s onwards about 1985 or so. The drop on
demand printing came into picture.
So, from the continuous inkjet printing the drop on demand printing came into picture and
one could achieve higher resolution and consumer printing and the small office and home
printing took into picture which is the most common way, we know inkjet printing.
Then, in the 21st century, there has been a drive to use inkjet printing for as a
manufacturing process. So, it is a bottom up approach for manufacturing and by depositing
the droplet of different substances, the potential of the technology is being explored to
develop different materials. It may be the chips, PCB circuits or it may be printing on the
deserts or it may be printing on the tiles or it may be printing on the carpets, it may be
printing on the notes or it may be printing off say 3D printing, what we have the digital
manufacturing; so the printing of tissues, printing of cells or printing of different materials
twice that we see in the market ok.
453
So, here are listed some of those different applications that we just talked about of inkjet
printing and some of these application has already been developed and some of them are
in the developing stage and others have potential to be develop and they are at the
researches stage only for example, the last ones.
So, if you look at inkjet printing technology, we can divide the technology into 2 or the
inkjet printers into 2 different categories. One is Continuous inkjet printer or which is
called CIJ in short or Drop-on-Demand inkjet printers or DOD. So, as the name suggests
454
the Continuous inkjet printer, a continuous jet comes out from the reservoir of the ink and
then, jet disintegrates into the droplets and then droplets are deposited on the substrate.
Whereas, in the droplet on demand systems, the droplet which are required only those
droplets are collected or only those droplets come out of the reservoir; whereas, in the
continuous ink jet printer a continuous stream of droplets are coming out. So, the droplets
which are to be printed on the substrate, they are printed on the substrate other droplets are
deflected and collected for the use.
So, that is a challenge in continuous inkjet printer, but in the drop on demand, only those
droplets which are required for the printing only those droplets, then and there are
generated. So, Continuous jets have again 2 types of technologies the first one is as we see
Single jet; so, in this one jet is generated.
And then, Multiple jets. drop on demand, their classification is based on the technology
that is used for the generation of the droplet. So, the 2 most common methods are Thermal
and Piezoelectric. There are other methods, but I have not listed those here.
So, in the continuous inkjet printer for the single continuous inkjet printer the jet of the ink
and this ink is generally it has a some solvents into it. So, it might behave (Refer Time:
22:07) like as a viscoelastic fluid a bit and so, the jet of ink it comes out from the nozzle.
455
And then, it is perturbed a bit this jet is perturbed a bit. So, that the droplets of a certain
size are generated from this jet, as we know from the relay plateau instability that a
continuous jet will disintegrate into a into the droplets. But the droplet size will be
different. So, by perturbing this continuous jet, one can manage the droplet size that is
required. And then, when it is passed to the charged electrode, these droplets are charged
this might be positively or negatively charged.
Now, when these droplets are passed through the electrode, once they have become
charged and they are passed through the deflection plates. So, the droplet which needs to
be used, they go through and the droplets which are not required or which are not to be
printed, they are deflected and collected in the gutter or other way round.
So, in this case what we see here, the droplets which are not required they go directly and
they are collected in the gutter and then, they can be reused. Whereas, the droplets which
are to be printed they are charged. So, this charging electrode can be done based on a
program there that the its electric charge the charging of the electrode can be in one
direction or it might be the electric field is such that that might be positively charged or
negatively charged.
So, the droplet material needs to be sufficiently conductive. So, that puts a constraint on
that ink material that can be used in this printer and then, these charged droplets are
deflected and printed up on the substrate. The typical nozzle velocities in these cases can
be as high as 10 to 30 meters per second ok.
Multiple jet is same technology, but in this case what we need is or what we have is
multiple jets and all of these are charged through an electrode and then, the deflection plate
plates can deflect the droplets or the droplets which are not to be used, they can be collected
in the catcher.
So, using multiple jet of course, will speed up the process the resolution of single jet to
inkjet printers is relatively poor and one can do the jobs which require high speed to be
done repeatedly, but the resolution or very good resolution is not mandatory.
456
Now, in the droplet on demand the Thermal inkjet printer as the name suggests the
technology for the droplet generation is based on thermal magnet. So, the in the jet or in
the nozzle, there is a heater and this heater might be placed on a at different positions.
So, the heater is placed and when a droplet is required the heater is heated from this, the
heat is given quickly. The ink becomes superheated and it evaporates and the bubble forms.
This bubble grows in size and sufficiently does grow that a droplet ejected once the droplet
has been ejected the heater switches off and then, the to fill that space the ink comes to the
heater.
So, by playing with the heat flux on this heater one can generate the droplets of different
size. So, this is the principle of thermal DOD inkjet printer. Then, another technology is
based on the Piezoelectric effect. So, Piezoelectric effect is based on the fact that for certain
materials as a result of electric field certain mechanical deflection can be achieved or based
on the mechanical deflection, the electric field can be generated for the from the
piezoelectric materials.
457
So, this Piezoelectric material is connected with the ink via diaphragm and this podia
position can be different as we can see in different images and once, the electric field once
we need a droplet, the electric field is applied.
And the diaphragm, the piezoelectric plate deforms or displaces the ink and once, the
droplet is generated and ejected the piezoelectric material comes back to its position or the
piezoelectric plate comes back to its position and the ink is filled in this system.
458
So, now the graph shows the properties of the material that should be used as printable
fluids or that can be used as inkjet. So, this graph is between Ohnesorge number and
Reynolds number. We can remember that Ohnesorge number is defined as the root of the
waber number divided by Reynolds number. So, this is equal to the waber number is ρu2d
over σ Reynolds number ρ u d over μ. So, we can cancel out u here and that will be equal
to μ over square root of σρd.
So, as you can see here that this Ohnesorge number is a function of the properties of the
fluids viscosity, surface tension density and the diameter of the jet or the droplet. And
Reynolds number of course, contains the velocity of the jet. So, all these effects surface
tension, inertia and viscosity.
So, inertia comes in the Reynolds number, they are important for the formation of liquid
jets and drops. Based on the experiments in the theoretical analysis it has been suggested
that the range for this printable liquids the inverse of Ohnesorge number which has been
defined as Z in the literature, it has to be according to one suggestion and it has to be
between 1 and 10. And according to another suggestion, it has to be between 4 and 14. So,
this according to that this is the range that we have in which the material can be printable.
Now, further on we also need that the jet that are being used that those jets must have
sufficient kinetic energy to be ejected from the nozzle. So, that is the line Re is equal to 2
by Ohnesorge number below which the energy is not sufficient for the formation of
droplets.
On the other side, as is shown by this dotted line, the material should also avoid or the
liquid should also avoid the splashing on impact on the substrate and for that the criteria
has been suggested as Ohnesorge number into Reynolds number raised to the power 1.25
is equal to 50. So, this is the region in which the fluid should be printable.
459
So, once what we have discussed in the technology that in the continuous inkjet printer,
one need to look at the generation of the droplets which is happening by giving
perturbation to the jets. And then the travel of this droplet from the print head to the
substrate, once it comes to the substrate; then, its impact on the surface and the droplet
solidification are 2 important factors.
Depending on the properties of the liquid and the surface and the velocity and the size of
the droplets different flow patterns can be can occur. And the droplet can have different
fates for example, it can deposit or it can promptly splash or the splash can be corona
splash or the receding breakup or it can rebound partially or it can completely rebound.
Apart from that so, we have seen in the previous slide that how we can or at least one
criteria that is therefore, to avoid splashing of the droplets.
And then, when these droplets they are splashing continuously and they are depositing side
by side. So, the droplets will interact with each other or may coalesce with each other. So,
the interaction of the droplets as has been suggested that they can have different forms say
at the form of beads or if they are close enough; then, they can have a pattern like this or
different patterns depending on the distance between the droplets.
Then, the next issue is the drying of droplets because the ink has a continuous fluid plus
some solute particle suspended into it. The ink also have a temperature effect for example,
460
in the Thermal DOD, the ink is being heated. So, when it deposits on the droplet, it will
have a temperature gradient the different.
So, the different regions of the droplet will have different temperatures and so, there will
be a temperature gradient set up which will cause a surface tension gradient. And result as
a result we will have marangoni flow there. The particles, the solute that may also cause
the change in surface tension and that may also have Marangoni effects and the solute
transport.
So, the drying also have what we call as coffee ring effect that the particles of the solute
they tend to take shape or they tend to reach out towards the ring, those effects. All those
effects need to take into account. So, the message here is that in the inkjet printing, it is an
interesting problem. We all know that it is being used very effectively for different printing
applications and it is finding newer and newer applications every day.
So, we need to understand the physics very clearly; while, some of this has already been
understood and lot yet need to be understood specially to develop applications. For
example, printing of issues or printing or 3 D printing using polymers etc.
So, there is lot that need to be understood from the fluid mechanics or multiphase flow
perspective.
461
So, now we come to the other application, Electronics cooling. As the Moore’s law
suggests which was given in 1965 and then again modified in 1975 that the number of
semiconductors or the performance of semiconductor doubles every month, every 18
months and this law has a stood test of time in past few decades. So, last 30- 40 years.
This has been happening that the performance of the semiconductor has been increasing
and as a result we see we have seen the tremendous growth of computing power in last 20
years. That is one of the reason that today, this video is being without any problems, it is
being transferred to you.
So, now soon it seems that if there are no further developments into the semiconductor
technology; then, this the this may saturate the Moore’s law may not no longer be valid
the same the semiconductor performance or may not grow at the same rate that we have
seen in past few years.
And one of the problems or one of the bottlenecks in this is the cooling requirement. So,
as we can understand that as the number of semiconductor chips that increase on a board,
the cooling requirement is also increased better because when all the chips work at high
performance rate.
They release lot of heat and that heat need to be removed for them to maintain a at a
particular temperature and this the temperature of the chips has to be below 85 degree
centigrade. So, an option is to increase the performance to be able to operate these at lower
temperature. So, once one can operate at lower temperature, one can afford to have higher
performance.
Now, according to an ITRS estimate by 2016, so in about 8 years, the power dissipation
from a microprocessor chip will exceed 800 watt and put this in perspective 15 to 20 years
back, it was predicted that by 2015, this value will be 270 watt and in 2011, ORACLE
SPARC T4 processor a high end processor. It has power dissipation of the order of 240
watts and the size of the die or the area was about 403 millimeter square.
So, this amounts to about 0.6, the heat flux of 0.6 megawatt per meter square; the INTEL
core i7 processor that we use in our desktops which was released in 2014. It has the power
dissipation of 88 watt and the die size in this is 177 millimeter square. So, it is about 0.5
megawatt per meter square heat flux. So, the heat flux that we are dealing with is about 1
462
megawatt per meter square already and once we achieve a higher performance, this what
we need to be looking at atleast 10 megawatt per meter square or higher those kind of heat
fluxes he should be able to remove.
Apart from that these chips will also have local hotspots, the heating is not uniform and at
these hot spots the heat flux that is need to be remove which may be 6 to 10 times or we
can say one order of magnitude higher than the average power and due to this it has been
estimated that due to this local hotspots, the performance of the chip decreases about 10 to
15 percent.
So, what are the methods that are used? Air cooling, when you see your laptop or your
desktop computers you can hear a fan running at the back and sometimes, when you will
laptop is old for example, then one fine day, you realize that the fan is not running anymore
and the laptop reaches to higher temperature. And if the laptop reaches to very high
temperatures or the set temperature and sometimes you may just see that the laptop just
turns off because of the heating.
So, this cooling is the air cooling is the common method that we find in most of the
computers and laptops. And it can dissipate power up to 1 megawatt per meter square. So,
we are approaching the limit of heat fluxes that can be removed by air cooling alone.
Another method that is used by using heat pipes and they are also their performance also
can, they can heat remove heat about same range 1 megawatt per meter square or lower.
463
Now, another technique that has been suggested in literature that Liquid jet impingement.
So, the jets of liquid that can be impinged on the on the surfaces and this can achieve high
heat fluxes up to 6.7 megawatt per meter square and, but it requires that the spray cooling
needs to be integrated into a closed loop system. So, designing such a system is a challenge.
And then, it also the creates an object require high pumping power and it also has the risk
of erosion of the surface at which it will be hitting.
So, a Single-phase and Two-phase flow in micro channel; so Single-phase means the flow
of water in micro channel because water have high thermal capacity than air. So, of course,
by changing the fluid to water or changing to fluid to coolants, one can have high heat
capacity removal. And this heat can be removed efficiently when we have Two-phase flow
then, the boiling.
So, due to the boiling the temperature of the surface can also be uniform. So, this uniform
the local hotspot problem can also be minimized. So, recently some of this has been the
recent research in flow in micro channels especially from the heat transfer perspective with
and without phase change has been driven by applications in electronics cooling.
And the single phase water flow in micro channels has been shown to achieve high heat
flux of 7.94 watt per meter square at 94 degree centigrade; quite few decades ago.
464
So, here is a image where IBM. They introduced in one of their high end servers or high
performance computers and evaporator for the cooling purposes. So, this evaporator this
had blowers, but the evaporators were introduced for cooling of some certain components
of the system and this was in about 2 decades ago.
There is another system again by IBM in which the water cooled IBM server has been
developed. So, this is based on micro channel and. So, the technology being utilized by
INTEL and IBM for the cooling of the electronic systems especially, the computers servers
data servers and so on.
In a comparison of air cooled and water cooled systems for a small supercomputer which
was rated at 300 kilo Watt. It was found that when the air cooling was being used, it used
44 percent of the power; whereas, in case of water cooling only 17 percent of the power
was being consumed for the cooling purposes.
In an another study for 3.7 kiloWatt system, small high performance system; it was
observed that single phase water cooling required 35 Watt; whereas, Two-phase pump
loop cooling 6.4 Watts and when the refrigeration cycle Vapour compression refrigeration
cycle acquired 747 Watt of.
465
So, the Two-phase where, the fluid is being pumped it required the minimum energy. So,
that is also an advantage of as we have seen here in the previous that the energy recovery
is also high in case of micro channels and the use of micro channels with pumped cooling.
These electronics cooling is also required for example, IGBTs the Insulated Gate Bipolar
Transistors which are used as a electronic switches in electric vehicles, rail tractions, wind
turbines, power supplies and motors. And it is often observed that the failure of this IGBT
chips occurs due to the thermo mechanical stresses which are which might be generated
because of local hot spots.
So, it where the high heat fluxes are being generated, the micro channel cooling is also
have applications in such IGBT; the cooling of IGBTs and other lot of different electronic
systems and servers and so on.
466
The other application of microfluidics that we are going to talk about in this lecture in the
is in the Oil and Gas Industry. So, in the Oil and Gas Industry, one application of
microfluidics is as we have seen in the chemistry is in the Analytical chemistry or
Analyzing the Oil and Gas.
So, for example, here is an image in which a fluorescence based method has been
developed by Professor Symptoms lab in Canada for the measurement of Minimum
Miscibility Pressure of CO 2 included our [vocalized-noise. So, what is Minimum
Miscibility Pressure? It is the lowest pressure at which the gas can develop miscibility with
oil at a given reservoir temperature.
Now, when the gas is not miscible, there is a surface tension or there is an interfacial
tension between the gas and liquid phases. So, there will be a sharp interface between the
2 phases. As we can see here that when pressure is less than MMP; then, the interfaces
sharp, as can be shown here this is gas and this is for the liquid phase.
When it becomes equal to MMP, then, there is no the surface tension is 0. So, in this case
what one will have there is no sharp interface and using fluorescent based method where,
the authors have taken the advantage of the inherent fluorescence of the crude oil.
467
So, based on this fluorescence microscopy, they could identify the MMP which is
comparable or even better than the conventional methods that are used for identification
of Minimum Miscibility Pressure. For example, the bubble observing the bubbles.
Another application is so, the first application that we talked about that in the oil and gas
industry. The different applications of microfluidics which have already been developed,
they can be used or they can be tailored for analysis different analysis of fluid and gases,
say viscosity of the oil mixture of the density of the oil mixture or the surface tension or
the dew point or the bubble point and so on.
The other application is that the oil and gas process in which we have reservoirs and the
transport of fluids, oil and gases across through the pores. And often specially for
secondary or tertiary recovery of the oil we need to put in the fluids in the reservoir for
enhanced oil recovery and that may be the polymers or may be a foam or it may be
specialty chemicals or it may be nano fluids or this is microbial based enhanced oil
recovery.
So, for all these enhanced oil recovery applications, we need to understand what is actually
happening at the pore scale. So, even before that when talk microfluidics there have been
development in people were developing the micro models.
468
Using and now, the developed and in micro fabrication techniques have a speeded or
speeded up or accelerated the process of investigation using these Micro models and one
can have the capability to develop novel micro models to understand the transport
processes the wetting phenomena, you know multiphase flow behavior in the channels.
For example, here is a calcite microfluidic chips. So, the researchers in this case they have
on a calcite material itself they have developed the micro pattern and studied the fluid flow
behavior to mimic and understand the flow behavior these micro channels.
Now, one of the challenges in this is because the pressure in the reservoir is very high and
it is often not possible for the microfluidics chips or microfluidics systems that are being
developed as of now to withstand such high pressures. So, that is one of the limitations,
but as we go ahead the solution to this problem will also be.
In another Micro model, there is this is say ‘Comb’ micro model for visualization of foam
behavior and heterogeneous reservoirs.
So, in heterogeneous, the flow of foam this is for the foam assisted enhanced oil recovery
application that how does the foam flow in a in heterogeneous micro structure which can
which can be a this can be a simplified or where is very simplified the structure of
heterogeneous structure in a reservoir.
469
So, these micro models gives one the power to understand the flow behavior with different
fluids. There have been also a lot of tricking around the wetting behavior of the of the
chips. So, that one can understand the flow of the fluid. The contact of the 2 fluids say oil
and water with the different areas in the chips.
With the applications of microfluidics in oil and gas industry, these are some of the chips
that have commercially available and they can be called as a reservoir chip where the pores
of the reservoir has been mimicked on the transparent microfluidic chips and these are
from Micronit Microtechnologies.
So, in summary, the Microfluidic technologies or Micro fluidics has lot of potential to
understand the actual flow behavior than that happens in reservoirs at the pore scale ok.
So, in this lecture, we have looked at the applications of Micro fluidics for Inkjet printing.
The relevant problem that we can apply the knowledge learned in this course to understand
and solve further the relevant problems inkjet printing, what are the applications or what
are the systems that can be cooled using electronics cooling we have talked about 2 or 3
systems in this.
And then, oil and gas industry the 2 applications, the characterization of the oil and gas
and the understanding the flow behavior at the pore scale, the 2 main applications of
microfluidics in Oil and Gas Industry.
In the next lecture, we will further continue some of these applications in other spheres.
Thank you.
470
Lecture – 23
Microfluidics: Applications
So in this lecture, we will continue the discussion of different Applications of Microfluidic

technology and in different areas. So, here is the list that we have seen or some of the
applications that we have listed of Microfluidics.
By no means this list is exhaustive actually while preparing for this lecture as I look upon
the applications of micro fluidics, I get to see more and more applications.
And then, I realized that if I would like to give a comprehensive overview of the fields that
Microfluidics can have an impact upon probably I would need a another course to discuss
the application, all the applications of micro fluidics. If I will be able to compile all the
applications in one lecture that will be humongous tasks in itself ok.
471
So, let us look at some more applications and in this, we will look at the some applications
in Aerospace Industry. So, in the Aerospace Industry, there can be different applications
of micro fluidics. One of them is in flow control of Aerodynamic flow control of the
aerospace vehicles.
These aerospace vehicles can be Aeronautical applications. For example, aeroplanes etc.
or it can be for space applications for rockets. So, these involve sensors, actuators or
thrusting devices. And these applications are generally based on MEMS. MEMS as you
would know that Micro Electro Mechanical Systems. So, the Mechanical electrical
Mechanical and electronic systems which are working at the micron size, they are called
MEMS and. So, some of those devices are say Synthetic jet for thrust vectoring.
So, here is a image of a Synthetic jet or it can be Supersonic micro jets, Oscillator, Fluidic
Oscillators or Turbulent boundary layer control. So, where these different jets or
oscillators, they can affect the overall flow and change the thrust vectoring for over the
aerofoils etcetera.
472
Another application is in Micropropulsion devices. So, with the application of satellite in

a number of areas, there have been a thrust towards developing smaller and smaller
satellites; so based on that the different names for satellites have also been standardized.
So, a satellite which raise about 100 kg is named as Micro satellite. A satellite which is
about 10 kg is a nano satellite. Pico satellite is 1 kg and femto satellite is less than 100
grams.
These this classification I have taken from the micro fluidics and nanofludics handbook
and these satellites they will have the MEMS based thrusters and these thrusters based on
maybe cold gas thruster or plasma thruster or combustion based thrusters.
Now, one of the challenges in this is that the impulse bit sizes or the small amount of
impulse that is required to control these small satellites. So, it is very important to have
that one can or one should be able to produce the amount of impulse accurately so that the
accurate or the desired control direction can be given to the satellite.
And so, the kind of impulse the kind of numbers that are required you can see from here
that for 1 kg mass the impulse required is about 10-4 to 10-6 Newton’s second. So, this
requires.
That one should be able to have the smallest bit of impulse. So, that one can have
increments over it, but if the minimum amount of impulse is required that can be given
473
that is the impulse bit size discussed here. So, that is a typical micro thrusters or a
schematic of a micro thrusters where there is a propellant and the heater element attached
to it.
This micro thrusters is only few mm in diameter and the exit diameter of the nozzle which
can be supersonic nozzle or subsonic nozzle. It is about 250 to 400 microns and as the
when this thruster is to be used and the element is heated.
And the propellant converts into gas, the pressure builds up and this diaphragm ruptures
and the gases come out and the desired thrust is given.
Another application in aerospace industry is in Micro power generators, in the in terms of

for example, fuel cells and micro turbines. So, these generators are expected to provide
power in the range of 1 to 15 Watt and this power might be required for miniature cameras
or mobile applications and other applications where power is required.
So, among the different fuel cells Direct methanol fuel cell is the preferred method. Here
is a schematic of the Direct methanol fuel cell and this Image is again taken from
Handbook of Micro and Nanofludics ok. So, these direct methanol fuel cells, they have
very high energy density of about 5960 Watt hour per kg.
In these, whatever we have learnt in this course, the gas liquid flow in micro channels. So,
some of these works in micro channels have been motivated by the fluid handling and
474
thermal management challenges in PMF fuel cell or direct methanol fuel cell. So, these are
the 2 biggest challenges in the in these fuel cells. So, in the aerospace industry we will
looked at 3 applications here.
The Aerodynamic flow control of aerospace devices so as to control the lift and drag on
the vehicles and then, micro thrusters for micro propulsion devices or for micro satellites;
then, micro power generators; for example, fuel cells and micro turbines ok.
Now coming to Chemical process industries which is the main area of this course because
this course has been offered under the Chemical Engineering discipline, so because this is
a multiphase flow course. So, what I have listed here is some of the Gas-liquid reactions
and in this image what you see is some of the micro reactors for Gas-liquid applications.
The first one is Falling film micro reactor, where the liquid film falls on the wall of the
channel and the gas comes between the 2 plates and the interface between the 2 gas and
liquid provide the contact area for the mass transfer between the gas and liquid phase.
Another application is Micro bubble column; you can guess or estimate the size of the
channel by looking at the match stick. So, the size of the channel window and we see here
is less than is smaller than a match stick.
475
Another one is a Packed bed reactor. So, where there are different channels and in which
it has been packed. So, these are different type of contactors and we have just I have just
listed or shown three contractors here.
So, one of the most investigated reaction in micro channels has been Halogenations and in
particular Fluorination. So, this is for example, Fluorination of toluene because this has
wide applications in a number of industries, pharmaceuticals, dyes, liquid crystals, crop
protection agents.
Actually every third drug contains fluorine moiety. The general in general the synthesis
route of these fluorine compounds are complex synthetic root there is no direct synthesis
route and these results in the selectivity of these routes is low and the amount of material
that is wasted is quite high.
So, the direct route it might be cheaper, but it requires precise temperature control because
reactions are fast and exothermic and if the overheating or the hot spot formation happens,
then, the undesired reaction may take place and one will again get poor selectivity, so as
these fluorine as we know that is a very reactive compound. So, the fluorination reactions
are very fast and so, they are often limited by the mass transfer between the gas and liquid
phases.
476
So, one will need high interfacial area density which is what we can easily get in micro
channels. So, it becomes a good candidate to perform gas liquid reactions in micro
channels. One also need because it is exothermic reaction. So, one need to have good heat
transfer and try to maintain the isothermal conditions in the reactor which is very difficult
in a large reactor. But it or a stirred tank reactor for example.
But it can be done in micro channels. One also need to have precise control over the things,
precise control over the residence time, precise control over the pressure and temperature
over which the reactor will be operating because the residence time need to be precise.
Because if it is too long; then, secondary reaction may take place and the selectivity will
again be poor.
So, by selectivity I mean the yield of the desired product or divided by the yield of the
undesired product which is only term generally at selectivity in chemical reaction
engineering. Another gas liquid reaction which has been studied it carbon-carbon
multibond multiple bond addition for example, cycloalkane hydrogenation.
So, here is an example of this. This reaction has been performed in trickle bed reactor and
it is well studied. So, actually this reaction has been used as benchmarking for micro
reactors.
477
It is again a fast reaction and exothermic. So, it can be performed in micro reactions. So,
that hot spots do not form and large interfacial area is provided for the reaction to happen.
There are number of other reactions that have been performed in a micro reactor. (Refer
Time: 14:04).
Now, one of the applications of chemical micro processing is in the Pharmaceutical

Industry; where, in general the trend is to have batch reactor. So, what I mean by batch
reactor that the 2 reactants or all the reactants are mixed in a vessel or in the reactor and
then, the reaction happens. After the reaction, the product is taken out and then, again the
vessel is cleaned and it can be used for another reaction or if the same reaction is to be
carried out.
So, same product is to be manufactured. Then, this without cleaning one can have the
reactions. Now, when one does this then it is highly likely that the product quality from
one batch to another batch may differ.
So, there might be a non uniformity in the product. But the use of batch reactors in
pharmaceutical industry is preferred because of the low quality or not low quality, but the
low quantity of the reactants and products that are involved. One do not need very large
scale reactors in pharmaceutical industry and the chemicals that are dealt with they are
generally of high cost.
478
So, they need to be used judiciously to reduce the production cost. Moreover, the same
reactor after cleaning can be used for another reaction. So, this these problems can be
circumvented or can be addressed by using continuous micro reactors. Micro reactors will
of course, offer the advantage of the continuous flow reactors at the larger scale.
But at the same time it will also give one the freedom to perform reactions with low volume
of fluids. So, one can have this and then, one can number up the process by or one can
scale up the process by having number of micro channel or micro reactors in parallel.
So, that the reaction can be scaled up and it can be performed in the micro reactors. So, in
recent years, in last 2 decades at least there have been a drive to for the to persuade the
pharmaceutical industry looking at the options changing from the Batch reactor mode to
the Continuous flow reactor mode at home as well as internationally.
So, for example, there has been about 15 or 20 years ago, there has been a Novartis and
MIT centre for continuous manufacturing where, they looked at different research
problems that are required to be addressed for continuous manufacturing of
pharmaceutical products.
There are some other news items and that there has been investment from Swiss firm Lonza
and DSM Pharma to convert to micro reactor process technology to produce
pharmaceutical chemicals.
Back home, there is a project running at National Chemical Laboratory, a CSIR lab in
Pune where they have the project is called IndusMagic, in which the goal is to develop
novel reactors for process intensification and one of the reactors that they are looking at
actively is Micro reactors for speciality and fine and speciality chemicals. Mostly, these
chemicals are used in pharmaceutical industry.
479
So, one of the issues that often comes up when one is looking at using micro reactors and
chemical processing that how to scale up. So, as I have discussed in quite a few times
during the course that this is scaling up is achieved by numbering up.
So, one example is Printed Circuit Heat Exchangers which were developed at the
University of Sydney in 1980s and then, it was commercialized by Heatric which is now
a group of a part of a Meggitt aerospace company.
So, in these micro reactors on the metal plates the flow, Flow channels are chemically
etched, as you can see here that on a metal plate the channels are etched and the these
channels can be straight channel or channels like this or they can be tortuous passage.
So, as to enhance the mixing and then these plates are diffusion bonded together and one
they will look like this and then form with the inlet and outlet manifolds one can create a
reactor of this form which will be very compact reactor. Now, this the technology which
they use has been used for developing heat exchangers and it is commercialized
successfully.
In recent decade they have been looking at and Heatric as well as other companies and
researchers have been looking at using the same concept for a scaling up the micro
reactors. So, this is because in this we are not changing the scale of the reactor rather than
480
the number of the reactors that are being used for the application is being increased. So,
for example, in this one can have thousands of micro reactors are running parallel.
So, one of the problems or one of the challenges in addressing this or in numbering up is
that the uniform distribution of the reactant in the channels because when one have the
fluid coming from the manifolds, it is very difficult to distribute the liquid or gas equally
in the channels and this becomes even more challenging when one is dealing with multi
phase flow systems. So, this is an active area of research that how one can have the equal
distribution or the control distribution of the fluid in number of reactors like this.
So, here are 2 sources of information for the students who want to further know about the
Chemical Micro processing in. So, so the chemical engineering or reactions in micro
reactors one can look at these 2 volumes by professor hassle Professor Schouten and
Rankin Sheila and others, Professor Kohl, Professor Lowe.
So, they are all who have been working on the micro reactor technology in its
commercialization for last few decades and this is these books having the several volumes
are a good source of information about the Fundamentals of Transport processes in micro
reactors, Computational and experimental techniques as well as the different Experimental
Protocols for different reactions and applications in for reactions in micro channels.
481
Now, let us look at some of the biological applications by no means the application that I
Show. Cover all the applications, but I am just showing few applications; one of this is
DNA analysis. So, DNA analysis involves a number of processes and it will be useful for
medical diagnostics as well as for point of use testing in agricultural.
So, here is a integrated analysis system in which all the steps of the DNA analysis, sample
loading, drop metering, thermal reaction, gel loading and electrophoresis which is used for
separation of the part of exists particles. So, this all have been integrated on a single chip
and this such a compact and small scale device can be very useful for analysis of DNA
and in these applications.
482
Then, another application is encapsulation of the Cells which is required for say example
drug delivery for treatment of hormone or protein deficient diseases or cancer therapy for
micro carriers or scaffold for cell cultures and tissue engineering.
So, this is a system for developing the 300 micron size the 300 micron size bits which can
encapsulate cells. Then, there is another application.
In understanding the Hemorheology. So, the Hemorheology or the Hemodynamics refers

to the study of flow of blood and Rheology as you might know that Rheology is the science
483
of looking at the relationship between stress and rate of strain or the viscosity of the
complex fluids.
So, as we know that blood is a suspension of red blood cells, white blood cells and platelets
in plasma. Now So, it is complex fluid consisting of plasma which have properties of
water, but it also have some proteins associated with it, so which makes it slightly complex.
But most of the time it can have the density and viscosity that of water. It has slightly visco
elastic nature as the recent research has shown. Now, when there are small deformable
particles most of them are red blood cells.
So, the suspended particles in the blood the 99 percent of the suspended particles are red
blood cells and in the total volume of the blood the volume of these red blood cells will be
45 around 45 percent in a healthy human being.
And this blood which flows in the vascular network which is the network of arteries
capillaries and veins in the body. Now, the size of the capillaries in the smaller arteries
which are called arterioles also and smaller veins which are called venules, it can be as
small as few microns about 8 to 10 microns.
The size of a red blood cells which is a disc shaped cell; it is about 8 micron in diameter
of the disk and 2 micron in thickness. So, as one moves in the smaller veins or the smaller
arteries or the capillaries, the size of the blood cells and the capillaries are almost same.
And the capillaries when these blood cells are moving through the capillaries, they need
to deform and flow through one by one.
Even in larger arteries where the diameter is 300 micron or smaller, as you can see here
there is an interesting effect that is observed in capillaries probably a micro channel effect
this effect is known as Fahreus which suggests that the viscosity of the blood; this is
relative viscosity of the blood is it decreases when the channel size decrease below 300
mm. So, understanding the processes that is happening at these smaller sizes is very
important for a number of biomedical applications.
For example, the techniques are being developed to estimate the viscosity of blood inside
the blood vessels and identify or diagnose the diseases from those for such application one
484
need to have very detailed understanding from the fluid mechanic as well as other transport
processes that is happening in these micro devices or in these micro channels.
So, one need to study first these things in the in the laboratory environment and gain
confidence over it for example, recently there has been a nano machine developed at Indian
Institute of Science in Bangalore; where, they plan to measure the viscosity of the blood
using a helical coil or nano helical coil using the magnetic field. Another interesting
phenomena that happens in micro or at low shear rates in blood that the red blood cells
they stack together.
So, as you can see here that at low shear rates which will often be the case in at low
velocities and low channel sizes because the velocities are very high when the blood comes
out from the heart say in our tiny large arteries. When it goes to capillaries the viscose the
velocity of the blood will be smaller.
So, the shear rates will be smaller. and these red blood cells they tend to aggregate and this
formation is called Roleaux Formation and when this formation happens? The viscosity of
blood is known to be or (Refer Time: 31:27) is known to be a newtonian fluid above a
certain shear rate. But at lower shear rate, it is it behaves as a complex fluid.
So, it shows the shear thinning behaviour. So, all these investigation of the flow of a
foreign material in the blood or flow of red blood cells their movement or the movement
of sick and healthy red blood cells; all this needs to be investigated in Microfluidic devices.
So, of course, there is lot of scope for micro fluidics in Hemodynamics and Hemorheology.
485
Similarly, it is also the language as we know that where the exchange of oxygen happens
in at a membrane exchanges of oxygen and carbon dioxide carbon dioxide happens
between the air and blood and this happens in smaller spherical kind of cells which is
which are called Alveolis.
So, there have been devices which are developed to understand the mass transfer processes
that happens at the lung and then, these devices had been known named as lung on a chip
devices another application which we just saw an example in a for DNA analysis is called
micro total analytical system.
So, in analytical chemistry one need to go through a different stages sample preparation,
liquid handling, detection and the reading out the electronics. It will be wonderful if one
can combine and integrate all these steps in a single platform and that is what these micro
total analytical systems are supposed to provide. So, these are also known as micro total
and system work mu TAS or lab-on-a-chip devices.
So, you have all the functions that you need to do in a chemistry lab integrated over a chip.
So, that is why the name and miniaturized analysis system and they have of course,
applications anywhere where you want to analyze chemically or biologically a particular
sample.
486
So, in medicine, drug discovery, forensic science, biowarfare defense, environmental
monitoring, oil and gas industry and so on and so forth. So, this is another example of
integrated Microfluidic cell culture analysis on a chip. We have already seen an example
for DNA.
So, a related application of this micro total analysis is has come from Professor
Gershenfeld lab in at MIT, where they have suggested that in micro fluidics one can use
bubble logic.
So, one can encode on chip process controlled using bubbles for Microfluidic devices;
where, the presence of bubble is 1 and the absence of bubble and 0. So, this bubble is a
single bit of information using this 1 and 0 and this bit of information it can also carry a
material or a chemical in these bubbles or droplets. So, the computation that happens in
these chips can also result in physical material processing at the same time.
So, one do not need electronics separately as they suggest that the micro fluidic devices
can carry their own process control information without the need for an external
microprocessor. So, that will be this is just a conceptualization stage, but if it can be done
and commercialized and this will be a wonderful thing having a lab-on-a-chip device
which has its own logic.
487
So, these are some of the applications that we have discussed and as I said at the start of
this lecture, there are a number of applications in a number of works of in our day to day
life as well as industry Biomedical applications, Aerospace applications, Automotive
applications, Cosmic industry applications, Pharmaceutical industry applications where
these micro fluidics are required and as micro fluidics depends on the control of the process
one need to have precise control of the systems. Most of the systems that we have discussed
are developed in micro channels, because they offer the advantage of precise control over
the processes.
So, whether one needs the precise control, one need to understand the processes that are
happening and one need to know the fluid mechanics, heat transfer, mass transfer reactions
that are happening in these systems. One also need to because most of the fluids that we
handle are multiphase fluids. So, one need to have a very good understanding of the
Multiphase Microfluidics that happen in these systems.
Thank you.
488
Lecture – 24
Concluding Remarks
So with this we are now, at the end of this 20 hours course titled as Multiphase
Microfluidics.
I have listed here the topics that we have discussed. We started with the Introduction to
Microfluidics. What are the physical effects or what is different or what can be different
in micro channels as well as the basics of Microfluidics and Multiphase flows?
So, in Microfluidics because the interfaces are generally large and capillary effects are
important. So, more than one need to understand the interfaces capillary effects in to
understand the Microfluidic at the Multi-phase flow behavior in micro channels. The
second part of this the interfaces in capillary effects; we have divided into 2 different
sections which were discussed at 2 different time times during the course or at 2 different
stages during the course.
The first stage, where we talked about the Surface tension and Gas-liquid or Liquid-liquid
interactions; we discussed at the start of the course. Once, we have developed a bit of
489
understanding in the next stage we discussed the Three-phase contact angles where we
also looked at the droplets sitting over a sphere and a contact angles and things like that.
So, once we have understood the basics of Multi-phase flows or the fundamentals, the
definitions, the equations n-Laplace law, Cassie Baxter law and the terms that we use in
this area.
Then, we looked at the flow regimes that can happen in Gas-liquid flow and Liquid-liquid
flows and it turned out that one of the most often and most important flow regime in terms
of the applications is the Taylor flow regime. So, we have devoted a large chunk of this
course to the hydrodynamics heat and mass transfer in the Taylor flow regime. We also
looked at that then because one need to precisely control the size of the bubbles and
droplets that are being generated.
So, we have looked at the we have in a brief lecture, we have looked at the bubble and
droplet generation that are and the mechanism of it and equations, the mechanism of
bubble generation at a T junction and in a flow focusing device has been discussed in this.
One while dealing with the Gas-liquid flows void fraction is a very important information
and designing any Microfluidics device, Microfluidic device one need to have an idea of
the pressure drop because that will determine the energy that is required for the pumping
in the fluid.
So, we have in a in a lecture, we have also discussed the void fraction and pressure drop
in Gas-liquid flows in micro channels. We have briefly touched upon the annular flow and
looked at the analytical solution for an ideal annular flow and a how we can calculate the
interfacial radius or the thickness of the liquid film and the pressure drop in ideal
conditions in annular flow. However, the conditions are not always ideal or almost never
ideal, when especially when we are dealing with Gas-liquid flows and that will be
instabilities at the interface.
So, one need to look at further the Hydrodynamics of a slug annular and annular flow and
this is an area which has not been explored fully in a Gas-liquid flow in micro channels.
Then, we have also looked at flow boiling in micro channels. So, the evaporation and the
nucleate boiling and the questions that are there, the to address the that which boiling it is
it the nucleate boiling or the evaporator boiling which is dominant is a still not addressed
490
and one need to further investigate to establish the dominant mechanism of flow boiling
in micro channels.
Taking a divergence from the entire theme of this course we have also looked at the inertial
micro fluidics where the main discussion was about inertial focusing. The focusing of the
particles in a channel and we looked at the rigid particle focusing, but one can also look at
the focusing of a the flexible particles, a cells, bubbles and droplets in the channels and
they it has as we have seen that it has a number of applications in Microfluidics. Then,
during the middle of the course, we have introduced the computational techniques for
modeling Multiphase flows and it because the interface size is large in Multiphase
Microfluidics.
So, one often prefers to capture the interface. So, the interface capturing techniques such
as volume of fluid method or level set method or phase field method or lattice Boltzmann
method are more popular in a Microfluidic application rather than say Euler-Euler method
where the interface cannot be captured and one need to rely on the closing or the closure
equations for the momentum exchange between the 2 fluids or the gas and a solid or gas
and liquid phases.
Then, we have a briefly discussed the different flow measurement techniques and in one
lecture we also looked at the basics and fundamentals of micro particle image velocimetry
491
which can be used to understand the velocity field and with laser induced fluorence;
fluorescence the temperature and concentration field in a micro channels.
And in the last 2 lectures, we have looked at the number of applications and the potential
of the micro fluidics, to impact the different sectors and our day to day life. So, I hope that
you have learned some of the basics. And this course has a motivated you to learn more
about Microfluidics.
We welcome the suggestions to further improve the course. I would also request you to
post any questions or doubts that you might have or any suggestions that might have in
that you might have about the course to post on the course website.
Thank you and good luck.
492
THIS BOOK IS NOT FOR SALE
NOR COMMERCIAL USE
(044) 2257 5905/08 nptel.ac.in swayam.gov.in

Microfluidics Lecture Index

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Microfluidics Lecture Index

Uploaded by

Copyright:

Available Formats

INDEX

S. No Topic Page No.

(Refer Slide Time: 07:55)

So, if we look at the continuum hypothesis. It is often described or determined by a non-

(Refer Slide Time: 14:55)

(Refer Slide Time: 18:15)

(Refer Slide Time: 20:56)

100 in conventional channels it can go up to 103 or more in microchannels. So, it will be

So, Reynolds number is generally defined as

stress is generally represented as 𝜌𝑈 2. So, the ratio of ρ U square and μ U by d is the

and buoyancy force is

So, ρ so, that will be

And surface stress or surface tension generated a stress is

So, this will be

(Refer Slide Time: 39:44)

(Refer Slide Time: 41:50)

(Refer Slide Time: 43:07)

(Refer Slide Time: 01:05)

Now, if we look at in the framework of continuum mechanics, or under the assumption of

(Refer Slide Time: 02:55)

So, that is a molecular or microscopic view of the interface.

(Refer Slide Time: 08:09)

Then the interest the energy loss is

(Refer Slide Time: 10:19).

(Refer Slide Time: 11:37)

(Refer Slide Time: 13:47)

(Refer Slide Time: 16:08)

(Refer Slide Time: 18:56)

(Refer Slide Time: 21:18).

(Refer Slide Time: 23:49)

if it represents a curve. Then the curvature is defined as

So, this is the divergence of the gradient of f.

And in many cases it can be approximated by d 2 y by dx 2 ok.

(Refer Slide Time: 25:33)

(Refer Slide Time: 28:10)

for a spherical bubble or droplet.

𝜕𝑊𝑒𝑥𝑝 = (𝑝1 − 𝑝0 )𝑑𝑉

(𝑝1 − 𝑝0 )𝑑𝑉 + 𝜎𝑑𝐴 = 0

(Refer Slide Time: 34:46)

(Refer Slide Time: 36:29)

So, in case of partial wetting 𝜎𝑆𝐺 is less than 𝜎𝑆𝐿 +𝜎𝐿𝐺 .

(Refer Slide Time: 01:08)

(Refer Slide Time: 10:36)

(Refer Slide Time: 12:04)

(Refer Slide Time: 16:18)

(Refer Slide Time: 20:32)

(Refer Slide Time: 24:51)

(Refer Slide Time: 33:41)

(Refer Slide Time: 36:29)

(Refer Slide Time: 38:26)

(Refer Slide Time: 44:23)

(Refer Slide Time: 58:19)

(Refer Slide Time: 60:39)

(Refer Slide Time: 00:42)

(Refer Slide Time: 04:51)

(Refer Slide Time: 19:28)

(Refer Slide Time: 21:15)

(Refer Slide Time: 23:01)

(Refer Slide Time: 00:42)

(Refer Slide Time: 03:08).

(Refer Slide Time: 07:08)

(Refer Slide Time: 12:32)

𝑉𝐺,𝐶𝑉 + 𝑉𝐿,𝐶𝑉 = 𝑉𝐶𝑉 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡