103106158

INDEX
S. No Topic Page No.

Week 1
1 Introduction 1
2 Fluid Stattics 6
3 Newton Law of Viscosity 13
4 Equation of Continuity Differential 19
5 Equation of Linear Momentum - 1 34
6 Equation of Linear Momentum - 2 43
Week 2
7 Bernoulli's Equation 52
8 Solution of Navier Stokes - 1 64
9 Solution of Navier Stokes - 1 74
10 Introduction to cylindrical coordinate systems 82
11 Continuity equation in cylindrical coordinates. 90
12 Solution of Navier Stokes in the Cylindrical co-ordinate system - 1 97
13 Solution of Navier Stokes in the Cylindrical co-ordinate system - 2 103
Week 3
14 circular poiseuille flow 110
15 Shear Stress Distribution 114
16 flow between two concentric cylinder. 122
17 taylor couette flow 130
18 Viscosity and Momentum Transfer 140
19 Device For Measuring Fluid Viscosity 157
20 Fluid Properties And its Behaviour 174
21 Tutorial 4 185
Week 4
22 Choice of Scaling Parameter 191
23 Non Dimensional analysis 200
24 Non-dimensional analysis-2 208
25 Non-dimensional analysis-3 - Buckingham Pi Theorem 227
26 Non-dimensional analysis-4 - Trinity test 244
Week 5
27 Non-dimensional analysis-5-Concept of similarity 255
28 Characterization Of Particles 1 272
29 Characterization Of Particles 2 296
30 Motion of a Particle in a fluid 317
31 Brownian motion and electophoresis 328
Week 6
32 Sedimentation and Seperation 346
33 Settling velocity - Stoke's regime and Newton's regime 363
34 Applications of settling - I 381
35 Applications of settling - II 392
36 Colloidal aggregates - Introduction 401
37 Settling of colloidal aggregates 414
38 Tutorial 5 424
Week 7
39 Settling of colloidal aggregates - free settling 430
40 Settling in Multiple Particles System 445
41 Flow Through Packed Bed 468
42 Pressure Drop Through Packed Bed 493
43 Tutorial 6 508
Week 8
44 Pressure Droped Through Packed bed Continue 511
45 Fluidized Bed - 1 528
46 Fluidized Bed - 2 542
47 Filtration - 1 562
48 Filtration - 2 580
49 tutorial 7 610
Week 9
50 Laminar and Turbulent Flows - 1 621
53 Turbulent Stress and Turbulent Shear Layer 647
54 Turbulent Flow near a wall and in a pipe 659
Week 10
55 Effect of rough Walls 672
56 Roughness in Turbulent Pipe Flow 677
57 Pipes of non-circular cross section 682
58 Minnor Losses, Sudden Expansion and Contraction 689
59 Friction Losses in Sudden Expansion 705
60 Tutorial 8 712
Week 11
61 Momentum and Kinetic Energy Correction Factor 721
62 pressure drop in pipes which connected in series 727
63 Pressure Drop in Pipes Which Connected in Parallel 730
64 Pressure Drop in Pipes Which Connected at Junction 737
65 Boundary Layer 743
Week 12
66 Boundary Layer - Momentum Integral Analysis -1 754
67 Boundary Layer - Momentum Integral Analysis -2 762
68 Boundary Layer - Differential Approach 768
69 Laminar and Turbulent Boundary Layer 779
70 Tutorial 9 797
Fluid Mechanics
Prof. Sumesh P. Thampi
Department of Chemical Engineering
Indian Institute of Technology, Madras
Lecture – 01
Introduction
So, in the first part of this course you are going to look at Mechanics of Fluids.
(Refer Slide Time: 00:22)
And in the second part, you will look at situations where the fluids come into contact with
particles; there is interaction between particles and fluid and that results into much more
complicated, but complicated scenarios, but that are often encountered in life. So, what do
I mean by fluid mechanics or what are fluids first of all? So, typically we have fluids as
constituted by liquids and gases, as opposed to solids.
So, what is the difference between solids and fluids? So, all of you know solids have a
definite shape, they do not change shape according to the container or the shape of the
container in which it is situated. Or more precisely if you apply a force on a solid the solid
would deform, it will change its shape and it will reach a new shape. But if you are applying
a force on a liquid or a gas; for that matter on a fluid the liquids and gases they will continue
to deform, they will continue to move as long as the force is applied.
1
So, that is the main difference between solids and liquid fluids. So, fluid mechanics there
are two things that we would like to look at; there is fluid statics and then there is fluid
dynamics. So, fluid statics is essentially study of the fluids at rest and dynamics is fluids
that are moving or are in motion. So, these are the two branches of fluid or two divisions
in fluid mechanics. So, where do we encounter fluid or where is this study important? So,
fluid mechanics is or fluids you encounter very often all around you.
So, imagine at the moment you are breathing you are the; you are inhaling the air you are
absorbing the things that are required in your lungs and then the rest is thrown out right.
So, you there itself the fluid mechanics come into action or look at the life ok; your entire
body it is the blood; the blood is moving all around taking things to the required parts of
your body.
So, and you know that blood actually travels through small and narrow pipes which are
distributed throughout the body of all shapes and sizes. So, again you have a fluid that is
flowing throughout your body and then again fluid mechanics actually plays a big role.
So, so just to list the applications for example, breathing, blood flow or any kind of
transportation that you think about.
For example, if its air transportation aero planes or ships or even for a racing car all of
them move in a fluid; it would be either water or air. And whenever these things are
moving in that particular fluid it basically experiences a drag force, a resistance to motion
ok; so that is where the fluid mechanics come into the picture.
Lots of sports whether its cricket football you know tennis or swimming everywhere again
you know for example, in cricket the ball interacts with air around it which decides the
trajectory of a cricket ball. Or when you are swimming you are again moving in a fluid
and then you need to optimize your swimming stroke since that you can get the fastest or
the highest speed.
Ah Then things like for example, wind mill where you know; where you are trying to
extract energy cases like rivers, dams then and so on ok. So, everywhere around if you
look at you see some kind of a fluid and then fluid mechanics become important. So, this
is a first course in which we will try to introduce you to various; what are the aspects that
are important when you look at the motion of the fluids, how do you know what are the
2
forces that are required to move a fluid from one point to another? So, that is the aspect
that we are going to concentrate.
And we are going to look at it from the perspective of continuum, so we are going to treat
fluid as a continuum. So what do I meant by a continuum? Let us say you have got a
microscope and you have some fluid you know air water anything that you like and then
you are going to zoom into the fluid.
You are zoomed in so much that you can now see individual molecules. So, you will see
that there are molecules that constitute a fluid and you have reached that kind of a
resolution in which you can see the molecules. Let us say let me represent some of them,
so let us say we do not know how they look like each of the circles; let us say represent
molecules at a given or the location of the molecules at a given time.
And my intention now is to plot a graph where on x axis, I will plot volume and on y axis
I will plot a density. So, what I am going to do? I am just going to take a small volume, let
us say that that is a volume that I have represented as δ V; a small volume where I can see
actually individual molecules. And then I am look I am going to calculate what is the
density and the way I would have calculated the density; would be I would say what is the
total mass divided by δ V and I will give you my density and I will mark that point
somewhere ok.
3
Let us say that volume is this is the point that I am looking at and that is a point that I have
gotten. Now, what I will do is that I will increase the volume of this let us say this. So,
initially I have chosen a cube which I have marked as red, now I will make it a little bigger
and I will calculate the volume again. Now, you can see that when I do that I have got
maybe I would have included one or two extra molecules.
But the point is that the number of molecules that you include when you are increasing
that volume is not uniform; maybe originally I had 3; now I have increased it into let us
say 4, 5. But that does not mean that your density has gone up proportionately; you have
increased your volume, but you have changed the mass by a little bit.
And therefore, you would see that you get a density which is different maybe something
else ok. And as I keep expanding this you will see that you know there will be lot of
fluctuations in the density that you are calculating and that is because you are actually
defining your volume in a scale similar to the molecular scale.
So, your volume that you have considered which is δ V is very small that when you are
increasing the volume; you will see that you will see that the number of molecules that
come up in each volume may be not sufficient to proportionately increase the density. So,
that is why you will start seeing a lot of fluctuations.
Now, as you increase your volume further you will see that the beyond some point ok; you
have gone let us say up to those kind of length scales where you are not able to see the
molecules further ok, you have you have got huge number of molecules that even when
you are increasing the volume that does not lead to a change in the density.
So, then what you would see is that these fluctuations will become smaller; then it will
become flat. So, that is the kind of density that you are familiar with, that is the density
that we measure because when we measure the density we will take a huge number of
molecules or in other words we will take a macroscopic volume and then we will measure
the density. So, if you measure density in that regime then the density becomes a constant.
And it becomes independent of the volume that you have considered, but let us say this is
this entire thing is part of a huge amount of fluid.
Let us say this is the fluid that you are considering and let us say temperature is changing
from one point to another. So, at the moment we had taken a small amount and that is what
4
we zoomed in and then looked at this aspect. But on the other hand if I continue to look
at, so that is the amount that we started with, if I go from one end to the other end. And let
us say temperature is increasing in the right direction then what you are seeing is that the
density should also change.
In other words, if I continue to increase my δ V, at some point I should see that the
temperature will come into the picture. There is a density change and that will result in to
and that will result into something of that sort ok. Whether they are the dense whether
where you will actually start seeing changes in the density because of the other effects ok.
So, this region where you start seeing the density changes is because of the macroscopic
variations. While in this region the density fluctuation is because of the because you are at
a length scale which are much comparable to the molecular length scale. So, this is a
molecular picture; while on the right hand side this is where you start seeing density
variations because let us say pressure changes, temperature changes and so on. In between
there is a region where you will see density being a constant and that is the region you
consider as in your continuum approach.
So, this region will constitute as a point in your continuum ok. In other words what I meant
to say is that if you have a fluid and if you say that you are looking at a particular point in
a fluid that point is really not of 0 size that point basically is of this length such that at that
point I can define a property to be constant. If I increase the volume of that point, then I
am going to see macroscopic variations, but if I decrease the volume of that point, then I
am going to see molecules.
So, the basic idea of continuum is that the point is defined on a length scale much larger
than the molecular length scale, but much smaller than the macroscopic variations. So, that
is the idea of treating fluid as a continuum. So, any quantity that you look at essentially is
going to be an average or any point any property that you assign to a particular point, that
point will have enough number of molecules that you will not see any fluctuations because
of the fact that it is constituted by discrete entities.
So, that is the definition of a continuum and we will treat fluid as a continuum. So, for us
then you know every property will be a well defined quantity. And we are going to be
staying away from the length scales of molecules at a sufficiently you know and we will
stay at length scales much larger than the molecular length scale ok.
5
Fluid Mechanics
Lecture – 02
Fluid statics
Now, let us talk about pressure. So, pressure is a special kind of stress, its also defined as
force per unit area.
So, let us look at pressure force on a fluid element. Let us take a coordinate system. So, let
us say this is x that is y and that is z and let us take a fluid element. So, this has got a length
d x in x direction, d y in y direction and d z in z direction. So, that the volume of this fluid
element:
𝑉𝑜𝑙𝑢𝑚𝑒 = 𝑑𝑥 𝑑𝑦 𝑑𝑧
Now, let us say this fluid element is in a given fluid where there is a pressure distribution,
so; that means, the pressure is changing from one point to another and let us say pressure
is given as a function of x, y, z and t.
𝑃 = 𝑃(𝑥, 𝑦, 𝑧, 𝑡)
6
So, pressure is changing at every point as well as in time ok, at any given point x, y, z and
t you know what is the pressure and what you are going to do is you are going to now
calculate what is the pressure exerted on each face of the fluid. So, let us say we are going
to look at this face. So, the face that I have marked with red color and I want to calculate
what is the force on that side. So, I know that on this side of this face, I can calculate the
force because pressure is defined as force per unit area therefore, if I want to talk about a
force which is acting in the y direction on the left side of the play on the left side of the
cube is simply going to be. So, let us let me write left is simply going to be:
𝑙𝑒𝑓𝑡
𝐹𝑦 = 𝑃𝑑𝑥 𝑑𝑧
So, that is the force that is going to be acting on this side of the element. Similarly, you
can calculate what is the force on the other side. On this side this exactly opposite side of
the red side that I have marked. Now, we do not know what is the pressure there, but we
can let us say do a Taylor expansion and write down the pressure there.
So, let us say if the pressure here is given by P, then the pressure here can be written as:
𝜕𝑃
𝑃+ 𝑑𝑦
𝜕𝑦
So, that is a simplest approximation that you can do to represent pressure on that side. So,
therefore, F y on the right side is given as:
𝑟𝑖𝑔ℎ𝑡 𝜕𝑃
𝐹𝑦 = [𝑃 + 𝑑𝑦] 𝑑𝑥 𝑑𝑧
𝜕𝑦
And your interest is to calculate what is the net force acting in the y direction, which you
will get if you subtract one from the other. So, if your interest is to calculate F y net which
is the force acting in the y direction is given as :
𝜕𝑃 𝜕𝑃
𝐹𝑦𝑛𝑒𝑡 = 𝑃𝑑𝑥 𝑑𝑧 − [𝑃 + 𝑑𝑦] 𝑑𝑥 𝑑𝑧 = − 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑦 𝜕𝑦
So, that is the force that is arising in the y direction due to a pressure. We can continue,
we can do the same thing in x direction as well as in z direction.
7
So, we can do the same exercise and then you find that:
𝜕𝑃
𝐹𝑥 = − 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥
𝜕𝑃
𝐹𝑦 = − 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑦
𝜕𝑃
𝐹𝑧 = − 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑧
So, in the given coordinate system these are the forces that are generated, I want to write
force as a vector. So, this is the component of force in x direction the second 𝐹𝑦 is the
component of force in y direction and 𝐹𝑧 is the component of force in z direction.
So, I can write the net force:
𝜕𝑃 𝜕𝑃 𝜕𝑃
𝐹⃗ = 𝐹𝑥 𝑥̂ + 𝐹𝑦 𝑦̂ + 𝐹𝑧 𝑧̂ = [− 𝑥̂ − 𝑦̂ − 𝑧̂ ] 𝑑𝑥 𝑑𝑦 𝑑𝑧 = −∇𝑃 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
So, what is grad? Grad is nothing, but the gradient operator gradient operator that is the
grad. So, that when grad is acting on some quantity that is nothing, but or we will use a
different one let us say ∇ phi; that means:
8
𝜕𝜙 𝜕𝜙 𝜕𝜙
∇𝜙 = 𝑥̂ + 𝑦̂ + 𝑧̂
𝜕𝑥 𝜕𝑦 𝜕𝑧
So, that is the definition of the gradient operator. So, we get:
𝐹⃗
= 𝑓⃗ = −∇𝑃
𝑑𝑥 𝑑𝑦 𝑑𝑧
So, what is this f now? This f is force per unit volume ok. So, in other words what we have
said so far is that pressure gradient differences in pressure leads to a force or more precisely
gradient on pressure gives rise to a force density. So, force per unit volume or force density.
You have shown now that gradient in pressure gives rise to a force density on the fluid
element ok.
So,
𝑓⃗ = −∇𝑃
𝜕𝑃 𝜕𝑃 𝜕𝑃
𝑓𝑥 𝑥̂ + 𝑓𝑦 𝑦̂ + 𝑓𝑧 𝑧̂ = − 𝑥̂ − 𝑦̂ − 𝑧̂
𝜕𝑥 𝜕𝑦 𝜕𝑧
Let us consider a fluid element under gravity and it is static. So, the fluid is not moving
the fluid element is static, the only force then that is acting on the fluid element will come
from gravity ok. So, in which case you can say that. So, we will use the same coordinate
9
system, so that is your x, that is your y, that is your z and gravity is acting in the downward
direction. So, there is no force that is going to act horizontally. So,
𝑓𝑥 = 0, 𝑓𝑦 = 0, 𝑓𝑧 = −𝜌𝑔
So, you know that, so gravitational force is nothing, but weight ok. So, the weight is given
by mass times g, but here we are going to write mass. So, weight as mass times g or it can
it can be also written as volume times density times gravity.
We are talking about force per unit volume or in other word or force per volume and that
is just going to be ρ g and I have given the minus sign because the minus sign tells you
that its actually acting in the minus z direction, because this is the diagram that you see on
the right that is the coordinate system that we have chosen. Therefore,
𝜕𝑃
𝑓𝑥 = 0 → = 0 → 𝑃 ≠ 𝑃(𝑥)
𝜕𝑥
𝜕𝑃
𝑓𝑦 = 0 → = 0 → 𝑃 ≠ 𝑃(𝑦)
𝜕𝑦
Therefore, P is only a function of z. This the equation of hydrostatic pressure.
𝜕𝑃
− = −𝜌𝑔
𝜕𝑧
𝑃 = 𝜌𝑔𝑧 + 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
10
Let us see where we can use this, let us look at application of a manometer. So, where do
we use manometer? Manometer is used to measure pressure differences between various
points. You know let us consider simple case where you have a pipe and let us say a fluid
is flowing in the pipe and for the fluid to go from the left side to the right side you need to
apply a pressure difference. So, let us say you have applied a pressure P1 here and let us
say the pressure on the other side is P2 and obviously, P1 has to be greater than equal to
P2 for the fluid to move from left to right and some point you wanted to calculate what is
the pressure difference between these two points.
So, the simplest way to do it is that you can connect a manometer. So, let me draw a
manometer, I will draw a U-tube manometer it is a big manometer and manometer has a
manometric fluid which let us say is as shown by the shades of that ok. So, the fluid inside
the pipe has a density of 𝜌1, let us say the density of the fluid of the manometric fluid is
ρm and there will be a height difference between two limbs in the manometer let us call
that h.
Now, from our previous analysis we have found that, so this manometric fluid right now
is a static fluid, even though there is a fluid that is flowing in the pipe ok. The manometric
fluid is a static fluid and therefore, the calculations that we just did is applicable for the
manometric fluid. So, let us use that fact and we found that the pressure does not vary in,
so we will choose the same coordinate system, so x y z ok. So, pressure can vary only in
11
in z direction because that is the only direction in which gravity is acting or weight is there
to balance your pressure forces, pressure does not change in x or y direction.
So, if I mark let us say this point as A and a similar point here at the same elevation as B,
then pressure cannot change between A and B. So that means,
𝑃𝐴 = 𝑃𝐵
But I also know that now I can write the pressure A as pressure at this point which is P 1
plus the pressure that is going to come from this height let us call that height h1. So,
𝑃1 + 𝜌1 𝑔ℎ1 = 𝑃2 + 𝜌1 𝑔ℎ2 + 𝜌𝑚 𝑔ℎ
So, this is the pressure force balance or rather the pressure balance at point A and at point
B and our interest is to calculate what is P1 - P2. So, I will write
𝑃1 − 𝑃2 = 𝜌𝑚 𝑔ℎ + 𝜌1 𝑔(ℎ2 − ℎ1 )
Now, if you look:
ℎ + ℎ2 = ℎ1 → ℎ2 − ℎ1 = −ℎ
So, I will use that as:
𝑃1 − 𝑃2 = 𝜌𝑚 𝑔ℎ − 𝜌1 𝑔ℎ = (𝜌𝑚 − 𝜌1 )𝑔ℎ
Δ𝑃 = Δ𝜌𝑔ℎ
So, pressure difference between two points is given by the difference in density times
gravity times the height and this is the height that we are interested in. And that tells you
how you can use a manometer to measure pressure difference between two points.
In other words the pressure difference just got translated into a height difference and height
difference is something that you can measure and from the height difference, you can go
back and calculate what is the pressure difference. So, this is a nice application of the
equation of the hydrostatics that we derived just now.
12
Fluid and Particle Mechanics
Lecture – 03
Newton’s law of viscosity
So, when you talk about problems in fluid mechanics, one of the property that comes up
often is the viscosity of the fluid.
So, let us just briefly look at what is viscosity? So, viscosity actually refers to the resistance
of the fluid when it is trying to flow. So, the concept is very simple, let us say you have
some solid object; let us say you have a solid object and that is placed on a another solid
substance. And let us say you are trying to move this box in a particular direction and what
happens is it because the and the object on which it is kept is not moving, its going to x
you know its going to resist the motion of the this the solid object that you have placed. In
other words this box is going to be experiencing a frictional force that is generated between
the box and the underlying substrate.
Similarly, if you have some wall let us say and then there is fluid on top of it, if the fluid
on top of this wall is trying to move the wall is going to offer some resistance it’s going to
generate some frictional resistance and viscosity exactly characterizes that resistance.
Now, the point is that you can imagine that the fluid that is above the wall is constituted
13
by various layers of the fluid, let us say the way I have drawn this lines ok. So, this is a
layer 1, this is layer 2, layer 3, layer 4, layer 5 and so, on.
So, the wall is going to give some resistance to the layer, the layer that we we have
numbered as 1 the layer 1 will offer a resistance to layer 2, the layer 2 will resist offer a
resistance to layer 3 and it keeps going ok. So, the viscosity of the fluid acts between any
2 layers of fluid that you are going to see or you will have it ok. So, the viscosity is nothing,
but the resistance that comes about when 2 fluid elements are moving with 2 different
velocities. If the fluid elements are moving with the same velocity then viscosity has no
role to play.
So, viscosity is nothing, but really the viscous or the resistance for the flow for the fluid
motion ok. So, that is a resistance to fluid motion or a measure of resistance to fluid motion.
So, how does one calculate this ok? So, that is where Newton’s law of viscosity comes
into picture, Newton’s law of viscosity it says that you take fluid between 2 plates. So, you
have a plate and another plate and then there is fluid in between as shown by the dashed
lines and let us say you apply a force F on the top plate ok.
So, what will happen is that the top plate will start moving, which means the fluid just
below it will start moving and in the fluid below it would start moving and finally, the
entire fluid would be moving except the fluid that is at the bottom of the plate. So, if you
look at the velocity profile meaning that the velocity at various locations it would look
something like this.
So, the top plate would be moving with a velocity almost with the velocity as that of the
top plate, then the next layer would move with a slower velocity, the next layer will be
even slower and that keeps going till the last layer is not moving. So, this is how the
velocity profile is going to look like. Let us say to get that you have applied a force F, then
you can define a shear stress as the force that you have applied per unit area where A is
the area of this plate ok.
𝐹
𝑆ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑒𝑠𝑠 =
𝐴
So, A is the area, so force per unit area tells you the shear stress. The strain rate is defined
as let us say this velocity at the top is u and the plates are separated by a distance h, then:
14
𝑈
𝑆𝑡𝑟𝑎𝑖𝑛 𝑟𝑎𝑡𝑒 =
ℎ
And from Newton’s law of viscosity:
𝐹 𝜇𝑈
𝑆ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑒𝑠𝑠, 𝜏 = =
𝐴 ℎ
𝜇 → 𝑑𝑦𝑛𝑎𝑚𝑖𝑐 𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦
So, that is what Newton’s law of viscosity, so that is how viscosity is typically measured
also.
Now, so what does? So, shear stress is nothing, but the force that you have applied right.
So, it is a measure of force per unit area what the strain rate indicate? One another name
for strain rate is that you know it’s changes in velocity per unit length right.
So, in that sense this is also a measure of velocity gradient. So, gradient typically refers to
things how much it is changing in some direction ok, so this is basically change in the
velocity. So, what does it mean? So, in order, to say that let us just take a fluid element,
let us say a small fluid element that we take from here. So, the one which I have shaded is
the fluid element that I have taken and I am going to make a slightly bigger drawing of
that.
15
Let us say that is the fluid element that we have considered, let us call that axis x, let us
call that axis y let us say this is small fluid element and elemental volume. So, this distance
horizontally is δ x and δ y is the vertical or the height ok. Now, what we want to do is that
we want to find out how what is the effect when there is a fluid, well actually what is the
effect or what is what is going to happen to this fluid element.
So, now imagine this is the fluid element the shaded one is what we have considered and
the velocity at the bottom of this layer is smaller than the velocity at the top of this element.
So, to represent that let us say that the velocity at the bottom of the layer is u is equal to 0
and let us say u is equal to δ y is the velocity at the top or otherwise you could say that u
is the velocity at the bottom and u plus δ sorry δ u u plus δ u is the velocity at the top ok.
Or the whole point is that the to the top layer and the bottom layer are moving with different
velocities.
𝐵𝑜𝑡𝑡𝑜𝑚 𝑙𝑎𝑦𝑒𝑟 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦, 𝑢 = 0
𝑇𝑜𝑝 𝑙𝑎𝑦𝑒𝑟 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦, 𝑢 = 𝛿𝑢
And let us say you are looking at things after a time t. So, at time t is equal to 0 this square
is the configuration and you want to find out what is the configuration after time t equal
after time t. So, now, we have assumed that the bottom line having a velocity u is equal to
0, so; that means that if I look at this corner after time t it’s going to be there. If I look at
16
this corner that corner is again going to be at the same place after time t. But if I look at
that this top corner that is going to move, that is going to move with a velocity u and it will
cover a distance u times t in time t.
So, this the point that I have marked as the red dot would be somewhere else after some
time ok. So, after some time this the point that is marked as the red dot would have moved
to a new point. Similarly let us say the point that I have marked as the cross would move
to a different point because the top layer is moving. In other words the shape would look
like that. So, the square does not remain the square anymore, the square would have
changed to a deformed shape.
So, this is what is going to happen in a small time. So, let us say the time that we are talking
about is really a small time δ t ok. What is the distance that the top layer would have
moved? So, since it has got to a velocity of u and then we are looking at a time of δ t. So,
this would have moved a distance δ u times δ t velocity times the time gives you the
distance that it has moved. So, this is the configuration or this is what happens to the fluid
volume. Let us calculate what is this angle let us call it δ θ.
So, I can calculate tan δθ as:
𝛿𝑢 𝛿𝑡
tan 𝛿𝜃 =
𝛿𝑦
17
And I am looking at δ θ being really small in which case tan δθ can be approximated as δθ
therefore:
𝛿𝑢 𝛿𝑡 𝛿𝜃 𝛿𝑢
𝛿𝜃 = → =
𝛿𝑦 𝛿𝑡 𝛿𝑦
But I will, I am interested in trying to find out when δt→ 0 δy→0. So, we are looking at an
infinitesimally small element which has got a differential velocity and in a small time ok.
So, each of this differences can be approximated as derivatives and therefore:
𝑑𝜃 𝑑𝑢
𝐹𝑜𝑟 𝛿𝑡 → 0, 𝛿𝑦 → 0; =
𝑑𝑡 𝑑𝑦
So, remember for the Newton’s law of viscosity, we defined τ or the shear stress is
proportional to viscosity times the strain rate. The strain rate which is du by dy, so this is
the strain rate right strain rate.
So, the physical meaning of strain rate is that it tells you what is the change in the angle in
unit time. In other words the change in angle, so if d θ by dt was 0 that would mean that
the shape does not change, the shape remains as a square which would happen if there was
no difference in the velocity. So, d θ by dt is telling you how fast the square element is
getting deformed ok.
So, shear strain rate is nothing, but the rate of deformation and Newton’s law of viscosity
simply says that the force that you are applying is proportional to the rate of deformation
and the coefficient that connects the force with the rate of deformation is nothing, but
viscosity. So, that is what Newton’s law of viscosity is that also tells you how you define
your shear stress, what is the physical meaning of strain rate.
18
Fluid Mechanics
Lecture – 04
Equation of continuity- Differential approach
So, to analyze fluid flow ok, we could take two different approaches one is an integral
approach and one is a Differential approach.
And we are going to actually follow the differential approach in this course because it is
going to tell you how the fluid flow is at various points in domain. And it is going to give
you actually the details of how does the flow look like and it is important to know what it
is ok. So, we are going to look at now the differential approach of solving fluid mechanics
problems. So, the first thing that I want to define is the acceleration of a fluid element.
So, let us say you have some arbitrary flow and you can define then a velocity field. So,
say let us say the velocity is some capital V, I am putting that squiggle to show that it is
actually a vector, because velocity is a vector it has a magnitude and a direction. But the
velocity will not be uniform, the velocity will be a function of space it will change with
time. So, in order to represent that I can say that it is a function of x, y, z,t so at every point
in space; it is going to be different:
19
⃗ (𝑥, 𝑦, 𝑧, 𝑡)
𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 = 𝑉
So, that is the velocity field that I have gotten in some domain, but remember we is said to
be a function of x, y, z and time, so; that means, that every point the velocity is changing
and it also going to change with time and my intention is to now calculate what is the
acceleration of the fluid element. Now, you could calculate acceleration in two different
ways.
But let us see first of all velocity ok. So, velocity; so since it is a vector, it has an x
component, it has a y component and an z component. So, let me write it explicitly:
⃗ (𝑥, 𝑦, 𝑧, 𝑡) = 𝑢(𝑥, 𝑦, 𝑧, 𝑡)𝑥̂ + 𝑣(𝑥, 𝑦, 𝑧, 𝑡)𝑦̂ + 𝑤(𝑥, 𝑦, 𝑧, 𝑡)𝑧̂

𝑉
So, this is basically nothing, but the velocity written in a vector form with each component
u v and w are the components of velocity in each directions. And our intention now is to
calculate what is the acceleration associated with this velocity field? There will be an
acceleration because velocity is changing ok; it is changing in space and time; so you will
definitely have an acceleration.
The question is how are you going to calculate the acceleration? You can do the calculation
in two different ways either you can talk about an Eulerian approach or you can talk about
a Lagrangian approach ok. So, the Eulerian approach is that if you are looking at a quantity
20
and you are fixed ok. So, the frame of reference is fixed, you are looking at a particular
point and you are looking at the changes; that is what you will have as an Eulerian field.
As opposed to the Lagrangian field in which you will be actually following a fluid particle.
So, let me give you an example; let us say you know you, you have a you are your
monitoring traffic ok. And let us say that you have got you are looking at various vehicles
which are moving and you are trying to find out what is the velocity with which different
things are moving ok. And you reach some junction; let us say some you know signal not
signal actually at some junction, you will stay there and you will look at what is the
velocity with which each vehicle is going ok.
So, you are looking at a particular point which is actually that junction and you are looking
at the speed of the vehicles. So, that is what is meant by Eulerian approach because you
are looking at a particular point and then trying to find out what is the velocity field ok.
On the other hand, let us say you are looking at a particular vehicle let us say you show
you picked one car and then you are trying to follow that car and trying to find out what is
the velocity of that car? Then it becomes a Lagrangian approach that is because you are
picking up a element a fluid element or in this case it is a car and then you are actually
following the car. So, then it becomes a Lagrangian approach ok.
Now, when I have written V as a function of x, y, z and t; I am actually defining my

Eulerian field because I am looking at a particular point and then I am trying to define
what is the velocity field ok. So, this is the difference between Eulerian and Lagrangian
approach and we would be actually using the Eulerian description because we have a fixed
domain and we are interested in telling how things change at a particular point and in time.
And now I want to define acceleration associated with this velocity field acceleration ok.
So, acceleration is going to be:
𝑎 = 𝑎𝑥 𝑥̂ + 𝑎𝑦 𝑦̂ + 𝑎𝑧 𝑧̂
Now, let us say you were throwing a stone and you are trying to follow the stone which
basically becomes the Lagrangian approach. In which case its very easy to calculate the
acceleration because you will find out how much does the velocity change in time when
you are looking at the stone. So, that is how you will calculate the acceleration.
21
Now here you have to do the same thing you want to look at a fluid element and you want
to find out how much does the velocity change when you are looking at a fluid element.
The only complication is that the fluid element that you are looking at time t would have
reached somewhere else at time t plus ∂ta t. And you are not interested in finding out what
is the velocity or the acceleration at that point you were originally interested in finding out
the acceleration of a fluid element at a particular point.
So, we can define let us say acceleration as the total change in velocity ok; which I am
writing it as dv by dt where v is my velocity field; I am going to differentiate my velocity
with respect to time and going to calculate what is the acceleration.
⃗
𝑑𝑉
𝑎=
𝑑𝑡
⃗
𝑑𝑉 𝑑 𝑑𝑢 𝑑𝑣 𝑑𝑤
𝑎= = (𝑢𝑥̂ + 𝑣𝑦̂ + 𝑤𝑧̂ ) = 𝑥̂ + 𝑦̂ + 𝑧̂
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡
So, therefore, if I want to take a total derivative; I should be really using the chain rule
𝑑𝑢 𝑑 𝜕𝑢 𝜕𝑢 𝑑𝑥 𝜕𝑢 𝑑𝑦 𝜕𝑢 𝑑𝑧
= (𝑢(𝑥, 𝑦, 𝑧, 𝑡)) = + + +
𝑑𝑡 𝑑𝑡 𝜕𝑡 𝜕𝑥 𝑑𝑡 𝜕𝑦 𝑑𝑡 𝜕𝑧 𝑑𝑡
I have just applied the chain rule because u is a function of x, y, z and t. Now,
22
𝑑𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢
𝑎𝑥 = = +𝑢 +𝑣 +𝑤
𝑑𝑡 𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
So, that is a component of acceleration in the x direction. This can be simplified further:
𝜕𝑢
𝑎𝑥 = ⃗ . ∇𝑢
+𝑉
𝜕𝑡
⃗ = 𝑢𝑥̂ + 𝑣𝑦̂ + 𝑤𝑧̂

𝑊ℎ𝑒𝑟𝑒, 𝑉
𝜕𝑢 𝜕𝑢 𝜕𝑢
∇𝑢 = 𝑥̂ + 𝑦̂ + 𝑧̂
𝜕𝑥 𝜕𝑦 𝜕𝑧
And what I am doing is; I am taking a dot product between this velocity vector and the
gradient vector.
𝜕𝑢 𝜕𝑢 𝜕𝑢
⃗ . ∇𝑢 = 𝑢
𝑉 +𝑣 +𝑤
𝜕𝑥 𝜕𝑦 𝜕𝑧
Now,
𝜕𝑢
𝑎𝑥 = ⃗ . ∇𝑢
+𝑉
𝜕𝑡
𝜕𝑣
𝑎𝑦 = ⃗ . ∇𝑣
+𝑉
𝜕𝑡
23
𝜕𝑤
𝑎𝑧 = ⃗ . ∇𝑤
+𝑉
𝜕𝑡
Or combining all of them; I can write the acceleration vector:
𝜕𝑢 𝜕𝑣 𝜕𝑤
𝑎 = 𝑎𝑥 𝑥̂ + 𝑎𝑦 𝑦̂ + 𝑎𝑧 𝑧̂ = ( ⃗ . ∇𝑢) 𝑥̂ + ( + 𝑉
+𝑉 ⃗ . ∇𝑣) 𝑦̂ + ( ⃗ . ∇𝑤) 𝑧̂
+𝑉
𝜕𝑡 𝜕𝑡 𝜕𝑡
So, if I combine this that and that I get that is equal to ∂ by ∂ t of this is the x component
of velocity, y component of velocity, z component of velocity; that is just velocity itself
plus the other three terms combine to give you:
⃗
𝜕𝑉
𝑎= ⃗ . ∇𝑉
+𝑉 ⃗
𝜕𝑡
So, that is the way you have to calculate acceleration in an Eulerian frame of reference ok.
So, really this kind of derivative is called material derivative or substantial derivative or
even sometimes total derivative ok.
24
So, this basically gives you how what is the; this expression gives you the acceleration,
but you could find out the material derivative of any quantity of interest. For example, you
can calculate the material derivative for temperature, pressure. So, if you want to calculate
it for temperature that would be; so, this is typically denoted with a simple D/Dt as:
𝐷𝑇 𝜕𝑇
= ⃗ . ∇𝑇
+𝑉
𝐷𝑡 𝜕𝑡
So, that tells you what is the material derivative of temperature; what is the physical
meaning of material derivative? It is just that if you are following a fluid particle; what is
the change that you are going to see? So, if you are going to calculate the material
derivative of temperature, if you are following a fluid particle; what is the change in
temperature that you are going to see; so, that is what material derivative represents. So,
in this box expression if you are following a fluid particle; what is the change in the
velocity that you are going to see and that is nothing, but the acceleration. So, that is the
way you have to define acceleration.
25
Now having found; acceleration, the next thing that you are going to you do; so you are
going to look at a fundamental law that we will keep using ok. And that is nothing, but
mass conservation; it is a very simple fact ok. Unless you know there is some reaction or
there is some source you know or unless; you know there is a sink the mass of the fluid;
mass of the fluid quantity that you are going to look at, it always remains fixed; it is not
going to change.
So, you basically want to express that fact as a mathematical equation and that is what we
are going to do now. So, what we will do; again remember we are actually doing a
differential approach. So, our intention is to write down this fact of mass conservation as
a differential equation. And the way you would do its will start by considering a fluid
element; so let us say we will take a fluid element ok. So, that is a fluid element which is
located at the origin of a coordinate system; so I will draw the coordinate system.
So, we will select this as x axis, that as y axis and this as the z axis ok. So, each of this
small elements; so to the small elements; let us say this side is dx, this is dy and this is d
z. And let us say there is some fluid that is flowing and you are looking at a small domain.
So, this is not a fluid element; this is a small element that you are considering in your entire
domain and there is a fluid that is must be flowing in the domain.
And then let us say that fluid is going to; so we will consider each face first let us say the
face that I am going to shade ok. So, that is the face that we are considering in the first,
26
which is actually face that is perpendicular to x direction ok; which is in located in the y z
plane. And what I am trying to calculate is; what is the total amount of fluid that is going
to go through that that face ok?
So, if I look at the face x and I want to calculate what is the total fluid that is entering; so
that is going to be. So, we what I want to write is the total mass that is entering and the
total mass is going to be equal to density times the volumetric flow rate; so density times
the volumetric flow rate. So, the volumetric flow rate is going to be ρ times u times d y, d
z. So, this d u times d; so u is the velocity that is perpendicular to the plate velocity times
area dy dz is a small area; u dy dz gives you the flow rate multiplied by density; gives you
the mass flow rate:
𝑚𝑖𝑛 = 𝜌𝑢 𝑑𝑦 𝑑𝑧
So, this is the mass that is going to be entering from this side. Now I want to again write
down what is the mass that is going to leave. Now in the absence of any other information
the best way to do it is to write down it has a Taylor expansion. So, if ρ u times dy by dz
is entering the left face, then let us say on the you move a distance dx the one which is exit
in is going to be:
𝜕
𝑚𝑜𝑢𝑡 = 𝜌𝑢 𝑑𝑦 𝑑𝑧 + (𝜌𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥
So, this is nothing, but the Taylor series expansion. So, it is basically that fact is that what
we have used to write down what is the; what is the amount that is going out. So, let us
actually try to make a table now based on this figure.
27
FACE IN OUT DIFFERENCE
x 𝜌𝑢 𝑑𝑦 𝑑𝑧 𝜕 𝜕
𝜌𝑢 𝑑𝑦 𝑑𝑧 + (𝜌𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧 − (𝜌𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥 𝜕𝑥
y 𝜌𝑣 𝑑𝑥 𝑑𝑧 𝜕 𝜕
𝜌𝑣 𝑑𝑥 𝑑𝑧 + (𝜌𝑣)𝑑𝑥 𝑑𝑦 𝑑𝑧 − (𝜌𝑣)𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑦 𝜕𝑦
z 𝜌𝑤 𝑑𝑥 𝑑𝑦 𝜕 𝜕
𝜌𝑤 𝑑𝑥 𝑑𝑦 + (𝜌𝑤)𝑑𝑥 𝑑𝑦 𝑑𝑧 − (𝜌𝑤)𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑧 𝜕𝑧
Now, if you have gotten what is the difference; you can add up this difference and what
can you say about this difference? So, see certain amount of fluid entered the volume and
then certain amount left, whatever be the difference that you have that difference what
should happen to that difference? That difference will constitute to change in the mass in
that particular element right.
28
So, in minus out should give rise to the accumulation. So, if mass is accumulating in that
fluid element; that means, it is going to essentially change the density of that fluid in that
volume.
So, we can say that if we write down the rate of change of density in the elemental volume
dx, dy, dz; that should be equal to In minus Out; because if In is exactly equal to Out, there
is no accumulation, density cannot change, there should not be anything else, but if there
is a change that should be reflected as a change in density.
So, then we can calculate the rate of change of density:
𝑅𝑎𝑡𝑒 𝑜𝑓 𝑐ℎ𝑎𝑛𝑔𝑒 𝑜𝑓 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑖𝑛 𝑡ℎ𝑒 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑙 𝑣𝑜𝑙𝑢𝑚𝑒 𝑑𝑥 𝑑𝑦 𝑑𝑧 = 𝑚𝑖𝑛

̇ − 𝑚𝑜𝑢𝑡
̇
𝜕𝜌 𝜕 𝜕 𝜕
𝑑𝑥 𝑑𝑦 𝑑𝑧 = [− (𝜌𝑢) − (𝜌𝑣) − (𝜌𝑤)] 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜌 𝜕 𝜕 𝜕
+ (𝜌𝑢) + (𝜌𝑣) + (𝜌𝑤) = 0
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
That equation is the equation that represents the fact that mass is conserved ok. So, if there
is a change in density that will be seen as a difference in the flow rate between two faces
or if there is a difference in the flow rate there should be seen as a change in density ok;
so this is the equation that describes the mass conservation.
29
We can write this equation in a much more convenient form:
𝜕 𝜕 𝜕
∇= 𝑥̂ + 𝑦̂ + 𝑧̂
𝜕𝑥 𝜕𝑦 𝜕𝑧
⃗ = 𝜌𝑢𝑥̂ + 𝜌𝑣𝑦̂ + 𝜌𝑤𝑧̂

𝜌𝑉
𝜕 𝜕 𝜕
⃗ =
∇. 𝜌𝑉 (𝜌𝑢)𝑥̂ + (𝜌𝑣)𝑦̂ + (𝜌𝑤)𝑧̂
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜌
⃗ ) = 0 [𝑀𝑎𝑠𝑠 𝑐𝑜𝑛𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛]
+ ∇. (𝜌𝑉
𝜕𝑡
Now at this point I will talk about an incompressible fluid.
30
So, incompressible fluid ok; let us assume that ρ is a constant density of the fluid that you
are considering is a constant. And if it is a constant, then it is; it is not changing with time,
𝜕𝜌
it is not changing with space. So, if you have ⃗ ) = 0 it is not changing with
+ ∇. (𝜌𝑉
𝜕𝑡
time and you can take ρ out because it is not changing with space.
𝜕𝜌
⃗)=0
+ ∇. (𝜌𝑉
𝜕𝑡
⃗ = 0 (𝐼𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑢𝑖𝑑)
∇. 𝑉
𝜕𝑢 𝜕𝑣 𝜕𝑤
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
So, this is the mass conservation equation and its very simple to write ok. So, because its
divergence of V; this is typically also called equation of continuity.
So, often equation of continuity and mass conservation ok; they refer to the same fact and
this particular form where we write divergence of V is equal to 0 only refers to the
incompressible fluids.
31
So, that is the equation for mass conservation; I am going to apply product rule to each of
the terms.
𝜕𝜌 𝜕 𝜕 𝜕
+ (𝜌𝑢) + (𝜌𝑣) + (𝜌𝑤) = 0
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜌 𝜕𝑢 𝜕𝜌 𝜕𝑣 𝜕𝜌 𝜕𝑤 𝜕𝜌
+𝜌 +𝑢 +𝜌 +𝑣 +𝜌 +𝑤 =0
𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧
𝜕𝜌 𝜕𝜌 𝜕𝜌 𝜕𝜌 𝜕𝑢 𝜕𝑣 𝜕𝑤
+𝑢 +𝑣 +𝑤 +𝜌[ + + ]=0
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝐷𝜌
⃗)=0
+ 𝜌(∇. 𝑉
𝐷𝑡
So, this is another form in which you can write the continuity equation ok. So, we wrote
continuity equation in two forms.
Also,
𝐷𝜌
𝐼𝑓 ⃗ =0
= 0 → ∇. 𝑉
𝐷𝑡
32
Now remember the density might be changing locally, but if you are following a particular
fluid element and if its density does not change; then is D ρ by Dt equal to 0, in which case
also you get divergence of V equal to 0:
𝜕𝑢 𝜕𝑣 𝜕𝑤
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
So, in this case you will say that the flow is incompressible; flow is incompressible as
opposed to the fluid being incompressible. So, earlier we said that the fluid is
incompressible and we threw away all the terms containing derivatives of ρ and then we
found that you basically get divergence of V is equal to 0.
But you will also get divergence of V equal to 0; if you find that following a fluid element
does not change its density ok; in which case again you will have divergence of V is equal
to 0. So, most often the cases that we will consider would be either the flow would be
incompressible or we will take the fluid to be incompressible. And in either case we will
have divergence of V equal to 0 and that is the equation that we will use.
33
Fluid Mechanics
Lecture – 05
Equation of linear momentum 1
So, the last class we looked at the differential equation that describes mass conservation
and that is because we wanted to express mechanics of fluids, using a differential approach
and mass conservation is a fact that is associated with any system that we are going to deal
with and therefore, we expressed that fact as a partial differential equation. And when we
did that we basically had two variables in that system one was density and the other was
the velocity field.
We also simplified the relation for a constant density and then derived the equation of an
incompressible fluid as well. So, we are going to continue and. So, we are trying to talk
about dynamics of fluids and as far as dynamics of fluids are concerned, these things are
moving they are in motion and therefore, the way they start moving is because you are
applying some external force onto the system. And so and the kind of fluid flow that is
generated is a consequence of how much is the force that you are applying.
So, what you are really looking for is some kind of a relation between the velocities that
are generated and connect that to the force that you have applied. Now for simple systems
like for example, if you are throwing a stone you know how to write down the velocity of
the stone at every point, you will actually write down the Newton’s second law which is
nothing, but force is equal to mass into acceleration. And if you know the force you can
calculate what is the acceleration and therefore, you will know what is the velocity at any
instant of time.
So, we need something very similar a corresponding expression, but for fluid flow and
also we need to do it in an Eulerian frame. Because we are going to look at velocity at
every point in space and in time and we want to talk about forces acting at a particular
point and see what is the kind of velocities that are generated. So, that is what we were
going to look at ok. So, we are going to derive now equation of linear momentum
conservation.
34
So, we are going to talk about how what is the law that is going to come about, then we
are going to talk about conservation of linear momentum. And we will proceed exactly the
way that we did yesterday, we will first write the will first consider a control volume let
us consider a control volume ok. So, that is a control volume and let us say that we are
going to fix it at the origin of our coordinate system. So, that is the coordinate system. So,
choose x y and z ok.
So, this length is dx that length is dy and this length is dz. So, these are the three lengths
and what we are going to say is that, the fluid actually has a density ρ. So, ρ is the density
of the fluid and ρ is the density of the fluid and let us say V is the velocity of the fluid so
it’s a velocity field ok. So, which varies at x y z and time and what we need to do now is
that we will find out, how much is the momentum that is going into the system and how
much is the momentum that is going out of the system and whatever is left out will give
rise to the accumulation. So, just like we proceeded yesterday, we will actually talk about
momentum in and out through each face.
35
So, let us mark let us say the first face that we are going to consider is what I have marked
with red color ok. So, that is a face that is perpendicular to the x direction and its area is
dy dz. So, if you, so, what is the momentum that is associated with the fluid? So, we need
to get let us say momentum flux; momentum flux in through the x face. So, that is going
to be equal to ρ times V. So, that is going to be the momentum density because its mass
times velocity is your momentum.
So, this is density is mass per unit volume. So, mass per unit volume times velocity is
going to give you the momentum density and multiplied by u times dy d z. So, u times dy
dz is going to be the flow rate. So, velocity times area is the flow rate. So, this is the flow
rate times the momentum is the total momentum that is getting in through the face dy by
dz and what will be the momentum that is out through the other face? Let us say the face
that I mark it as green again we can use the Taylor series expansion:
⃗ )𝑢 𝑑𝑦 𝑑𝑧
𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚𝑖𝑛 = (𝜌𝑉
𝜕
⃗ )𝑢 𝑑𝑦 𝑑𝑧 +
𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚𝑜𝑢𝑡 = (𝜌𝑉 ⃗ 𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑥
So, similarly we can do it for each face. So, let us do that.
36
So, let us say face momentum flux in momentum flux out and the difference which is in
minus out. Now, we can do the same thing for the y face. So, let us look at the y face now.
So, the y face is the momentum that is entering from the bottom and then the momentum
that is going above from the top. So, similar calculation.
FACE MOMENTUM MOMENTUM FLUX DIFFERENCE

FLUX IN OUT
x ⃗ )𝑢 𝑑𝑦 𝑑𝑧
(𝜌𝑉 ⃗ )𝑢 𝑑𝑦 𝑑𝑧
(𝜌𝑉 𝜕
− ⃗ 𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑥
𝜕
+ ⃗ 𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑥
y ⃗ )𝑣 𝑑𝑥 𝑑𝑧
(𝜌𝑉 ⃗ )𝑣 𝑑𝑥 𝑑𝑧
(𝜌𝑉 𝜕
− ⃗ 𝑣)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑦
𝜕
+ ⃗ 𝑣)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑦
z ⃗ )𝑤 𝑑𝑦 𝑑𝑥
(𝜌𝑉 ⃗ )𝑤 𝑑𝑦 𝑑𝑥
(𝜌𝑉 𝜕
− ⃗ 𝑤)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑧
𝜕
+ ⃗ 𝑤)𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑧
37
So, this is the quantity that is interesting to us, because that tells you what is the momentum
in minus momentum out. And that momentum in minus momentum out should be the
should be equal to the rate of change of momentum in that particular volume. So, if there
is in is not equal to out; that means, the momentum of the fluid in that particular volume
is going to be changed and we know how to do that.
The rate of change of momentum.
𝜕
𝑅𝑎𝑡𝑒 𝑜𝑓 𝑐ℎ𝑎𝑛𝑔𝑒 𝑜𝑓 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 = ⃗ )𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑡
So, rate of change of momentum must be then equal to the parts that I have marked in that
red box, but then if you think about it a little more seriously, what you know is that the
momentum or the rate of change of momentum is not just because there is a difference in
the momentum IN minus momentum OUT, you can get a rate of change of momentum if
you apply a force also right.
So, for example, if something is moving with a certain momentum and if you apply a force,
you are actually going to change its momentum. So, the rate of change of momentum is
not just equal to IN minus OUT, it actually has to be also equal to the total forces acting
on that fluid element ok. So, therefore, that right equation is some more forces plus
38
momentum in minus momentum out has to be equal to the total rate of change of
momentum.
So, therefore, we find that our equation should really be sum of forces that is acting plus
all the quantities, that we have found out as in this red curve red box. So,
𝜕 𝜕 𝜕
∑𝐹 − ⃗ 𝑢)𝑑𝑥 𝑑𝑦 𝑑𝑧 −
(𝜌𝑉 ⃗ 𝑣)𝑑𝑥 𝑑𝑦 𝑑𝑧 − (𝜌𝑉
(𝜌𝑉 ⃗ 𝑤)𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕
= ⃗ )𝑑𝑥 𝑑𝑦 𝑑𝑧
(𝜌𝑉
𝜕𝑡
So, that is the expression that now we have gotten when we have said that the sum of the
forces plus momentum in minus momentum out is equal to rate of change of momentum
and now we need to simplify this expression. So,
𝜕 𝜕 𝜕 𝜕
∑𝐹 = [ ⃗)+
(𝜌𝑉 ⃗ 𝑢) +
(𝜌𝑉 ⃗ 𝑣) + (𝜌𝑉
(𝜌𝑉 ⃗ 𝑤)] 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
⃗
𝜕𝑉 𝜕𝜌 ⃗
𝜕𝑉 𝜕(𝜌𝑢) ⃗
𝜕𝑉 𝜕(𝜌𝑣) ⃗
𝜕𝑉
∑ 𝐹 = [𝜌 ⃗
+𝑉 + 𝜌𝑢 ⃗
+𝑉 + 𝜌𝑣 ⃗
+𝑉 + 𝜌𝑤
𝜕𝑡 𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧
𝜕(𝜌𝑤)
⃗
+𝑉 ] 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑧
𝜕𝜌 𝜕(𝜌𝑢) 𝜕(𝜌𝑣) 𝜕(𝜌𝑤) ⃗

𝜕𝑉 ⃗
𝜕𝑉 ⃗
𝜕𝑉
∑ 𝐹 = [[ + + + ⃗ +𝜌
]𝑉 + 𝜌𝑢 + 𝜌𝑣
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑡 𝜕𝑥 𝜕𝑦
⃗
𝜕𝑉
+ 𝜌𝑤 ] 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑧
39
So, these are the terms that are in underlined as red in the above expression and other
terms. Now what is this thing term in this square bracket that I have put that is nothing,
but continuity equations. If you look at your notes it’s essentially equal to 0.
So,
⃗
𝜕𝑉 ⃗
𝜕𝑉 ⃗
𝜕𝑉 ⃗
𝜕𝑉 ⃗
𝜕𝑉
∑𝐹 = 0 + 𝜌[ +𝑢 +𝑣 ⃗ ] 𝑑𝑥 𝑑𝑦 𝑑𝑧
+ 𝑤 ] 𝑑𝑥 𝑑𝑦 𝑑𝑧 = 𝜌 [ + ∇. 𝑉
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑡
40
So, that is where we have ended up with. So, now, we need to talk about the forces acting
on the fluid element.
The forces can be actually two types; one is called body forces and the other is called
surface forces. So, when I say body forces it’s an external force that acts on the entire the
on or the force that depends on the volume of the fluid element that you are looking at and
you know one such force for example, gravity ok. Gravity depends upon what is the
amount of the fluid that you are looking at and here its obvious that if I take dx dy dz to
the left hand side, I really need what is the force density or the force acting per unit volume
and the force acting per unit volume is nothing, but ρ times g ok.
So, that I, so, the really the forces m times g, but if you are talking about the force per unit
volume its ρ times g ok. So, ρ g is the force that is going to come up as a body force. Now
surface forces are the force that acts on the surfaces of the fluid element. So, if you have
let us say let us look at the same fluid element, will make another diagram ok. So, that is
the fluid element and we had it has x y, I will just system x y z ok.
Now on any face of it let us say this fluid element is part of the fluid this fluid element is
surrounded by other fluid elements. So, the other fluid elements basically will be exerting
some forces on this fluid element as well and that force is exerted on the surface of this
fluid element. So, what I am trying to say is that let us say if that is one of the fluid element
above that draw again ok.
41
So, that is one; then above it there is going to be another fluid element, then on the side
there is going to be another fluid element and so on. So, the red fluid is going to exert some
force on the black fluid, the black fluid is going to exert some force on the green fluid. So,
this is all forces exerted on the surface not on the not its that force exerted is essentially
proportional to the surface area because that force is getting transferred across the area ok.
So, those are the kind of forces that you call as surface forces and there are two surface
forces one is something that you already know and that is the pressure the second is the
shear stress.
So, if you remember the definition of the Newton’s law of viscosity, we defined it as stress
is equal to a constant times strain and that constant is what we called as viscosity. So, stress
is again defined as force per unit area. So, the pressure forces shear stresses are examples
of surface forces ok. Because you can see that that force is defined per unit area unlike the
body force which is defined per unit volume, because it depends upon the mass of the
object that is under consideration. So, since we now understand what are the body forces
and what are the surface forces.
So, as far as body force is concerned, we know what is the expression that we need to use
in the equation that we have. So, this is the form of the equation at the moment, this is the
form of the equation at the moment ok. So, the left hand side we have simply written a test
total forces, but we need to be more explicit about it. So, ρ g is the body force contribution
that is going to go into the sum of f and then we need to find out what are the what is the
form of surface forces that should go in the expression for capital F.
42
Fluid mechanics
Lecture - 06
Equation of linear momentum 2
So, the surface force is typically represented using a second order tensor, which is called
as 𝜎𝑖𝑗 . 𝜎𝑖𝑗 is actually a force acting in i direction on an area directed in j direction. So, you
can always define a vector normal to any area that you consider and let us say j is the
direction and i is the direction of the force ok. So, that is how you will represent your
surface forces for example, if you talk about σ x y ok; that means, you have a force which
is acting in the x direction, but it is on an area that is perpendicular to y direction. So, that
is the idea of defining it. So, if you want to mark here, let us say in our diagram we want
to represent σ.
43
So, remember. So, how many components of σ would be there? There will be nine
components. There will be nine components of the stress and is often conveniently
represented as a matrix and that can be simplified or that basically constitutes two parts
this is the total stress, that stress can be either due to pressure or due to shear stresses ok.
So, we said there are two kinds of surface forces pressure and stress shear stress. So, the
pressure is typically denoted by P the shear stress is let us say we will use τ to represent
that.
𝜎𝑥𝑥 𝜎𝑦𝑥 𝜎𝑧𝑥 −𝑃 + 𝜏𝑥𝑥 𝜏𝑦𝑥 𝜏𝑧𝑥

[𝜎𝑥𝑦 𝜎𝑦𝑦 𝜎𝑧𝑦 ] = [ 𝜏𝑥𝑦 −𝑃 + 𝜏𝑦𝑦 𝜏𝑧𝑥 ]
𝜎𝑥𝑧 𝜎𝑦𝑧 𝜎𝑧𝑧 𝜏𝑥𝑧 𝜏𝑦𝑧 −𝑃 + 𝜏𝑧𝑧
So, this is if you want to write your surface stresses as pressure and shear stresses viscous
stresses separated. So, I can also show it schematically. So, for example, let us look at 𝜎𝑥𝑥 .
Its a force that is acting in x direction on a plane in x direction. So, if you looked at
𝜎𝑥𝑥 would actually be this. Now you I can also define of this. So, the area that I am looking
at is this the one which is perpendicular to x direction ok. So, I can define two more forces
I can define a force that is acting in y direction on this area x. So, a force that is acting in
x direction, so, acting in y direction on this area x that I can write it as σ y x, I can talk
about another stress that is acting in z direction.
44
So, that will be σ z x. Similarly if I consider another point let us say that this on this plane.
So, that plane is a plane that is perpendicular to y. So, if I define a σ yy, that will be a force
that is acting in y direction on the area y, I can define a force that is acting in x direction
on this plane y. So, that we will write it as σ x y or I can write down a force that is acting
in the z direction. So, that will be σ z y ok. So, these are the various components of forces
or various components of this matrix the stress that we are talking about.
Now, what we really need is that, we need to write down how these stresses come as part
of the force that we are talking about, because that is the expression that we are really
looking for. In other words we need to write down the stresses as forces. So, we have
already done that in the context of a equilibrium system a fluid which was under you know
equilibrium which was not flowing a static fluid and there we found that the force is
essentially written as gradient of pressure. So, we found that its not pressure that is really
important, its a differences in pressure that is what is going to be important its the gradient
of pressure that generates a force. Similarly, we need to we will give something similar
here we need to find out how is stress going to be reflected as force.
So, let us make another diagram let us consider a fluid element or your cell notation x y z,
let us look at one of the faces say the red face ok. So, there I have σ x x. So, σ xx is acting
on that side ok. So, if σ xx is acting on this side, on the other side we can say that is the
force this stress is going to be σ x x plus ∂ by ∂ x of σ xx or if you want to talk about the
45
force is going to be multiplied by dy d z and on this side as well its going to be multiplied
by dy dz. In other words the forces that is going to come in x direction from this particular
component, so, difference ok.
The difference between the force acting on the left side and the force acting on the right
side is the force that is acting in x direction, :
𝜕 𝜕 𝜕
𝑑𝐹𝑥 = (𝜎𝑥𝑥 )𝑑𝑥 𝑑𝑦 𝑑𝑧 + (𝜎𝑦𝑥 )𝑑𝑥 𝑑𝑦 𝑑𝑧 + (𝜎𝑧𝑥 )𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕 𝜕 𝜕
𝐹𝑥 = [ (−𝑃 + 𝜏𝑥𝑥 ) + (𝜏𝑦𝑥 ) + (𝜏𝑧𝑥 )] 𝑑𝑥 𝑑𝑦 𝑑𝑧
𝜕𝑥 𝜕𝑦 𝜕𝑧
So, that is the total force that is going to be acting on this fluid element ok. So, the forces
defined positive in the positive x direction and therefore, the net force that is acting in the
plus x direction is given by that extra contribution. So, that is the force in x direction.
Similarly, you can derive a force in y direction and you can derive a force in z direction as
well.
In other words this is going to be the force density this is going to be the force density that
we really need in x direction in the expression that we had derived here. So, that was the
expression that we had derived here. So, there is a ρ g that we have we have it from the
46
body force and then we have a shear stress term that is going to come here. So, we can
now use all this fact and simplify and write down our momentum equation as:
𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑃 𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥

𝜌[ +𝑢 +𝑣 +𝑤 ]=− + + + + 𝜌𝑔𝑥
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧
So, there is a full expression for the momentum conservation in x direction. Similarly we
can write down the y part:
𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑃 𝜕𝜏𝑥𝑦 𝜕𝜏𝑦𝑦 𝜕𝜏𝑧𝑦

𝜌[ +𝑢 +𝑣 +𝑤 ]=− + + + + 𝜌𝑔𝑦
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝑤 𝜕𝑤 𝜕𝑤 𝜕𝑤 𝜕𝑃 𝜕𝜏𝑥𝑧 𝜕𝜏𝑦𝑧 𝜕𝜏𝑧𝑧

𝜌[ +𝑢 +𝑣 +𝑤 ]=− + + + + 𝜌𝑔𝑧
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
So, this is what the equation of linear momentum written for the fluid element in the
Eulerian frame of reference. It just looks complicated, but actually it’s not that is ok. So,
the left hand side if you look at that is nothing, but the substantial derivative the substantial
derivative of velocity and the substantial derivative of velocity is nothing, but acceleration.
So, the left hand side really represents the acceleration and if the left hand side is
acceleration the right hand side should be force remember we have written it for unit mass
ok. So, this side is going to be nothing, but total forces.
So, the forces consist of two parts, one is that you have got gravity forces and the other is
that you have got viscous forces and then we have got pressure forces. So, in each equation
acceleration is equal to pressure plus viscous forces plus gravitational forces that is all it
is; now it’s convenient. So, remember we did not write what τ is we just defined τ as a
shear stress, we defined it as a surface force and we said that is what is acting on the fluid
element, but we need to further simplify it in order to you know use it further ok. Because,
if you write it in this particular form there are too many unknowns and what are the
unknowns? For example, velocities are unknown the way its written pressure we do not
know, shear stress we do not know that is where Newton’s law of equation comes into
picture.
Newton’s law of viscosity comes into the picture we can use Newton’s law of viscosity to
write stress in terms of velocity gradients ok. So, stress if you write it in terms of velocity
47
gradients the unknowns the stress being unknowns will go away and instead we are going
to get velocities. So, we can try to do that now.
So,
𝜕𝑢 𝜕𝑢 𝜕𝑣 𝜕𝑢 𝜕𝑤
𝜏𝑥𝑥 = 2𝜇 ; 𝜏𝑦𝑥 = 𝜇 ( + ) ; 𝜏𝑧𝑥 = 𝜇 ( + )
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑧 𝜕𝑥
Remember, when we derived the Newton’s law of viscosity we simply connected stress
with rate of strain here I have written down in a much more general fashion, where each
of the components of the shear stress is related to what you call as the velocity how it is
connected to the velocity gradient ok.
48
And this can be further simplified.
𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥 𝜕 𝜕𝑢 𝜕 𝜕𝑢 𝜕𝑣 𝜕 𝜕𝑢 𝜕𝑤

+ + = (2𝜇 ) + (𝜇 ( + )) + (𝜇 ( + ))
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑧 𝜕𝑧 𝜕𝑥
𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑣 𝜕 2𝑢 𝜕 2𝑤
= 2𝜇 2 + 𝜇 2 + 𝜇 +𝜇 2 +𝜇
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑧
𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑣 𝜕 2𝑤
= 𝜇[ + + ] + 𝜇 [ + + ]
𝜕𝑥 2 𝜕𝑦 2 𝜕𝑧 2 𝜕𝑥 2 𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑧
𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢 𝜕 𝜕𝑢 𝜕𝑣 𝜕𝑤
= 𝜇[ 2
+ 2 + 2] + 𝜇 [ + + ]
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥 𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢

+ + = 𝜇 [ 2 + 2 + 2] + 0
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
So, this this entire expression is 0 for an incompressible fluid, that is what we had derived
incompressible fluid so; that means, it’s not there or in other words it’s not there. So, what
I wanted to tell you is that the right hand side of the equations that we derived essentially
simplifies into this particular form and therefore, I can substitute that back into the
equations that we derived. So, the one that you are seeing in this purple color we can now
replace. And if you replace it and we can do that for other equations as well and then the
equations simplify further.
49
And the simplified equations would be
𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝐷𝑢 𝜕𝑃
𝜌[ +𝑢 +𝑣 +𝑤 ]=𝜌 =− + ∇2 𝑢 + 𝜌𝑔𝑥
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝐷𝑡 𝜕𝑥
Now, similarly for the y and z momentum equation
𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑣 𝐷𝑣 𝜕𝑃
𝜌[ +𝑢 +𝑣 +𝑤 ] =𝜌 =− + ∇2 𝑣 + 𝜌𝑔𝑦
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝐷𝑡 𝜕𝑦
50
𝜕𝑤 𝜕𝑤 𝜕𝑤 𝜕𝑤 𝐷𝑤 𝜕𝑃
𝜌[ +𝑢 +𝑣 +𝑤 ]=𝜌 =− + ∇2 𝑤 + 𝜌𝑔𝑧
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝐷𝑡 𝜕𝑧
So, that is the equations of linear momentum expressed as you know only in terms of
velocity. So, if you look at now you only have u v w and pressure as the unknowns ok. So,
let us see unknowns are u v w and p. So, that is 4 in number how many equations are there?
You have three equations here equation of x y z linear momentum and then you also have
conservation of mass. So, you will have four equations, four unknowns and given any
problem you will be able to solve and find out the solution ok.
The equations look a little more detailed ok, but there are many cases where it is it can be
so simplified and solved and we will look at some of the cases later on, which will give us
some insight into what does the fluid flow look like in a given situation. So, these equations
that we have derived we have assumed Newton’s law of viscosity, we have assumed
incompressibility. So, this is really applicable for incompressible fluids and we have also
used Newtonian assumption in other words we have used Newton’s law of viscosity and
of course, we are using we have expressed this in Cartesian coordinate system.
51
Fluid and particle mechanics
Lecture - 07
Bernoulli's Equation
So, in the last few classes we have been trying to take an approach based on differential
equations to understand fluid motion. We only derived the governing equations, one of
them the fact that you know mass is conserved there is a mass balance; we express that as
a differential equation. The second one was the momentum balance or force is equal to
mass into acceleration, we represented that fact also as a differential equation.
Now, we should learn how to use these equations for the kind of problems that we want to
analyze, but before that there is one more you know balance that we often use and that is
about energy right the energy is conserved or there is an energy balance. So, one can use
that fact also and write down an equation, that can be quite useful. Now the we will not do
it as rigorously as we did for the mass balance and for the momentum balance, we are
going to only do it so, that it looks very similar to something that you are already familiar
with and that is Bernoulli’s equation ok.
So, we are going to see how energy conservation is nothing, but similar to what you already
know the Bernoulli’s equation and how does it get modified really for a real fluid flow.
52
So, we are going to now look at Bernoulli’s equation we will try to derive that ok. So, let
us say you have some flow field so; that means, there is a velocity, there is a pressure and
let us also in general assumed that you know the density is also changing. So, that we don’t
have to restrict ourselves for incompressible fluid flows and let us say we will consider a
differential element.
So, for this particular one we are going to really write down just energy balance as a 1D
equation ok. So, we will not consider therefore a cubic element, we will consider
something simpler let us say a fluid element which is made of which is actually a stream
tube I will explain what is a stream tube. Let us say that is a stream tube stream tube is
something that is made up of stream lines. So, you take the surface of the stream tube, the
surface of the stream tube is made up of stream lines are made of stream lines what are
stream lines?
Stream lines. So, if you know a stream line and if you draw a tangent on this on the stream
line, that is actually going to give you the velocity field. So, let us draw some stream line
let us say that is a stream line, if you take a particular point and let us say a tangent there,
that is actually going to give you a velocity field. If I take another point and draw a tangent
there. So, let me draw the tangent of the different color red ok. So, that is going to give me
the velocity there if I have here that is going to give me the velocity field here ok. So,
stream line is essentially connected by those velocity vectors. So, if there is a change in
the direction of the flow stream line will show that. So, you can draw stream lines at every
point in the flow field. So, stream tube is basically a surface made up of streamlines. So,
for us; that means, that if this is what the stream tube is if I take any point and draw a
tangent that is basically going to be the direction in which the fluid is flowing. There cannot
be a fluid motion that is perpendicular because by definition stream line is the one which
is drawn such that the velocity is tangential to it ok.
In other words if you consider a stream tube because there you cannot that is made up of
stream lines, there cannot be a flow from In or Out. So, there is a flow that is going to go
In and there is a flow that is Out. So, this is in that is OUT. So, this is like really a tube and
then there is fluid that is coming from the left side and then the fluid is going to go outside,
but on the surface because its made up of stream tube there is no flow through the surface.
So, its the stream tube is as if its like a rigid object for that matter because there is no flow
that is going to go through the stream tube, but its really made up of fluid element ok.
53
So, that is the stream tube that we are considering. Let us say this side the pressure is p
density is ρ velocity is v and outside its P plus dp is the pressure ρ plus d ρ is the density
v plus d v is the velocity and there is a change in area. So, there could be a change in area
because fluid may not just go straight fluid probably would be expanding and flowing like
that the fluid might be flowing like that. So, you could you have a inlet area let us say A
and let us say the outlet area is A plus dA. So, this is the system under consideration.
𝐼𝑛: 𝑃, 𝜌, 𝑉, 𝐴
𝑂𝑢𝑡: (𝑃 + 𝑑𝑃), (𝜌 + 𝑑𝜌), (𝑉 + 𝑑𝑉), (𝐴 + 𝑑𝐴)
And what we will do is we’ll write down a mass balance and a momentum balance for this
particular fluid element that we have considered. So, we already know how to do it, we
can say that the mass In minus mass Out should be equal to rate of change of mass in the
system. So, if you think about mass balance. So, mass In minus mass Out has to be equal
to rate of change of mass inside right. So, what is the mass that is flowing into the system
that is going to be volumetric flow rate times density, but volumetric flow rate is nothing,
but the area times velocity. So, ρ times A times V is the mass that is growing In minus the
mass that is coming Out is going to be ρ plus d ρ times A plus d A times v plus dv and
that has to be equal to the rate of change of mass which is ∂ by ∂ t of the mass in that area.
So, let us say the, so, sorry in that volume. So, the volume is given by some area times
particular length. So, let us mark the length also, let us say from the center to here the
length is given by d s and also let us say that this is actually at an angle θ with respect to
the horizontal. So, the rate, so, that the volume of this is going to be approximately equal
to A times d s times density is going to be the rate of change of mass in the system. So,
this is nothing, but really change in a ρ into A V. So, let me rewrite it as:
𝜕
𝜌𝐴𝑉 − (𝜌 + 𝑑𝜌)(𝑉 + 𝑑𝑉)(𝐴 + 𝑑𝐴) = (𝜌𝐴𝑑𝑆)
𝜕𝑡
𝜕
𝑑(𝜌𝐴𝑉) = (𝜌𝐴𝑑𝑆)
𝜕𝑡
𝜕𝜌
𝐴𝑑𝑆 = −𝑑(𝜌𝑉)𝐴
𝜕𝑡
54
So, A is a particular area that you have considered and d is the total length of the
differential length of the fluid element that you have considered ok. So, that is what the
mass conservation equation gives you.
So, now we will write down the linear momentum equation. So, linear momentum equation
is nothing but again momentum In minus momentum Out has to be equal to the rate of
change of momentum, but we also know that the rate of change of momentum can come
from forces. So, we could write that In momentum minus Out momentum has to be equal
to rate of change, but that rate of change can also come from all the forces acting in the
system. Remember there is a fact that we used to write down the momentum conservation
equation.
𝐿𝑖𝑛𝑒𝑎𝑟 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛: ∑ 𝐹 + 𝐼𝑛 − 𝑂𝑢𝑡 = 𝑅𝑎𝑡𝑒 𝑜𝑓 𝑐ℎ𝑎𝑛𝑔𝑒
So, for this particular fluid element that we are looking at ok. So, there is a force there is a
force that is coming from pressure, there is gravity that is acting and then there could be
viscous forces and then we can also calculate what is the momentum that is going In what
is the momentum that is going Out ok. So, let us start with the right hand side the rate of
change of momentum can be written as ∂ by ∂ t of momentum of the fluid. So, that is mass
ok.
55
So, that is then density into volume times v. So, that is the rate of change of momentum in
that volume, that is equal to the differential of the momentum in minus a momentum out.
So, just like the mass conservation we can write it as:
𝜕
∑ 𝐹 − 𝑑(𝜌𝐴𝑉𝑉) = (𝜌𝐴 𝑑𝑆 𝑉)
𝜕𝑡
Now, here and we need to write down what are the forces one obvious force is the body
force ok. So, the body force which is coming from gravity. So, what will be the body force
is nothing, but just the weight of the system. So, here we have A as the inlet area d s as is
the length and what and then gravity is acting downwards only thing is that the. So, gravity
is acting downwards, but the element is actually you know up like that in at a particular
angle. So, we need to find out the component of gravity that is acting along the d s direction
ok.
So, this is the direction that we are trying to see now, its the direction ok. So, we are writing
down the force balance in that direction. So, in the direction of that pink arrow. So, we
have to find out what is the way what is the component of gravity, that is acting along it
and that is going to be nothing, but so, because gravity is acting downward ok. So, that is
the direction in which gravity is acting. So, its going to be g sin θ that is going to act in
that particular direction. So, body force is going to be area times length, A ds is going to
56
give you the volume times ρ the volume times density that is the mass times g sin θ and
its acting.
So, g is acting downwards. So, its minus g sin θ, because its acting at an angle θ with
respect to the horizontal now. So, let me make a smaller diagram here. So, that is the
direction in which g is acting, that is a kind of fluid that we were really looking at and we
said that is nothing, but d s right the length of the element. Let us say this in edge from a
particular base is at a height is z 1 and out edge is at a height is at z2.
So, if I make a triangle there with that angle θ. So, this is ds this is z 2 minus is z 1 let me
call it d z then sin θ is equal to d z divided by d s or d s sin θ is simply d z. So, I can use
that d s sin θ here and write this as minus A ρ g times dz ok. So, that is what the body
force is.
Now, we have got surface forces and we know that surface forces are two types; one is
pressure the other is shear stress ok. What we are going to do so we are going to neglect
shear stresses. So, we know that shear stresses is basically arising from viscosity or it is
due to friction. So, by really neglecting shear stresses we are talking about frictionless
fluid. So, if there is a fluid which was not having any viscosity does not feel any friction,
that was the case then there is no shear stresses and that is the case that we are want to look
at and then we will correct it later on how to account for friction. So, the only force that
we are this surface was that we have worried about really is the pressure force.
57
So, let me make a diagram again to tell, what is the best way to do it. So, that is our diagram
that we are looking at, we said there is an inlet pressure and then there is an outlet pressure.
So, p there is a p that is acting on the inlet and there is a p plus ∂ta p that is acting at the
outlet, but we know that its only the difference in pressure that is going to create a force
and therefore, let us say that the force at the inlet or the pressure here is 0 and the pressure
at the other side is simply dp because dp is the one that is going to really constitute the
pressure.
Here we have an area A and here we have an area A plus dA ok. So, the net force acting
is simply going to be dp times A plus d A in this manner right. But the point is that if you
write it as A times d p plus dp times d A and we are looking at really a differential element
where any of these differentials actually the values go to 0 where dp goes to 0 dA goes to
0. So, now, this quantity is going to be much smaller than A dp or in other words the
effective pressure force that is going to act is going to come mainly from A dp.
So, look at this when I write down the force this is the force arising from pressure, force
from pressure is what I have calculated which is basically based on the inlet side and the
outlet side, but there is actually pressure contribution that is going to come also from the
periphery from the lateral sides, I have neglected it because that is again going to be a
small number and you could actually show that that is going to be much smaller compared
to the contribution that I have identified here ok.
But let us not worry about how to show that at the moment let us see that its A into dp is
the net force that is going to be acted and that is going to act in the direction opposite of
your d s and therefore, the pressure contribution is going to be nothing, but minus A into
dp.
So, putting all of them together, we have minus A ρ g d z that is a force. So, where am I
substituting I am substituting it here and substituting it into that expression. I have the
force I have the change in the momentum and the rate of change. So,
𝜕
−𝐴𝜌𝑔 𝑑𝑧 − 𝐴𝑑𝑃 = 𝑑(𝜌𝐴𝑉 2 ) + (𝜌𝐴𝑉)𝑑𝑆
𝜕𝑡
Now,
𝑑(𝜌𝐴𝑉 2 ) = 𝑑(𝜌𝐴𝑉)𝑉 + 𝜌𝐴𝑉𝑑𝑉
58
𝜕 𝜕𝜌 𝜕𝑉
(𝜌𝐴𝑉)𝑑𝑆 = 𝐴𝑉𝑑𝑆 + 𝐴𝑑𝑆
𝜕𝑡 𝜕𝑡 𝜕𝑡
If you look at the continuity equation:
𝜕𝑉 𝑑𝑃
𝐴𝜌 (𝑑𝑆 + 𝑉𝑑𝑉 + + 𝑔𝑑𝑧) = 0
𝜕𝑡 𝜌
𝜕𝑉 𝑑𝑃
𝑑𝑆 + 𝑉𝑑𝑉 + + 𝑔𝑑𝑧 = 0
𝜕𝑡 𝜌
So, this is really an unsteady Bernoulli’s equation written in a differential form. On

integration we have,
𝑆2 𝑃2
𝜕𝑉 𝑉22 − 𝑉12 𝑑𝑃
∫ 𝑑𝑆 + +∫ + 𝑔(𝑧2 − 𝑧1 ) = 0
𝜕𝑡 2 𝜌
𝑆1 𝑃1
59
Now if I take the assumption that let us say were worried about steady state flows or steady
flows where time dependency is not there, then I will not consider this term and will also
say that ρ; ρ is a constant.
𝐹𝑜𝑟 𝑠𝑡𝑒𝑎𝑑𝑦 𝑓𝑙𝑜𝑤𝑠, 𝜌 → 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑉22 − 𝑉12 𝑃2 − 𝑃1
+ + 𝑔(𝑧2 − 𝑧1 ) = 0
2 𝜌
𝑃1 𝑉12 𝑃2 𝑉22
+ + 𝑧1 = + + 𝑧2
𝜌𝑔 2𝑔 𝜌𝑔 2𝑔
Now let me just tell you the assumptions that went in deriving that equation, because that
tells you when you should be using Bernoulli’s equation ok. So, we started of course, with
a simple you know differential element, we are actually doing here in a 1 D fashion we are
just worried about the direction which we called as d s we calculated the mass balance we
calculated the momentum balance and the first assumption that we made is about the
frictionless fluid we said the fluid has no friction ok.
So, that is one that is important because then; that means, Bernoulli’s equation should be
applied only when your fluid has really no friction. Now the second assumption that we
made is that it is applicable only for steady flows, third is that we are assuming that the
60
density is constant ok. So, these are the three cases these are the three assumptions that
have really gone in in deriving the Bernoulli’s equation now.
So, let us see what this means or let me just give you this. So, pressure this first term is
typically called pressure head ok. This is called velocity head and then of course, then
height some height ahead. So, head really means some height. So, why is it called before
that. So, what are these terms ok? So, this is really nothing, but the pressure work the work
this is the work is done well. So, by the pressure right. So, whenever the fluid is moving
pressure is a force that is acting and there is the fluid is moving. So, there is a 4 times
velocity that gives rise to a work in the work or the fluid. So, pressure p this first term
really represents that work that is done on the fluid.
Ok, represents the potential energy. So, this is really kinetic energy, this is really potential
energy and then there is a pressure work. So, its says that if you look at pressure work
kinetic energy and potential energy that sum remains constant along any point that you
along any streamline that you consider ok. So, if you consider two different points and
write down the kinetic energy plus potential energy plus pressure work, if the fluid has no
friction and if its steady state and its incompressible, then you can apply Bernoulli’s
equation and that is why this is actually an energy balance.
So, this equation that we have derived really is an energy balance. So, z 1 has the
dimension of length then the v 1 square by 2 g should also have the dimension of length
61
and you can verify that. Similarly, pressure by ρ g will also have the dimension of length
ok. So, this equation now is written in such a way that each term the such as pressure head,
velocity head all of them have the dimensions of length ok. So, we are really expressing
energy in terms of length. So, for example, when you have potential energy and you say
the potential energy is really dependent upon the height at a particular height it is, similarly
kinetic energy is expressed as a height pressure work is also expressed as a height.
Now, therefore, if you say that your fluid is actually not frictionless, but there is a friction
ok. So, if it has friction; that means, that certain amount of energy is lost due to frictional
forces because of frictional forces some energy is converted into heat. So, in other words
the action of viscosity between fluid layers resulting generation of heat ok. Typically that
number is very small and therefore, you will not worry about I mean the number need not
be small, but if you actually convert it into temperature that may be very small. So,
typically you do not consider temperature variations, but still there is an energy loss ok.
So, we can say that if you had a pressure head a velocity head and then you know there is
a potential energy and a part is lost due to friction, then you say will say there is a head
loss in the system ok. So, for a frictional fluid you can say that your equation is:
𝑃1 𝑉12 𝑃2 𝑉22
+ + 𝑧1 = + + 𝑧2 + ℎ𝑓
𝑤ℎ𝑒𝑟𝑒, ℎ𝑓 → 𝑓𝑟𝑖𝑐𝑡𝑖𝑜𝑛𝑎𝑙 ℎ𝑒𝑎𝑑 𝑙𝑜𝑠𝑠
62
So, that is energy that is lost in the system ok. So, you can see that loss would essentially
be a positive quantity because there will always be friction invariably. So, that particular
number or the particular quantity that I am comes up will always be positive.
On the other hand let us say if you are applying this equation to a pipe ok a long pipe and
then you have fluid in so, which you will call 1 and then you have 2. Then you can actually
apply Bernoulli’s equation between point 1 and point 2 and you could write down the
equation where now you have you can you have to say what is the hf or your idea most of
the times would be to find out what is hf. On the other hand let us say if you put a pump
some pump in between. So, that is a pump what does the pump do? Pump would give you
energy into the system.
So, one way of writing it is that, if you want to write you could say you have an hf then
you can say minus h pump because hf is the loss in energy pump is going to give you
energy into the system. So, you can again talk about a pump head or head thanks to the
pump ok. So, that energy that is input into the system, you can call it a pump head. So,
typically we will worry about that equation if you have a pump you know how to add it or
if you have something which is extracting energy like for example, a turbine. So, then you
have to remove that turbine head ok. So, depending upon what you have in the system,
you can keep adding or subtracting different heads. So, that is really the energy balance.
63
Fluid Mechanics
Lecture – 08
Solutions of Navier Stokes – 1
Look at some simple situations to start with so, that you all get into the flow. So, the first
thing that we are going to do today is to look at an exact Solution of Navier Stokes
equations.
So, what you have got is fluid between two parallel plates ok. So, what you have got is
two plates ok; two solid plates and then there is fluid in between ok. And then we are
saying that there is a fluid flow that is going to come from this side and it is going to enter
this channel and then go and come out of this channel ok. And let us say that the fluid that
is going to come from the left side of the screen is going to be a uniform flow, or the
velocity profile is going to be a uniform; velocity profile is uniform. So, what do I mean
by velocity profile? Can somebody say what is velocity profile?
So, velocity variation ok; so, when you want to plot velocity as a function of distance that
is what you call as velocity profile. So, when I say the velocity profile is uniform or the
velocity is uniform, the way I could represent it is that for example, I will say that the
velocity which basically says velocity is a vector. So, it has two things. It has got a
64
magnitude, it has got a direction neither of them are changing in space. So, there is a fluid
that is coming from the left with a constant velocity. So, the typical way to do it is actually
that way ok.
So, two parallel lines with there is nothing changing in between its approaching the plate.
And what would happen when this fluid touches the tip of the plate; tip of the channel?
So, I am talking about this region. So, if I think about a fluid element that is there, it is
going to come like that and touch the tip. What would you expect to happen? It would slow
down right. So, it would slow down so; that means, the velocity right there is going to go
to 0; the velocity right there is going to go to 0, but the velocity anywhere else would
remain as whatever it had come with right.
So, then so, you could say that if I take a little bit ahead in this channel. So, the fluid is
coming from this side so, if I take let us say you know a point there; this point would be at
what velocity? With the velocity would be same as the velocity with which it came or the
velocity would be smaller.
Smaller right. So, you could say that if I take this point and let us say this point the velocity
profile would look something like that. So, what do I mean by that? So, here in fact, I
should draw it a little bigger. So, that is going to be my velocity profile. So, the region
which is very close to this wall, the fluid is going to see the effect of this lower wall and
is going to be slowing down. The fluid above you know in this region is going to see the
effect of this wall and is going to slow down, but in between it may not have.
So, if I say that this is what it is then as you know it progresses this region where the
influence of the wall is seen becomes thinner and thinner ok. And finally, so this region
basically will thin, thin, thin and at some point it will disappear or in other words the even
a fluid element like this which is at the center of the channel would have seen the effect of
two walls. Now, therefore, it would develop a velocity profile from then onwards the
velocity profile would be slightly different like that.
You see what I am saying. So, the fluid which has come with a constant velocity enters
and then the fluid experiences the friction from the wall ok. And how do you quantify it in
terms of any of the fluid properties?
Student: Viscosity.
65
Viscosity ok. So, you could say that because the fluid is viscous. It has a viscosity, it is
going to be seeing the friction from the wall and it will slow down and that slowing down
is what is going to result in this particular profile. Is that clear? So, there is; so you could
say so, this is a stream of uniform velocity, this region where the fluid was developing its
velocity field ok. So, it entered like that and it finally, became like this that region is known
as entrance length ok.
So, that is the entrance length and the region from here onwards is known as fully
developed flow. Is that clear, any ∂bts? This I have drawn this line right, I did not say
anything about that. What does that line indicate? That is really a boundary of the boundary
in which the fluid is getting affected by the solid ok. There is a name for that and that is
known as boundary layer and we will talk about boundary layer in detail at a later point.
So, right now what we are going to do is we are going to look at only the fully developed
flow.
So, we will define a coordinate system, let us say that is your x direction, that is your y
direction and z is perpendicular to the screen and we will assume that your flow is right
now two dimensional. What is what do I mean by the flow being two dimensional? So,
there is no velocity perpendicular to the screen that is in the z direction and there is no
variation in the z direction. So, what do I mean by that? v is that the z component of
velocity is 0, there is nothing that is going to change in the z direction ok.
So, I do not care about my z direction anymore I am going to restrict all my analysis in
two dimensions, I only have an x component and the y component alright. So, now, the
question is if I want to find out let us say this profile, how do I proceed and you also you
told me that you know that you have to solve Navier Stokes equations to find out what this
velocity profile going to be. So, let us just do that. So, Navier Stokes equations consists of
two parts ok. The first is about mass conservation which is what you called as continuity
equation on that day ok; the second is about a force balance ok.
So, you have two components; one is mass conservation and then you have a force balance.
So, the mass conservation ok; you know that it is one of the fundamental principles and
then whatever flow whatever mechanics that you are going to talk about, it should be
obeying the mass conservation. So, somebody that they said it is A1 V1=A2 V2 right so;
that means, whatever fluid is going to enter your domain has to get out of the domain, there
66
is no escape unless there is something else going on like you know some reaction some
fission fusion and so, on the mass conservation is always going to be obeyed ok.
So, it is a one fundamental principle and it has to be obeyed by the fluid flow. So, whatever
you do, you have to respect that it is going to obey the mass conservation. Second is that
you are going to talk about something that is moving. Why does it move? Why would the
fluid that is coming from this side is going to move to the other side? Because you must
have applied some force. Unless you have applied a force, it is not going to move right
and the simple way you would have known that is equal in terms of Newton’s second law.
What is Newton’s second law? Force is equal to mass into acceleration that is all what we
write in the force balance.
So, we are saying that the fluid is you know obeying its mass conservation, one of the
fundamental principles that it has to respect, second is that it is obeying Newton’s law
which is force is equal to mass into acceleration. And all that you have done in your
continuum mechanics is you have written force is equal to mass into acceleration in terms
of a differential equation. So, force is equal to mass into acceleration was really a particle
and you are applying a force. And if the particle had a mass m, you would have written
that force as:
𝑑2 𝑥
𝐹 = 𝑚𝑎 = 𝑚. 𝑑𝑡 2
Now, when we talk about fluid, the fluid is there everywhere in the domain. We can’t talk
about such particles or otherwise, we need to do this force is equal to mass into acceleration
for every particle in the fluid and things might be changing in x direction; things might be
changing in y direction. So, by writing the force balance in a differential form, you have
really generalized this equation. So, all you are doing is mass conservation and force
balance. Is that clear? So, now, let us write down the equations in the differential form ok.
67
So, continuity equation all of you wrote it down the other day. So, let us write it again
continuity equation and as we said, we are going to deal with two dimensions.
So, we will only worry about the x component and the y component. Let us say u and v
are the x component of velocity and the y component of velocity respectively, then
continuity equation says:
⃗ =0
𝐶𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛: ∇. 𝑉
𝜕𝑢 𝜕𝑣
+ =0
𝜕𝑥 𝜕𝑦
So, that day somebody interpreted divergence as you know something a net, you know
flux of something through ok. So, the fact that there is no net flux is what is giving rise to
divergence of u being 0 so; that means, we are assuming something and we are assuming
that what property is constant.
Student: density.
Density is constant so; that means, we have already assumed that we are going to look at
the constant density fluid which you also call as incompressible fluid.
So; that means, in general if you want to talk about a system in which the density is
changing; for example, you want to model a reactor and then the reaction is happening so;
68
that means, at every point in space the density is going to change, then you have to worry
about a general equation or it could be that the temperature is varying. Let us say you are
talking about a heat exchanger and you are looking at a heat the fluid which is going
through one of the tubes in heat exchanger. The temperature is going to change along the
length of the tube; that means, the density is going to change and you have to worry about
the general form of this equation which of course, you know what it was.
So, right now we are restricting ourselves to a simple situation. So, we have ∇. 𝑢 is equal
to 0 so, that is the case. Now, we are only going to look at this fully developed part ok, we
are not going to right now worry about what is happening in the entrance length. So, by
definition fully developed flow is something in which things are not changing in the x
direction. All of you agree with that that is by definition. So, we are going to
incompressible flow and fully developed.
𝜕𝑢
𝐹𝑢𝑙𝑙𝑦 𝑑𝑒𝑣𝑒𝑙𝑜𝑝𝑒𝑑 𝑓𝑙𝑜𝑤: =0
𝜕𝑥
So from my continuity equation:
𝜕𝑣
= 0 → 𝑣 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝜕𝑦
What is v? The velocity of the fluid perpendicular to the plates right. So, the velocity of a
fluid any fluid element that is perpendicular to the plate ok; now and we are saying that is
a constant. So, does; that means, that if I talk about if the fluid element let say fluid element
here, the velocity in this direction is what we called as v and if that is constant everywhere
in the domain, then what should that constant be? Why should that be 0? Because there is
nothing that can come from the wall to outside.
So, any fluid element that you are going to look at on the wall should also have that
constant velocity and we know that that velocity is 0 at this wall. We know that velocity
is 0 at that wall; that means, the velocity there cannot be a velocity that is normal or that
is normal to the walls ok. So, what is that boundary condition? So, that is really a boundary
condition and that boundary condition is called.
69
So, the fact that v at walls is equal to 0 is called no penetration boundary condition. So,
this would change, let us say if you were talking about a porous plate or let us say you
want to do a filtration setup and you want to talk about fluid that is flowing on a you know
a membrane, ok. Then there will be things that are coming from one side to another and
then this condition will break down. So, why do you need a boundary condition? Because
we are solving differential equations which will be when where and you need to so, you
will be ending up with constants; just like this constant that we ended up with. And we can
find those constants with the help of boundary conditions ok.
Now, there will therefore, be a boundary condition about u also right. What is the, what is
the value of u? So, what is u? The component of velocity parallel to the wall. What would
that be? So, why is that 0?
So, any fluid element that you are going to again look here; so, if you look at this fluid
element the fluid element will essentially be sticking to this wall that fluid element is not
going to go anywhere, you know that that is a physical situation. So, this is no slip
boundary condition is just a mathematical representation of saying that u the component
of velocity parallel to the wall is 0 and you call it no slip boundary condition. Alright, any
questions?
𝑣𝑤𝑎𝑙𝑙 = 0 (𝑁𝑜 𝑝𝑒𝑛𝑒𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝐵𝐶)
70
𝑢𝑤𝑎𝑙𝑙 = 0 (𝑁𝑜 𝑠𝑙𝑖𝑝 𝐵𝐶)
So, we have now learned something that the velocity normal to the channel is 0. So; that
means, there is only one particular component of velocity that we need to worry about and
that is the u velocity ok. There is just one component and of course, that is what we plotted
also, that there is only one component of velocity that is parallel to the channel wall and
we want to now find out that and to find out that. So, this much we have learned only from
the mass conservation or the continuity equation.
So, now, let us take you know utilize the force balance and see what more we can get. So,
we are going to write.
Momentum equations:
𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑃 𝜕 2𝑢 𝜕 2𝑢
𝑥 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ +𝑢 +𝑣 ]=− + 𝜇 [ 2 + 2]
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦
𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑃 𝜕 2𝑣 𝜕 2𝑣
𝑦 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ +𝑢 +𝑣 ]=− + 𝜇 [ 2 + 2]
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑦
So, we know that v is 0. So, that makes our y momentum equation very simple almost
everything goes away except pressure because we have not said anything about the
pressure. And the reason why we did not say anything about the pressure is because we
71
did not say anything about the force so far because pressure is nothing, but force per unit
area right. So, it will come only in a force balance, all our analysis was about continuity
equation that is why pressure never came into the picture. So,
𝜕𝑃
= 0 → 𝑃 = 𝑃(𝑥)
𝜕𝑦
So, in our channel pressure is not going to change in that direction, pressure can change in
this direction. So, pressure is a function of x; pressure is not a function of y. So, like
pressure. so, pressure is not a function of y ok. So, that is done. Now so, that was the use
of writing down the y momentum equation even though the v component itself was 0 still
you should have analyzed it because it is not just about velocity, it is also about knowing
pressure. So, when we say that you know you want to solve a flow problem, the thing is
that you are actually trying to find out both velocity field and pressure field ok.
So, when you say you want to talk about fluid flow in some you know, in some domain in
some geometry anywhere and you want to talk about whether you know everything about
the flow. The description becomes complete when you talk about both pressure and
velocity, because both pressure and velocity can change with the space and time. Was there
a question? Any questions? So, yeah so, we have now figured out that pressure is only a
function of x. Now let us see what does x momentum equation tell you. So, we have a fully
developed flow right, that is what we said. So, would u be function of time?
No, u is not a function of time. How about u do u by ∂ x? It is 0 because ∂ u by ∂ x is 0, u

is not changing with x v ∂ u by ∂ y. It is 0 no because v is 0 ok. So, v is 0. So, that is also
gone and actually the simplification that you see here is very important because if you look
at this entire equation, these are the terms that are like product terms, u into u its like u
square this is like u into v ok. These terms are really non-linear terms and you know how
difficult it gets when you have to solve non-linear equation. Do you know any method of
solving non-linear differential equations? Everything that you would have learned would
be about linear differential equations ok.
So, you can use all of them when you have so, we you really have governing equations as
linear differential equations ok. So, whenever you have a system in which these non-linear
terms go away, you can actually expect to solve things analytically otherwise it becomes
difficult ok. So, this problem is a nice problem because this has already thrown away all
72
the non-linear terms. And therefore, you can expect to get linear equations and therefore,
you should be able to know you are you should be able to utilize your mathematics to solve
the system. Is that clear? Ok. So, that is there, ∂ p by ∂ x anyway we do not know anything,
but we know that P is a function of x. So, we have that then d square u by dx square 0. So,
that is also gone d square u by dy square, it is non-zero and that is what externally we want
to find out.
So, we find that:
𝑑 2 𝑢 𝑑𝑃
𝜇 2=
𝑑𝑦 𝑑𝑥
Now, if you look at this so, what does this tell you? Can you say what this could be?
Anything more? Can this be a function of so, if I write this f should I write it as a function
of x or y? Or x? So, this part says that u is not a function of x, u is only a function of y.
This part says that P is only a function of x, not a function of y. So, the most I can expect
is this is a function of x, this is a function of y; a function of x can be equal to a function
of y only if that is a constant. So, there is nothing possible like f of x comma y, the best
that you can say is that this is going to be a constant in this system. Agreed?
𝑑 2 𝑢 𝑑𝑃
𝜇 2= = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑑𝑦 𝑑𝑥
So, that is going to be useful because now I can integrate this equation ok. Now what is it
saying? dP/dx is a constant so; that means, if I go back to our original system ok, we are
looking at regions from here onwards ok, that is the region of interest; in that region dP/dx
is a constant ok. So, P is going to change along the x direction and it’s only going into
whatever variation is there, that variation is going to be constant. Which side will be
pressure more? Pressure will be more on this side or that side? Pressure will be more on
this side because that is why fluid is moving.
So, you have applied pressure on this side and then that fluid in between experiences that
pressure and is going to go there. So, this is where pressure is going to be larger inlet
pressure is going to be larger than the outlet pressure and it is just going to vary linearly.
In fact, because dP/dx is a constant; that means, P is a linear function of x, P will just
change linearly with x. Clear?
73
Fluid and particle Mechanics
Lecture – 09
Solutions of Navier Stokes - 2
So; that means, now I what I am interested in is now to find out. So, we have already
figured out now what pressure is, we understand a lot about pressure. We still do not know
enough about velocity and we can figure that out if we integrate this equation because this
is a is this now a partial differential equation or a ordinary differential equation. Its an
ODE you know how to solve it, you can go ahead and solve it. So, its a second derivative
and because this is a constant I can simply integrate it as:
𝑑2 𝑢 𝑑𝑃
∫𝜇 2
=∫
𝑑𝑦 𝑑𝑥
𝑑𝑢 1 𝑑𝑃
∫ =∫ 𝑦 + 𝑐1
𝑑𝑦 𝜇 𝑑𝑥
1 𝑑𝑃 𝑦 2
𝑢= + 𝑐1 𝑦 + 𝑐2
𝜇 𝑑𝑥 2
74
So, that is what my velocity field is going to be and C 1 and C 2 will be the constants that
are coming out of integration and you have to find that out from the boundary conditions.
And you know what those boundary conditions are right we already wrote it down, we had
written it down as somewhere there u at walls right so; that means, walls is either at y is
equal to 0 and let us say some y is equal to h where h will be the width of the channel now.
So, let us define all those things. So, that is my let us make another picture.
75
So, that was my original board that is what I called as x this is what I am calling as y let
us say that is equal to h. So, this wall is y is equal to 0, this is y is equal to h and when y is
equal to 0 you know u is equal to 0 at y is equal to h also that is going to be 0. Now let us
do a slightly different problem because that is that will do you know two things in one go.
So, what I am going to do now is that the same problem. So, what we have done really is
that you have applied a pressure gradient in other words you have applied a large pressure
here for example, you know you will put a pipe pump you put a pump, that will exert a
large pressure here which will move the fluid in the channel right.
So, along with that let us say that you are not only applying a pressure gradient, you are
also going to move this upper plate. So, this upper plate is going to go with a velocity
capital U what would that do? That can also drive the fluid right because you take fluid
between two plates; you are moving one of the plate the fluid will automatically move.
So, that basic basically what I am trying to say is that, there are two ways of moving fluids
typically one is that you put a you know pump, in other words you apply a pressure
gradient which can generate a flow or you move one of the boundaries like this and then
it will also drive a flow ok. So, we are looking at a situation, now where you have applied
a pressure gradient as well as you are moving a boundary ok. And then let us calculate
what is the velocity profile and see you know how that how different those velocity profiles
are going to be is that clear.
So, I am changing the gear a little bit now. So, so that is the situation. So, we know that so
far so, till writing down that equation, we only cared about the. So, we did not really
impose the boundary condition, only thing that I am doing is by imposing u is equal to you
know capital U at y is equal to h I have just changed the boundary condition. So, this
equation whatever we have derived so, far should remain ah. So, right is that clear ok. So,
I am going to use that and apply these boundary conditions can you calculate C 1 and C 2
now? So, you have a fluid which is driven by a pressure gradient whose velocity profile is
going to be this, but its also that fluid is in between the plates and one of the plate is moving
with a certain velocity capital U.
So,
𝐴𝑡 𝑦 = 0, 𝑢 = 0 → 𝑐2 = 0
76
𝐴𝑡 𝑦 = ℎ, 𝑢 = 𝑈
1 𝑑𝑃 ℎ2
𝑈= + 𝑐1 ℎ
𝜇 𝑑𝑥 2
𝑈 1 𝑑𝑃 ℎ
𝑐1 = −
ℎ 𝜇 𝑑𝑥 2
Substituting the constants of integration we have,
1 𝑑𝑃 𝑦 2 ℎ𝑦 𝑈𝑦
𝑢= ( − )+
𝜇 𝑑𝑥 2 2 ℎ
So, dP by dx will it be greater than 0 or less than 0?
Student: Less than 0.
So, you can see if you want to recognize the fact that dP by dx is less than 0 and write
down this quantity as a minus dP by dx, this term will reverse that is typically the way you
would write it. Let us see different cases:
𝑑𝑃 𝑈𝑦
𝐶𝑎𝑠𝑒 𝐼: =0→𝑢=
𝑑𝑥 ℎ
77
So, that means, I have a plate this was my y is equal to 0 and this was my y is equal to h
at y is equal to 0, u is equal to 0 here this plate is moving with a capital velocity U. So, at
cap y is equal to capital H u is indeed equal to capital U. So, it has some finite velocity and
the variation is linear so; that means, my velocity profile is going to look like that right.
And that is something that you have already seen probably and this flow is known as one
of the simplest flow fields that you can get and this is very important in terms of
determining viscosity and so on this particular flow field and you will see that probably
next week and week after how do we utilize this further.
𝑑𝑃
𝐶𝑎𝑠𝑒 𝐼𝐼: 𝑈 = 0; ≠0
𝑑𝑥
1 𝑑𝑃
𝑢= (− ) (ℎ𝑦 − 𝑦 2 )
2𝜇 𝑑𝑥
So, you have plate y is equal to 0, y is equal to h, you have applied a pressure gradient dP
by dx is less than 0 ok. So, because pressure will be larger pressure, here smaller pressure
and fluid is flowing from larger pressure to smaller pressure and the flow profile is going
to look like this, at y is equal to 0 u is 0 at y is equal to h u is again going to be 0 somewhere
in between its going to be a maximum and that maximum can only be at the center then.
So, it will look like that parabolic velocity profile. So, this flow is known as plane
poiseuille flow doubts.
78
Student: Sir, suppose we do not make a pressure difference because of a pump or
something, due to the fluid flow itself can there be pressure difference generated?
So, its the same thing. So, if you have a. So, in this case what you are asking is that you
have generated a flow field, in which case there can already be a pressure difference ok.
So, its a, so, because. So, if you look at this equation we ended up saying, so, that one ok.
It says that a pressure gradient generates a fluid flow that is what I interpreted it as or you
could say that yes there is a flow that will generate a pressure gradient and the way we
typically do it is we apply a force and get a motion we do not say that we actually get a
motion and because of that that flows existed which is an equivalent description but its
probably less physical than interpreting it the other way ok. So, flow will be typically
associated with a pressure variation or a pressure variation would generate a flow and.
Student: Sir, why did you neglect the gravity in the y momentum equation?
You could. So, right now I am looking at a pipe, that is actually perpendicular to the gravity
ok. So, the only thing that would have happen is that, there can be hydrostatic pressure that
is perpendicular to this ok. So, in reality that would mean that when we said this quantity
right dP by dy is really rho g if you wanted to include that. So, the pressure will split into
two parts, pressure will have a hydrostatic part which will balance the you know gravity
force and then there is a pressure variation in the x direction, which is going to balance the
viscous forces. Does that make sense? Because hydrostatic pressure will not generate a
flow. So, its just that you know just that fact that is why we do not care about gravity at
all.
Student: Is there gravity in the x direction?
Gravity is in the y direction.
Student: You should in the y direction you should write the g?
You can and what that will give you is that pressure will just come out to be rho gy which
will have nothing to do with what we do next ok. P is only a function of y we did not care
because I am saying that. So, read this pressure really can be interpreted as the total
pressure minus the hydrostatic pressure.
79
P is, but we know everything about what is it in y, P is just. So, see the y. So, we have a
uniform channel.. So, therefore, any point that you take if pressure is varying that is just
according to this what is interesting for us is, what is happening in that direction which is
what we did not know, which is what we calculated. So, now, you could if you like you
could expand by saying that I am going to add my rho gx term here, I will add my rho gy
term here which would give me dP by dy is equal to rho gy, where I am only talking now
my gx is 0, I only have gy which gives me P is equal to rho gy.
Student: I am telling that it won’t be constant then because this is function of y also there
in a gravity like if mu d square u by dy square is a function of y.
So, that is what I am saying that the pressure has two components now, one is due to
gravity the other is due to flow what we have calculated is the pressure due to the flow not
due to gravity ok. So, in some, so, this pressure. So, let us do it this way, let us say that
this is some let us I think this is called a dynamic pressure; that means, we have already
removed the hydrostatic part from it. So, we will call it a Pd then its ok? And that is because
gravity will not do anything, but if you were looking at a pipe which is perpendicular to
gravity, then you should have been careful because these two components will add up in
the same direction now ok.
So, we will see how to. So, see that is exactly what you do it in Bernoulli’s equation; you
have a pressure head, you have a velocity head, you can have a gravity head right if you
talk about a pipe in a you know in a horizontal direction, then there is nothing gravity will
do ok. Gravity will not generate any velocity, but if it was in the perpendicular direction
or in the direction that is parallel to gravity, then gravity could induce a flow and which
we will see how we will write down gravity included in terms of pressure field and so on
at a later point. So, at the moment let us just say that either we have a situation in which
our flow is perpendicular to gravity and therefore, the hydrostatic pressure is not going to
do anything. Anything else? I need to do two more cases.
80
So, the next case you can plot and then come back next class.
𝑑𝑃
𝐶𝑎𝑠𝑒 𝐼𝐼𝐼: 𝑈 ≠ 0; <0
𝑑𝑥
𝑑𝑃
𝐶𝑎𝑠𝑒 𝐼𝑉: 𝑈 ≠ 0; >0
𝑑𝑥
81
Fluid Mechanics
Lecture -10
Introduction to cylindrical coordinate systems
What we are going to do today is to do the same in cylindrical coordinates because finally,
we want to do a flow through a pipe and that is a very important problem. But, we should
know what are the governing equations in various coordinates systems because then we
can try to attempt various geometries ok.
So, what we did in the last class was the simplest of the case where you know you deal
with the Cartesian coordinate system and then you could therefore, solve for the flow
profile in a 2D channel and you do know that you know the governing equations look a bit
different when you are in other coordinate systems. So, let us just look at why it happens
so before we actually attempt to solve the problem in a pipe.
So, what we are going to try doing today is to find out how do we find out what is the right
governing equation in various coordinate systems. So, we will use cylindrical coordinate
system as an example because that is going to be most useful for this course.
82
So, what we have is a Cartesian coordinate system. Let us start with that because that is
the most familiar case ok. So, you have x, y, z a right handed coordinate system and any
point that you look at, you will describe that point using a x, y, z right that is actually what
we mean by Cartesian coordinates. Now, in cylindrical coordinates we want to replace it
with an r θ z ok. How does one define r? The way to define r is you project this point onto
the x y plane and this distance is what actually is known as r and how does one define θ.
Θ is essentially this angle ok. So, what so that means, that z so, things are changing so. So,
really you have z in the Cartesian coordinate system remains as the same z in the
cylindrical coordinate system instead of x and y you are going to use r and θ to describe
any point. So, r is going to tell you what is the radial distance from z-axis. So, that is a z-
axis it tells you how far radially it is from z-axis. Θ is going to tell you if you were to have
an x-axis how much rotated it would be. So, therefore, r and θ will give you the same
information as x and y give. That is clear?
So, if you look at just x y coordinates x. So, we can just since z is not changing we can
actually look at all the analysis by looking at the x y plane. So, we will just draw x y plane
that is our x that is our y. Any point that was actually described using x y we are going to
change it into r θ coordinate system. So, that is going to be our r now and that is going to
be our θ now. So, if this point was denoted using x y then what is the relation between x r
and θ?
𝑥 = 𝑟𝑐𝑜𝑠𝜃; 𝑦 = 𝑟𝑠𝑖𝑛𝜃; 𝑧 = 𝑧
Now, the important thing is to realize what happens to the unit vectors in these coordinate
systems. So, in Cartesian coordinates system, if I have a unit vector that I denote using x
cap; here you know a unit direction y cap and this is z cap any vector that I talk about I
can resolve it into components in x y z direction you all know that ok. Now what we really
need is when we have a vector we also need to resolve it in r direction θ direction and z
direction if you want to really make use of cylindrical coordinates system ok.
So, therefore, we need to define unit vectors in r θ z coordinate system which is going to
be simple because this is your r and then how what you defining a r is essentially the
distance from the z-axis to any point ok. So, r is really that ok. So, r cap if I define, r cap
is just going to be a vector ok; a unit vector which tells you know what is the direction
which your radius is going. So, it could be so, this is right now in this plane y is a plane, r
83
cap would be like this if I talk about an x z plane, r cap will come out of the plane and so
on you understand that. So, if you have an z-axis r cap could be in these directions just the
way you have radial directions going away ok.
And θ cap is basically the direction in which θ is changing. So, θ cap would be something
of that sort ok; a unit vector that is defined in the θ cap direction. So, you imagine my hand
is the z-axis ok, r cap would change like that θ cap is going to tell you how things are
changing in this direction. So, this can be so, if I am looking at this point θ cap could be a
direction if I am to talking about a point here θ cap will be this direction and so on ok.
So, r cap is the unit vector that is coming out of z θ cap is going to tell you how things are
going to change this way and z cap of course, will remain as usual along the z-axis. So,
that are those are the unit vectors in cylindrical coordinate system ok. Now, we are going
to do is we are going to connect this unit vectors in cylindrical coordinates with unit vectors
in Cartesian coordinates ok. That is going to be useful this going to give us useful relations.
So, again we can a just stick to the x y plane; so, x y the point at a distance r at an angle θ.
So, by definition that is my unit vector r cap, that is going to be my unit vector θ cap, this
is how x cap would have looked like that is how y cap would have a looked like and this
angle is going to be θ all of you agree. So, then how can I write down r cap, r cap is a unit
vector. So, if I draw finish the triangle this length is going to be just cos θ.
84
This length is going to be just sin θ. So, if I do cos θ x cap plus sin θ y cap that is just going
to give me r cap right; so,
𝑧̂ = 𝑧̂
𝑟̂ = cos 𝜃 𝑥̂ + sin 𝜃 𝑦̂
𝜃̂ = − sin 𝜃 𝑥̂ + cos 𝜃 𝑦̂
So, then now write down what is x cap and what is y cap? You can write down using the
same thing or you can solve these equation whichever way you like. You can multiply the
first equation with cos θ, multiply the second equation with sin θ add up and you will get
x cap and the other way ok.
𝑥̂ = cos 𝜃 𝑟̂ − sin 𝜃 𝜃̂
Ok and y cap.
𝑦̂ = sin 𝜃 𝑟̂ + cos 𝜃 𝜃̂
Yeah. So, that is basically the relation the inverse relation. So, now, we know given
Cartesian coordinate system how to convert it into cylindrical coordinate unit vectors and
the other way ok. Now, so these are just some relations that we need the most important
difference that make Cartesian coordinate system different from all other coordinate
systems is this. What is ∂ by ∂ x of x cap that is when I am moving along the x direction
how does my unit vector x change? Does it change? Is just does not change by magnitude
by definition and it does not change the direction so it is 0.
What is ∂ by ∂ y of x cap? If I am moving in y direction does anything happen to my unit

vector x cap; no ∂ by ∂ z of x cap again 0 and so is true with whether I use y cap or z cap.
In other words none of the unit vectors there are no gradients whatever differences I am
going to take it is all going to be 0 ok.
𝜕𝑥̂ 𝜕𝑥̂ 𝜕𝑥̂

= 0; = 0; =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
That is not true in any of the curvilinear coordinate system.
85
So, why I am saying that let us say this is my z-axis. Let us say this pen represents by r
cap ok. If I am changing in this direction my r cap direction is changing right, then I am
going like this I am going in θ direction. When I am going in θ direction; so, if I have a
point here this is the direction which I have defined my r cap if I go to a different point my
r cap has changed the direction even though it is magnitude has not changed ok.
So; that means, that if I am see if I am looking at the derivative of unit vectors that need
not to be 0 the way we have written it down for the Cartesian coordinates system. So, we
need to actually find out how unit vectors change when I go in different directions and that
is important because we talk about all differential equations, we talk about spatial
gradients, we talk about divergences, curves and gradients and so on. So, the change in
unit vectors therefore, become important that is clear. So, let us calculate that it is very
easy.
So,
𝑟̂ = cos 𝜃 𝑥̂ + sin 𝜃 𝑦̂
𝜃̂ = − sin 𝜃 𝑥̂ + cos 𝜃 𝑦̂
Now θ is not a function of r by definition θ and r are 2 independent coordinates, x cap and
y cap are constant unit vectors. So, this is basically a constant as far as r variation is
concerned and therefore, that is equal to 0 ∂ by ∂ r of r cap equal to 0 and that has to be the
86
case because if I define an r cap like this if I go in this direction the unit vector is not going
to change ok. So,
𝜕
(𝑟̂ ) = 0
𝜕𝑟
Now, let us do ∂ by ∂ θ of r cap. What is ∂ by ∂ θ of r cap?
𝜕 𝜕
(𝑟̂ ) = (cos 𝜃 𝑥̂ + sin 𝜃 𝑦̂) = − sin 𝜃 𝑥̂ + cos 𝜃 𝑦̂ = 𝜃̂
𝜕𝜃 𝜕𝜃
Also,
𝜕
(𝑟̂ ) = 0
𝜕𝑧
𝜕
(𝜃̂) = 0
𝜕𝑟
𝜕
(𝑧̂ ) = 0
𝜕𝑟
𝜕
(𝑧̂ ) = 0
𝜕𝜃
𝜕
(𝑧̂ ) = 0
𝜕𝑧
So, this is basically how the unit vectors are changing in the cylindrical coordinate system
ok. So, there are non non-trivial 2 non-trivial quantities; one is ∂ by ∂ θ of r cap another is
∂ by ∂ θ of θ cap ok. And, that can actually bring in some additional terms whenever you
talk about special gradients.
87
So, we are going to derive now gradient operator in cylindrical coordinates. In Cartesian
coordinates we have,
𝜕 𝜕 𝜕
∇= 𝑥̂ + 𝑦̂ + 𝑧̂
𝜕𝑥 𝜕𝑦 𝜕𝑧
Substituting expressions we derived earlier:
𝜕𝑟 𝜕 𝜕𝜃 𝜕 𝜕𝑟 𝜕 𝜕𝜃 𝜕 𝜕
∇= (cos 𝜃 𝑟̂ − sin 𝜃 𝜃̂) ( + ) + (sin 𝜃 𝑟̂ + cos 𝜃 𝜃̂) ( + ) + 𝑧̂
𝜕𝑥 𝜕𝑟 𝜕𝑥 𝜕𝜃 𝜕𝑦 𝜕𝑟 𝜕𝑦 𝜕𝜃 𝜕𝑧
We can write r as:
𝑟2 = 𝑥2 + 𝑦2
Now,
𝜕𝑟 𝑥
= = cos 𝜃
𝜕𝑥 𝑟
Now we need ∂ θ by ∂ x. So,
𝑦
𝜃 = tan−1
𝑥
88
𝜕𝜃 1 𝑦 𝑦 𝑦 1
= (− 2 ) =− = − 2 = − sin 𝜃
𝜕𝑥 𝑦2 𝑥 𝑥2 +𝑦 2 𝑟 𝑟
1+
𝑥2
Also,
𝜕𝜃 cos 𝜃
=
𝜕𝑦 𝑟
So, now we can substitute and we can simplify:
𝜕 sin 𝜃 𝜕
∇= (cos 𝜃 𝑟̂ − sin 𝜃 𝜃̂) (cos 𝜃 + )
𝜕𝑟 𝑟 𝜕𝜃
𝜕 cos 𝜃 𝜕 𝜕
+ (sin 𝜃 𝑟̂ + cos 𝜃 𝜃̂) (sin 𝜃 + ) + 𝑧̂
𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧
𝜕 𝑟̂ cos 𝜃 𝜕 𝜕 𝜃̂ sin2 𝜃 𝜕 𝜕
2
∇= cos 𝜃 𝑟̂ − ̂
− sin 𝜃 cos 𝜃 𝜃 + + 𝑟̂ sin2 𝜃
𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑟
𝜕 𝑟̂ sin 𝜃 cos 𝜃 𝜕 𝜃̂ cos 𝜃 𝜕
2
𝜕
+ 𝑟̂ cos 𝜃 sin 𝜃 + + + 𝑧̂
𝜕𝑟 𝑟 𝜕𝜃 𝑟 𝜕𝜃 𝜕𝑧
𝜕 𝜃̂ 𝜕 𝜕
∇= 𝑟̂ + + 𝑧̂
89
Fluid Mechanics
Lecture – 11
Continuity equation in cylindrical coordinates
So next is Continuity equation in cylindrical coordinates ok. So, we have what is continuity
equation in general? Is that ok. So, what we have done now we just derived the gradient
operator in cylindrical coordinates ok.
And, you can see that it came out to be slightly different than what you would have
expected. You would have thought that you would first write r cap del by del r plus theta
cap by delta theta plus z cap del by del z, that is the instinct that we would have, but that
is not correct you have an extra r that should come and that will come because if you look
at actually the units this theta is basically an angle it is in radians ok. It has no dimensions
of length ok. So, that is why you have an extra r that is coming in to make this whole thing
dimensionally consistent ok.
90
Both r and z have dimensions of distance ok. They basically talk about some length while
theta is in terms of radiance and r theta is the one that basically makes it a length that is
why you have that extra r coming in cylindrical coordinates. If you go to spherical
coordinates then you would have more problems because you have two angles there and
each of them will do in a different way you know you will have r combined with theta in
different forms you might see there some r sin theta might come, r cos theta might come
and so on. So that is not right.
So, considering that we have gotten this it is very easy to actually do it in cylindrical
coordinates now because the velocity; so, what is velocity now? So, you have any point
velocity you can write as u r r cap plus u theta theta cap plus u z z cap where u_r, u_ theta
u_z are the three components in the cylindrical coordinate systems.
So, what do I mean by that?
91
Like if you go back to the original coordinate system that we started with x, y, z if I am
looking at a point there and if I am defining my velocity I would have resolved it in x, y,
z if I were doing the Cartesian coordinate system. Instead of that now I will do the u r u
theta that is perpendicular to the plane and a u z that is you know along the z-axis. So, if I
have a velocity here which is flowing in this direction I will resolve it in r direction, I will
resolve it in theta direction and I will resolve it in z direction, that is all and each component
is what I wrote it as u r u theta u z ok.
So, now, we can easily do continuity equation, if you start by the general form which is
divergence of u is equal to 0 right, that is the continuity equation and when it is written in
this form this does not care about what coordinate system you have used actually right
because there is no information about the coordinate system in this way when you write it.
So, that is the advantage of writing this equations in this particular form ok. In this form it
is independent of the coordinate system.
So, we can now use our expression for divergence and u in cylindrical coordinate system.
So, we just said this is the gradient operator remember it is the gradient operator in
cylindrical coordinate system. So, that is r cap del by del r plus theta cap divided by r del
by del theta plus z cap del by del z dotted with u r r cap plus u theta theta cap plus u z z
cap ok.
92
So, we have this is really two vectors and dot product ok, but in you have to remember
that you cannot just do this times that because this is actually differential operator ok. So,
you have to operate each of these terms.
So, for example, this term has to operate on this, this and this and so on and you should be
writing down all such operations. In Cartesian coordinates when you do it only few of
them remain which will be just you know the simpler du x by dx du y by dy and so on, but
that will not be the case when you do it here. So, let us do it and see that it is different. So,
that is what we want to do.
Is equal to let us start this I am going to operate on all of them r cap ok. Now, what is this?
r cap del by del r dotted with this thing now. What is the dot operation between? The dot
operation is between this r cap and this r cap ok. So, r cap dotted with del by del r of u r r
cap plus let us do each of them r cap del by del r cap dotted with del by del r of u theta
theta cap plus r cap dotted with del by del r of u z z cap plus theta cap divided by r del by
del theta. So, theta cap dotted with del by del theta of u r r cap plus theta cap by r dotted
with del by del theta of u theta theta cap plus theta cap by r dotted with del by del theta of
u z z cap plus z cap del by del z dotted with u r r cap plus z cap del by del z is dotted with
u theta theta cap plus z cap del by del z dotted with u z z cap right.
I just expand it wrote down all the terms and let us look at each of the terms now. Let us
start with the last term, z cap del by del z of u z is z cap. Now, I am doing derivative of
93
this quantity with respect to z. So, as far as this quantity is concerned only u z can depend
on z z cap is a unit vector. So, this is really z cap dotted with z cap del u z divided by del
z, but z dot dot z cap dot z cap is unity and therefore, what you basically have is del u z
divided by del z that is only term that is coming from the last term.
How about z cap dot del by del z of u theta theta cap? So, you have to do del by del z of u
z u theta theta cap. Does theta cap depend on z? no. So, I can pull out theta cap outside
which is going to give me z cap dot theta cap del u theta divided by del z, but what is z cap
dot theta cap? They are orthogonal. So, they will go away ok. How about this term z cap
dot del by del z of u r r cap? Does r cap depend on z? No. So, you can pull out r cap that
is going to give you an z cap dot r cap which will be 0. So, that will go away.
So, you can see that out of all the z derivatives only dou u z by dou z remained. Agreed?
Now, let us look at the next term. So, let us look at this. So, this is del by del theta of z
cap. Does z cap depend on theta? No, you can pull that out, but that will make it 0. So, that
term will go away. Does theta cap depend on theta? Yes? No? Yes. So, that means, you
need to apply the product rule there to differentiate. So, we have theta cap divided by r
dotted with del u theta divided by del theta times theta cap plus u theta times del theta cap
divided by del theta.
What is del theta cap by del theta? Minus r cap ok; so this r cap so, if I take dot product
of theta cap with r cap that is going to go away. Only this remains I have a theta cap dot
theta cap that is 1 and therefore, what is left out is 1 by r dou u theta divided by dou theta,
clear? How about this term? r cap is independent of theta. So, I can pull that out so I get
theta cap dot r cap that will go away. This term I can pull out z cap. So, that will go away
this I need to be careful because theta cap is dependent on r and I should do a product rule.
So, I have to do r cap dotted with dou u theta divided by dou r into theta cap plus u theta
into dou theta cap divided by dou r.
So, this one r cap dot theta cap that will go. What is dou theta cap by dou r? What is dou
theta cap by dou r? Oh yeah it is with respect to r. So, that is again 0 k. So, there is nothing
to worry about. This one what is here again r cap is r cap dependent on r? No, r cap is
independent of r so, you can pulled out r. So, that means, this will just leave you dou u r
by dou r. So, you get that term, you get that term I have made a mistake.
Where did I make the mistake?
94
The first term or the second term? Second term it should be. Here right? So, that cannot
be right because r cap is dependent on theta, theta cap divided by r times del u r by del
theta times r cap plus u r del r cap divided by del theta. So, this is 0, but del r by del theta
is theta cap. So, that is theta cap and therefore, that gives me theta cap dot theta cap
basically gives you 1 by r u r. So, I will write down all the terms together that I have gotten.
I have gotten that del u r divided by del r plus I have this term 1 by r u r because I have
that term 1 by r dou u theta by dou theta plus dou u z by dou z is equal to 0. So, that is
what we get as the continuity equation in cylindrical coordinates.
Often it is written as 1 by r del by del r of r u r plus 1 by r dou u theta by dou theta plus
dou u z by dou z is equal to 0, because if you differentiate this you are going to get these
two terms ok.
So, that is where that extra you know r factor and r dependence comes from ok. This will
be useful exercise and once you do that then you will be clear what when you get the next
expression how to transform between different coordinate systems.
So, well, this is one way of doing it, there are multiple ways of doing it, but this is
something probably that something that you are familiar with because you are familiar
with equations in Cartesian coordinate system and given that you will be able to convert it
95
into any coordinate system that you want. So, you can imagine how it would get a bit
lengthier if you want to do the momentum conservation equation. So, let us just take it for
granted that we will be able to do it because the procedure is simpler, it is only the algebra
that is lengthier ok. So, we will just take those equations and try to go ahead with the
solving pipe flow problem in the next class that is tomorrow.
96
Lecture – 12
Solutions of Navier Stokes in the cylindrical ordinate system - 1
So, what we are going to do today is pipe flow.
In terms of calculations the calculations are very simple straightforward you would have
also probably seen it, but the concepts that get introduced during this calculation they are
very important and that is because it is the same concept that will that we will keep talking
again and again in various contexts throughout the course and probably throughout the you
know your degree here ok. So, that is the importance of pipe flow.
So, the calculations essentially are very similar to what we did in the 2D channel ok. You
basically want to find out how you can describe flow in a cylindrical pipe ok. So, when I
say pipe flow it could be actually any cross section and when you go to a real system ok.
For example, you go to an industry you need not always find a cylindrical pipe you might
find a square pipe rectangular pipe and so on.
In fact, you could ask the question what is the pipe that I should choose when I have an;
when I have an application. When I want to let say transport things from one point to
97
another let us say crude oil or some chemical what should be the shape that you should be
choosing? At the moment you do not know and we probably will not talk much about it,
but we will see what various shapes would mean that will come at a later part, but this
forms a basic you know example.
So, you have a pipe and there is a fluid that is going in and coming out and in order to
attain that what you can do is you can apply a pressure gradient. You will apply a pressure
p plus ∂ta p on this side and this side can be pressure and if as long as ∂ta p is greater than
0 fluid is to go in to go from one side to another and again we are going to look at fully
developed flow.
So, as we said that day when the moment the fluid enters here there will be an entrance
region where the flow is developing; that means, if I am going along this axis along the
pipe axis I would see that at every point the flow is changing, but beyond a point flow does
not change right and that is what we termed as fully developed flow. So, fully develop
flow is that flow which is coming after the entrance length and we are basically dealing
with fully develop flow because entrance length may not be so long while we might be
talking about pipes which are like you know kilometers long then we do not really care
what happens at the initial portion. So, let us as well neglect that initial portion and work
only with the fully developed flow for the time being and let us use now which coordinate
system would we choose for this problem? Why would we choose cylindrical coordinate
system?
Because it is a cylinder. So, let us say now I want to solve flow in that Cartesian ok. So,
what decides what you know how should we choose the coordinate system? We will see
that I will answer the question once we go along. So, let us choose the cylindrical
coordinate system. We will say that that is our z axis that is our r and then θ is going to be
you know perpendicular to the plane and around the z axis.
We can, so, what is the first thing to do you will write down the governing equations which
is continuity and momentum equation. We will start with the continuity equation and we
will make few more assumptions. So, we derived continuity equation yesterday what was
that? I hope you will tell. Are you going to tell?
No. I am going to write it down in cylindrical coordinate system.
98
1 𝜕(𝑟𝑢𝑟 ) 1 𝜕𝑢𝜃 𝜕𝑢𝑧
𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑖𝑡𝑦 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛: + + =0
𝑟 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧
This one what you see on the screen. So, that is the continuity equation. So, you have u r
u θ u z. So, these are the three components of velocity as we described yesterday. So, we
have fluid which is going from one side of the pipe to the other ok. So, the component of
velocity that we are interested in is u z ok. So, we really want to know what is u z and if
you remember what we did in the 2D channel what was the component of velocity that we
had. We had a component which is parallel to the channel axis we did not have a
component that is perpendicular to it. So, here also it is unlikely that we will have a u r ok;
u r is likely to be 0 and we will right now assume that u r is indeed 0 you can question that
assumption, but at a later point.
So, let us assume that u r is equal to 0 we will also assume that u θ is equal to 0 ok.
𝑢𝑟 = 0; 𝑢𝜃 = 0
Now assuming u θ equal to 0 is much more sensible because what is u θ? U θ is essentially.

So, let us look at a cross section of this pipe. Let us say that is a cross section ok. So, flow
is coming from behind the screen through the screen to outside and let us say you know
you have an x, you have a y and that is what you define as θ right. So, then I could say that
my u θ which will come out to be some function of θ if I say I could also draw my
coordinate system in that fashion right that is also an acceptable coordinate system in
which I would define my θ that way right.
So, the θ that I would have defined in my original coordinate system and the θ that I will
define in my new coordinate system would have been different ok. So, it is unlikely that
therefore, there can be u θ dependence because you would not be able to find out a
mathematical expression that will do this, that any coordinate that you choose whether you
know you choose this x o, y o, x n, y n whatever coordinate system you choose there will
not be any mathematical function that is going to tell you that you know you are going to
get the same number ok.
So, it is unlikely that u θ will be coming out to it will come out as a non-zero quantity.
That is one reason to a priori assume that I can actually start by saying that u θ is 0. Is that
99
clear ok? So, even if it is not it is I mean let us just assume that u θ is equal to 0. We will
also assume that ∂ by ∂ θ of anything is equal to 0.
𝜕
(𝐴𝑛𝑦 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒) = 0
𝜕𝜃
What is meant by that? As I go in the θ direction. So, as I am going in this direction again
I am assuming that there is nothing that is changing ok.
So, that is, so, imagine, so, you exactly have a circular pipe ok. So, why should any point
be different than any other point that is the idea. So, therefore, you are going to assume
that ∂ by ∂ θ of anything is also going to be 0. So, these two assumptions are known as
axisymmetric. So, we are essentially assuming that flow is axisymmetric. So, flow is
axisymmetric. So, basically meaning that around the axis around the z axis flow is
symmetric that is the assumption that we have gone in or we have taken and we are only
looking for the u z component. The u z component is typically called the axial velocity
that is velocity along the axis.
You can, so, you cannot assume both of them to be 0 as a consequence of a fully developed
flow. You can take one of them another can be shown that fully developed flow is
consistent with this approximation. So, let me just show you. So, let us say if you take u r
is equal to 0 u θ equal to 0 the continuity equation says that is gone, that is gone and only
this is left which is whatever fully developed flow is ok.
So, it is consist. So, only thing I am saying is that fully developed flow and these two
assumptions are consistent ok. Now on the other hand if you wanted to start by saying that
I have fully developed flow. So, ∂ u z by ∂ z is equal to 0 then suppose you started with
continuity equation and you could have said ∂ u z by ∂ z is equal to 0. So, this I am going
to put to 0 then you have to either assume u r equal to 0 or u θ equal to 0 to get the other
is equal to 0.
So, if I say u r is equal to 0, I will pull push that out then I have ∂ u θ by ∂ θ which will
come out to be a constant ok, but it will be 0 at the wall and therefore, it will be 0
everywhere.
So, by definition this is our definition of fully developed flow. If u z is changing with z
then it is not called fully develop flow ok. For example, the flow that is coming from here
100
to here things are changing in z direction. So, you can say the flow is actually speeding up
or slowing down that is speeding up is in space as well it could be in time also, but it is.
So, it is accelerating in space. So, wherever it is accelerating or decelerating you will not
call it a fully developed. It is a developing flow against a fully developed flow ok, alright.
So, those are the assumptions. Now we can just go ahead write down our you know force
balances throw out all the unnecessary terms and then see what is left out. Any ∂bts
otherwise on this? So, this one thing. So, we have ∂ u z by ∂ z is equal to 0; that means, u
z is a function of r and θ, but because of the assumption of axis symmetry by definition
things are not dependent on θ.
𝜕𝑢𝑧
= 0; 𝑢𝑧 = 𝑓(𝑟, 𝜃) = 𝑓(𝑟)[𝑎𝑥𝑖𝑠𝑦𝑚𝑚𝑒𝑡𝑟𝑖𝑐 𝑎𝑠𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛]
𝜕𝑧
So, we do not have we have just u z as a function of r. Just the way in the other problem
we had a velocity nor parallel to the channel is just a function of you know perpendicular
direction. So, similarly here velocity is along the pipe axis and it will vary only in the radial
direction and that is what we are gonna derive.
So, let me write down the equations r momentum equation.
𝜕𝑢𝑟 1 𝜕𝑃 𝑢𝑟 2 𝜕𝑢𝜃
𝑟 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ ⃗ . ∇𝑢𝑟 − 𝑢𝜃2 ] = −
+𝑢 + 𝜇 [∇2 𝑢𝑟 − 2 − 2 ]
𝜕𝑡 𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝜃
101
So, now, we should not be surprising to you why we get all these extra terms right. So, this
right hand side is actually ∇2 𝑢 right it is second derivative of u. So, we already saw what
divergence did just second derivative. So, there are many more differentiations. All those
unit vectors are going to contribute to many things. That is why you get all these extra
terms here and some of the extra terms there, alright ok.
𝜕𝑢𝜃 1 𝜕𝑃 𝑢𝜃 2 𝜕𝑢𝑟
𝜃 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ +𝑢
⃗ . ∇𝑢𝜃 − 𝑢𝑟 𝑢𝜃 ] = − + 𝜇 [∇2 𝑢𝜃 − 2 − 2 ]
𝜕𝑡 𝑟 𝜕𝜃 𝑟 𝑟 𝜕𝜃
𝜕𝑢𝑧 𝜕𝑃
𝑧 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ +𝑢
⃗ . ∇𝑢𝑧 ] = − + 𝜇[∇2 𝑢𝑧 ]
𝜕𝑡 𝜕𝑧
𝜕 𝑢𝜃 𝜕 𝜕
𝑢
⃗ . ∇= 𝑢𝑟 + +
2
1𝜕 𝜕 1 𝜕2 𝜕2
∇ = (𝑟 ) + 2 2 + 2
𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧
From first two equations, we see that
𝑃 ≠ 𝑃(𝑟, 𝜃)
𝑃 = 𝑃(𝑧)
That is going to be nothing, but the pressure force that you have applied ok. Pressure is
high on one side and pressure is low on the other sides just exactly that. So, p is only a
function of a z.
On simplification, the final differential equation that we obtain is:
𝜕𝑃 1 𝜕 𝜕𝑢𝑧
− +𝜇 (𝑟 )=0
𝜕𝑧 𝑟 𝜕𝑟 𝜕𝑟
What kind of a this is a partial differential equation or an ordinary differential equation

that is the question? So, we said p is not a function of r and θ. So, I can write this ∂ p by ∂
z as really d p by d z and therefore, u z is only a function of r. So, that is again an ordinary
derivative ok. So, it is no more a partial differential equation it is an ordinary differential
equation and you know how to solve it.
102
Lecture – 13
Solutions of Navier Stokes in the cylindrical coordinate system – 2
Before solving it let us do one thing, let us do something called non-dimensionalization.

So, you have done non dimensionalization in the last semester’s course, no you heard about
non dimensional numbers, Reynolds number that is a non dimensional number right. So,
we are going to see this the next part in which we will see how to do non
dimensionalization, what is meant by non dimensionalization, but at this point itself we
can actually start doing that and I think it is worth doing it at this point. So, before solving
let us do something called non dimensionalization of equations.
So, what is meant by non dimensionalization? You are going to make all the variables non
dimensional. So, for example, what is the dimensions of velocity? It is just length per time,
you want to now make it a non dimensional quantity; that means, you get rid of the units
from that that is what is meant by non dimensionalization ok. So, we will actually non
dimensionalize each term. So, how do we do that, how can we get rid of you know
dimensions of velocity?
103
You have to divide it with another quantity which has the units of velocity ok. So, that is
one thing what is the other quantity that we need to non dimensionalize length r ok, radius
see what are we expecting to see we are expecting to see u z as a function of r that is our
objective ok. So, we want to find we want to non dimensionalize our velocity, we want to
non dimensionalize our length.
So, now we have to do that we need to find out what can be used to non dimensionalize
velocity first or okay let’s non dimensionalize length first, what can be used to non
dimensionalize my length. So, what we need to do is to choose something called a
characteristic length ok, some constant length from the problem. So, one can you think
about some constant length that we have in the problem right now.
Radius ok, it could be radius or it could be diameter depending upon whichever you want
to do we can do that let us just choose radius. So, let us choose characteristic length as
radius of the pipe. Then what we do, we will define a non dimensional radius
𝑟
𝑁𝑜𝑛 − 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑎𝑙 𝑟𝑎𝑑𝑖𝑢𝑠: 𝑟 ∗ =
𝑅
So, this r, small r was our original radius which had the dimensions of length we have
divided it with the pipe radius and we have gotten an r star which would now vary from?
0 to 1 that is a good thing about another good thing about non dimensionalization, you
have gotten rid of the length and you would basically say that things are going to change
between 0 and 1. If you had taken diameter you would have found that it would go between
0 and half that is all ok, but it is always going to go in that range it is not going to go from
0 to 1000 or 10000 to some 100000 ok, that that does not come out it is basically going to
be a number which is going to be of the order of 1, when I say also order of 1 it is around
1.
104
So, that so, radius is done, the other one which we need is velocity. So, we need to get
something called a characteristic velocity, what can be a characteristic velocity pipe?
Maximum velocity is a possibility, but we do not know what it is at the moment, but let us
say that let us just define a velocity say U which let it be the maximum velocity where
does the maximum velocity appear? At the center ok. So, we will use U which we do not
know, but let us just take it. So, and therefore, I can define my u z star the non dimensional
velocity
𝑢
𝑁𝑜𝑛 − 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛𝑎𝑙 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦: 𝑢𝑧 =
𝑈
Now can you substitute this into your differential equation:
𝑑𝑃 1 𝑑 ∗
𝑑
=𝜇 ∗ (𝑟 𝑅 (𝑢∗ 𝑈))
𝑑𝑧 𝑟 𝑅 𝑑(𝑟 ∗ 𝑅) 𝑑(𝑟 ∗ 𝑅) 𝑧
𝑅 2 𝑑𝑃 1 𝑑 ∗
𝑑
= ∗ ∗
(𝑟 ∗
(𝑢𝑧∗ ))
𝜇𝑈 𝑑𝑧 𝑟 𝑑𝑟 𝑑𝑟
So, you look at this equation now. So, this tells a whole lot of thing ok, the right hand side
r by μ u d p by d z. So, you have this pipe, you have applied a d p by d z you have applied
it, it has got this radius R the fluid has a viscosity μ and it has a velocity maximum velocity
105
U, which must be related to the flow rate ok. So, this quantity, so, what will be the
dimensions of this quantity? What would be the dimensions on the quantity on the left
hand side, what is the dimension of r star?
Nothing what is the dimension of u z star? So, what is the dimension on the left hand side?
Nothing. So, the dimensions on the right hand side also be nothing. So, the right hand side
is actually a non dimensional number ok, it is a non dimensional number that tells you how
the flow is going to be it is ok. So, the entire characteristic of the flow determined by that
non dimensional number. So, that non dimensional number is nothing, but R by 𝜇 u d p by
d z ok, if you say that number then the entire flow can be characterized in other words,
you know u double r and then you change your you double your u also in some way then
the right hand side does not change; that means, the left hand side does not change, the
solution that you are going to get will have just this thing as a parameter.
So, if you it is like this. So, you have heard about parametric equations right, x is equal to
r cos theta y is equal to r c sine theta is one of the parametric equation. So, what do we
mean by that? So, it just says that, this is a differential equation this is an ordinary
differential equation ok, there is only 1 constant in that entire ordinary differential equation
which is this right hand side, which is a unit less constant it tells you everything about the
flow. Originally, when we had we did not know this.
106
So, here if you look at the original equation that was our original equation we had a
pressure gradient, we had a viscosity and that is it and each of them had it is own units and
so on ok, but now we have gotten rid of all that and we said oh look at our entire equation,
the entire equation has become non dimensional.
And we have also figured out, if you want to change flow in some fashion this is the
quantity that we should look at nothing else we do not have to really care about anything
else because we have. So, it is so, in fact, this is going to be related to Reynolds number,
we will see that at a later point maybe we will see. So, we you get a little more idea about
this when we substitute for u which we will do at a later point. So, this is the non
dimensional version of the equation. So, we can go ahead and solve our original equation,
which is that can you solve it and tell me what is the solution? You have to do 2 integrations
and the expression will be:
1 𝑑𝑃 𝑟 2
𝑢𝑧 = + 𝑐1 ln 𝑟 + 𝑐2
2𝜇 𝑑𝑧 2
So, now we have 2 constants c 1 and c 2 and we need to calculate what is c 1 and c 2 right
ok? So, what do we need for that boundary conditions. At the pipe wall:
𝐴𝑡 𝑟 = 𝑅, 𝑢𝑧 = 0
107
And I think, that is sufficient because we can already throw away one another term from
this. So, how do we know anything about at r is equal to 0 what is u z.
So, but what about this term c 1 ln r, at r is equal to 0 what is log 0? C 1 must be 0,
otherwise it will go to infinity at r is equal to 0 and you cannot have infinite velocity. So,
that can immediately throw away and you will have only c 2. Now left out and therefore,
u z will only be I mean you can use this boundary condition to determine what is c 2,
agreed. Now this is where the choice of the coordinate system becomes important you are
able to write r is equal to capital R u z is equal to 0 suppose, you had done Cartesian
coordinate system what would you write instead?
𝐹𝑜𝑟 𝑐𝑎𝑟𝑡𝑒𝑠𝑖𝑎𝑛 𝑠𝑦𝑠𝑡𝑒𝑚, 𝑡ℎ𝑒 𝐵𝐶 𝑤𝑜𝑢𝑙𝑑 𝑏𝑒: 𝐴𝑡 √𝑥 2 + 𝑦 2 = 𝑅; 𝑢𝑧 = 0
So, the boundary condition become complicated, if you actually had solved it in Cartesian
coordinate system such a thing is much easily written, because now you have chosen a
cylindrical coordinate system ok. So, it is really the shape of the boundary that determines
the choice of the coordinate system, because application of boundary condition becomes
simpler in that coordinate system there is no other reason to prefer any coordinate system
over anything else ok, it is a application of boundary condition how simple it can be, clear?
108
So, substituting that
1 𝑑𝑃 𝑟 2 − 𝑅 2
𝑢𝑧 = ( )
2𝜇 𝑑𝑧 2
So, we have gotten the velocity profile, now I think we will continue with the calculation
in the next class now any doubts? Here, I will stop and we will meet tomorrow.
109
Fluid Mechanics
Lecture – 14
Circular poiseuille flow
So, this is how it is known as ok pipe flow or in. So, if you have flow between two parallel
plates and if it is driven by a pressure gradient you call it a planar flow, with the name
associated with him is Poiseuille law.
Or circular Poiseuille flow. So, what we have is a pipe right and fluid flow through it and
we found that:
𝑅2 𝑑𝑃 𝑟 2
𝑣𝑍 = (− ) (1 − ( ) )
4𝜇 𝑑𝑧 𝑅
So, we will analyze this equation a little bit; the first thing that we want to do would be to
find out what is the maximum velocity. Can you find out what is the maximum velocity?
𝑑𝑃 𝑅 2
𝑣𝑧𝑚𝑎𝑥 =−
𝑑𝑧 4𝜇
110
The next thing that, we want to calculate is the flow rate through the pipe. Yes volumetric
flow rate yes, volumetric flow rate.
𝑄 = ∫ 𝑣⃗. 𝑑𝐴⃗
2𝜋 𝑅
𝜋 𝑅 4 Δ𝑃
𝑄 = ∫ ∫ 𝑣𝑧 𝑟 𝑑𝑟 𝑑𝜃 =
𝜃=0 𝑟=0 8𝜇 𝐿
I want v average. So, how do I calculate? So, I want to; so we find that the velocity profile
is like that right, it has a shape of a parabola, but actually it is a paraboloid because it is a
3D parabola. It is a paraboloid with velocity being like that ok. The velocity is maximum
at the center and we also calculated the volumetric flow rate, it is also good to know what
is the average flow rate. So, v average is given by:
𝑎𝑣𝑔 𝑅 2 Δ𝑃
𝑣𝑧 =
8𝜇𝐿
Also, pressure drop is given by
8𝜋𝐿𝑄
Δ𝑃 =
𝜋𝑅 4
This is one of the most used expressions ok. Why is it one of the most used expressions?
Because given a flow rate, and pipe length, and pipe radius you will know which pump
you should use to pump a fluid of viscosity mu. That is what that expression is given giving
you. Therefore; it is used highly ok. what else yeah. So, when; can we use this expression
always?
There are original like assumptions ok, that u r is equal to 0 u theta equal to 0, there can
be situations where these thing can develop ok. And one of the simplest place, where that
can develop would be a turbulent flow ok. Turbulent flow meaning, flow is highly
fluctuating, it will have all components ok. So, there is a limit in which all our analysis is
correct and that is decided by the Reynolds number. When is the flow laminar in terms of
Reynolds number in case of a pipe?
𝑅𝑒 < 2100 → 𝐿𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤
111
So, whenever you use this expression, it is important to check whether Reynolds number
is actually less than 2000 or 2100. If it comes out to be less than that, your analysis will be
ok. You will be finding out the correct pressure drop, otherwise you should be changing
your analysis ok. Let us do one more thing, let us go back to our non dimensional equation.
𝑅 2 𝑑𝑃 1 𝑑 𝑑
= ∗ ∗
(𝑟 ∗ ∗ (𝑢𝑧∗ ))
𝜇𝑈 𝑑𝑧 𝑟 𝑑𝑟 𝑑𝑟
Substituting in the expression we get,
1 𝑑 𝑑
∗ ∗
(𝑟 ∗ ∗ (𝑢𝑧∗ )) = 4
𝑟 𝑑𝑟 𝑑𝑟
So, in fact so this is our what non dimensional equation looks like right k one. So, this is
very interesting, because if you look at this; this is really the governing equation, that told
you what the velocity profile was right. And this velocity profile has no information about
the fluid properties, no information about the pipe dimensions. You see that so that means,
this is like you know whatever you know a pipe you take as long as it is going to be a
cylinder. Whatever fluid you take ok, if it is in the laminar regime this is always going to
be the governing equation, you are always going to get a parabolic shape with a maximum
of. So, what will be the value of U z star maximum value of U z star?
112
We defined U z star as U z divided by U r star as r divided by capital R, U z star will vary

between 0 and 1 r star will vary between 0 and 1 ok. So, the velocity profile is always
going to be a parabola with a minimum value of 0 a maximum value of 1, in a space which
will go from 0 to 1 ok. As long as you non dimensionalize that profile with these things
ok. So, that is what it says irrespective of the fluid property and so on.
113
Fluid Mechanics
Lecture – 15
Shear stress Distribution
So the next thing is Shear stress Distribution. So, Tom Joseph yeah; so, what is shear stress
Tom Joseph? Stress is force per unit area; so what is shear stress?
Shear force per unit area that is all it is ok; it is essentially nothing, but the viscous forces.
So, we want to calculate how is shear stress distributed and looking at that is going to be
little more useful because it will tell you know some nice detail about the flow.
114
So, can you calculate; so what is the relevant shear stress that I will have. So, shear stresses
has how many components? It has nine components in general be in a 3d system ok; so
because we are talking about a cylindrical coordinate system; we will have
𝜏𝑟𝑟 , 𝜏𝑟𝜃 , 𝜏𝜃𝜃 , 𝜏𝑟𝑧 and so on ok.
The only quant only shear stress that is really relevant for us is here because you have 𝑢𝑧
as a function of r. So, the only shear stress that we need to really care about is:
𝜕𝑢𝑧
𝜏𝑟𝑧 = 𝜇
𝜕𝑟
And that so and that will be a non-zero quantity because u z is a function of r anything else
you are going to take most likely we will end up being 0 or a constant. So, that is the
quantity; so can you go ahead and calculate
𝑟 𝑑𝑃
𝜏𝑟𝑧 = ( )
2 𝑑𝑧
So, that means, the shear stress varies linearly in the pipe ok; the velocity varied
quadratically, the shear stress is going to vary linearly in the pipe. So, if these are walls of
the pipe and this is at the center; you will find that it is basically going to. So, what will be
it at the center? What will be the shear stress at the center? It is 0; why is it 0?
115
Because you have an r max; you have a velocity maximum there. So, dv by d r has to be 0
that is why the shear stress is 0. So, shear stress is 0 and then it will decrease sorry it will
increase with a negative sign linearly towards the wall. So, that is how it really the shear
stress distribution is going to look like. So, this is really; so that is my r this is my pipe that
is my velocity u z and the profile would for τ r z would look like that any doubt, Rakshitha?
So, shear stress at the wall:
𝑅 𝑑𝑃
(𝜏𝑟𝑧 )𝑟=𝑅 = ( ) → 𝑊𝑎𝑙𝑙 𝑠ℎ𝑒𝑎𝑟
2 𝑑𝑧
So, this is really the shear stress at the wall and this is known as wall shear. And it is an
important quantity because that is what actually is you know resisting; your flow. You are
actually pushing the fluid with a pressure gradient, the wall is opposing it. The wall is
opposing it by applying a shear stress on the fluid.
So, typically when you want to talk about a force that is resisting your fluid motion; it is
really the wall shear ok. Now that is true now in case of a laminar flow and this is a quantity
that is going to change when you are going to talk about other flows. Like you know
turbulent flow or you will see few more examples later, but wall shear is actually an
important quantity that is also clear.
Let us write down that from a cell balance. So, you have done the cell balance approach
in your continuum mechanics; is that simpler than what we do here? Or is it was a difficult?
116
It is actually not very hard, let us just do a quick one ok. So, we have fluid flow through
the cylinder the reason I am trying to do that is because it gives you a little more insight
on what is happening. Let us say the pipe and let us consider a cylindrical fluid element; a
cylindrical fluid element of radius r of length let us say small l ok; you have applied a
pressure gradient.
So, let us say on this side the pressure is P plus delta P and on this side the pressure is P
ok. And what happens because it is basically in a fluid; the surrounding fluid will exert a
shear stress on the cylindrical element, which is externally going to be 𝜏𝑟𝑧 which is acting
in the opposite direction everywhere on the surface of the cylinder right. Then you can say
Δ𝑃. 𝜋𝑟 2 = 𝜏𝑟𝑧 2𝜋𝑟𝑙
Δ𝑃 𝑟 𝑟 𝑑𝑃
𝜏𝑟𝑧 = = ( )
𝑙 2 2 𝑑𝑧
So, it is one and the same thing that is what I wanted to tell you; that if you have it actually
started from a cell balance; you would have ended up with the same equation. You would
have written down this equation and then you would substitute τ r z using Newton’s law
of viscosity, you can find out then what is the velocity profile; that is clear?
Look at this force balance; this force balance tells that any fluid element you are going to
see; it experiences a force because you have applied a pressure from one side. So, the fluid
element is actually trying to move in presence of the rest of the fluid, but when it is trying
to move in presence of other fluid elements, it is experiencing a shear stress in the opposite
direction. Because viscosity does not like things to move with a differential velocity;
viscosity does not like that viscosity will try to resist it, it will try to resist it by applying a
shear stress ok.
So, this pressure is trying to push it in one direction; the shear stress is trying to push pull
you know push it in the other direction and that fluid element therefore, is in a equilibrium
state whether it is; it is not being acted by a net force. But it is still continuing to move
because we have already assumed its steady state; so, it is moving with a steady state
velocity. So, that is what is actually happening its clear.
So, it is just a convention ok. So, in this pipe we are saying that this is going to go with
that way; that means the pressure you have applied is larger compared to here ok. So, in
117
this direction we have pressure that is decreasing ok. So, when we define dP by dz this
was our positive z direction. So, dP by dz is just you know or other way to think about it;
it is really P minus P plus delta P divided by L because the pressure on that side minus
pressure on this side.
Depending upon whether you want to talk about whether the fluid element is exerting shear
stress on the fluid outside or it’s the other way; there exist one convention. There also
exists another convention in which whether you want to interpret shear stress as a force or
momentum flux ok. So, the minus is essentially to do with that, but the point is that; the
when you consider the fluid element, pressure is the one which is going to push the fluid
on the right side, shear stress is the one which is going to push the fluid in the other side
and it is a balance of that.
We will do quickly comment about pressure now. So, I kept saying that we will you know
always do this; I mean I have been doing this with the horizontal pipe right. And many of
you actually had this concern that why aren’t we talking about gravity. And I kept saying
that ok let us talk about horizontal pipe let us in ignore gravity.
But gravity is always there; the pipes need not be horizontal and you know that you we
have gravity present in that equation, but we need not solved again because there is a nice
way of interpreting things. So, let us just look at that interpretation and before moving on.
So, we will start with again Navier stokes equation, but we will stick to Cartesian
coordinates because that is simple and we will just look at a 1d equation.
118
So x-momentum equation:
𝜕𝑢𝑥 𝜕𝑃
𝑥 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚: 𝜌 [ +𝑢
⃗ . ∇𝑢𝑥 ] = − + 𝜇∇2 𝑢𝑥 + 𝜌𝑔𝑥
𝜕𝑡 𝜕𝑥
𝑑𝑃
𝐹𝑜𝑟 𝑠𝑡𝑎𝑡𝑖𝑐 𝑓𝑙𝑢𝑖𝑑 𝑖𝑛 1𝐷: = 𝜌𝑔𝑥
𝑑𝑥
𝑃 = 𝑃0 + 𝜌𝑔𝑥 𝑥 [𝐻𝑦𝑑𝑟𝑜𝑠𝑡𝑎𝑡𝑖𝑐 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑑𝑢𝑒 𝑡𝑜 𝑔𝑟𝑎𝑣𝑖𝑡𝑦]
So, it is as if there exists a pressure gradient that supports the weight of the fluid and that
is what gives rise to buoyancy forces and so on right. So, what we can do is we can say;
so what our intention right now is to take care of this gravity. And what I am going to do
is I am going to bring this gravity in defining a pressure and that pressure is called dynamic
pressure:
𝐷𝑦𝑛𝑎𝑚𝑖𝑐 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒: 𝑃𝑑 = 𝑃 − 𝜌𝑔𝑥 𝑥
𝜕𝑢𝑥 𝜕𝑃𝑑
𝜌[ +𝑢
⃗ . ∇𝑢𝑥 ] = − + 𝜇∇2 𝑢𝑥
𝜕𝑡 𝜕𝑥
So, dynamic pressure essentially is defined in order to accommodate gravitational force

and this is the equation in fact, that we solved throughout right. So, the pressure that we
have really calculated; even in presence of a gravity was really the dynamic pressure. If
119
you get dynamic pressure and if you want to talk about the total pressure; now you know
the relation, you can use this relation and find the total pressure. So, that is the way to
easily accommodate gravity without really caring about gravity is that clear.
So, what would it look like if you write down let us say velocity profile for our you know
pipe flow?
𝑣𝑧 = (− ) (1 − ( ) )
4𝜇 𝑑𝑧 𝑅
What pressure is this? Will this be absolute pressure or dynamic pressure? This is the
dynamic pressure that we really used. So, can you substitute for in terms of absolute
pressure and gravity what would that mean? That is what it is going to be right that is all
it is ok.
𝑣𝑧 = (− + 𝜌𝑔𝑧 ) (1 − ( ) )
4𝜇 𝑑𝑧 𝑅
So, this is nice because let us say you consider two walls and then there is fluid in between
ok. This side is P atmosphere, this side is also P atmosphere; that means, dP by dz is 0;
you only have ρ gz because of which it is going to develop a flow and the fluid is going to
go down right; that is all it says.
120
On the other hand, if you had applied a pressure gradient in this direction; let us say by
putting something below like fluid in a container. Then what are you going to do? You are
going to apply a upward force ok; you are going to apply a dP / dz upward and these two
terms will cancel each other, when dP / dz is exactly equal to ρ gz in which case there is
no flow. If dP by dz is more than the ρ gz then there will be an upward flow that is what it
is.
So, it is really that competition that these two terms are showing; that is also clear, next
problem.
121
Fluid Mechanics
Prof. Sumesh P.Thampi
Lecture – 16
Flow between two concentric cylinders
So we started looking at the two concentric cylinders with fluid in between and we
essentially calculated the velocity profile that is all we did yesterday, right. And, we also
looked at the limit how to take the limit of such a problem, so that you know if you want
to talk about a thin film or thin gap how does it look like. We will look at that a little bit
more in detail.
So, this flow is known as Taylor Couette flow. So, right and we found that there is only
one component of velocity which was u θ right and then you found out an expression for
that which happened to be a function of r.
𝑢𝜃 = 𝑢 𝜃 (𝑟)
The inner cylinder was rotating with certain velocity omega which means:
𝑢𝜃 (𝑟𝑖 ) = Ω𝑟𝑖
𝑢𝜃 (𝑟0 ) = 0
122
So, if you look at the velocity profile; the velocity profile would basically decay something
like that that you like to actually plot and see, but that is what. Now, this profile is not a
very stable profile. In fact, what happens is that if you define something called a Taylor
number:
𝑟𝑖 (𝑟𝑜 − 𝑟𝑖 )3 Ω2
𝑇𝑎𝑦𝑙𝑜𝑟 𝑛𝑢𝑚𝑏𝑒𝑟, 𝑇𝑎 =
𝜇 2
(𝜌 )
𝜇
→ 𝐾𝑖𝑛𝑒𝑚𝑎𝑡𝑖𝑐 𝑣𝑖𝑠𝑐𝑜𝑠𝑖𝑡𝑦
𝜌
Kinematic viscosity is it is basically defined as the ratio of dynamic viscosity to the density
of the fluid and it is a very useful quantity and it is like diffusivity in the. So, you have to
heard about mass diffusivity, you have heard about thermal diffusivity, have you heard
about? Ok. You will hear more about in the mass transfer and the heat transfer course and
this quantity is a analogous quantity this is mu by rho is more like diffusivity, it is called
a mu by rho is called a kinematic viscosity.
So, what happens is that if this quantity is approximately 1700 or below then whatever you
derived is going to be the flow profile when you do it experimentally. But, if it exceeds
that, then this flow happens to be what you call unstable in other words you will start
seeing more or different things.
123
So, to illustrate that so, let us say let us just look at the gap now, so I mean it write in the
next page ok. So, that is my r i, that is my r o. So, if I am increasing my Taylor number
which is defined in that particular way, so Taylor number just includes the what you call
the inner radius, the gap the angular velocity and some fluid properties. So, let us say you
are doing an experiment and you calculate this and it came out to be let us say less than
1700, you will see the flow profile exactly what you derived in the last class.
But, if it happens to be bigger than that, then the flow profile changes. For example, if I
take a cross section ok, a perpendicular cross section and look at what is the flow that is
going to look like there will be a primary flow which will actually have a u θ which would
be like that, but it also develops a secondary flow. So, u θ will be there as a primary flow
and then it in addition it develops a secondary flow and that secondary flow will have both
u r and u z components. And, the flow in fact, looks something like this and so on this side.
So, that means, the fluid is not only going like this it is also trying to do rotate like that.
So, I am just looking at a cross section and you will see that there are like you know things
are going circle in the in round and round. So, each of these are called Taylor vortices;
vortices, so when you say vortex you know you have some feeling of vortex right; vortex
is something which is going round and round. So, here you see that the fluid element will
actually start going in round and round. So, these things are called Taylor vortices each of
them.
Can you imagine how a fluid element will actually be going? So, so see when it was less
than 1700, it only had a u θ component ok. Now, I am saying that if I am looking at a cross
section with a Taylor number greater than 1700, you would actually start seeing things
going round and round, but in that is only a secondary flow. The primary flow is still in
the u θ direction. So, what happens if this is a fluid element it was originally going like
that, but now it had to do a rotation. So, it would actually go like that; it would go like a
helix ok.
So, any section you look like any section you take if you only look at that radial cross
section it would look like a you know vortex, but in addition things have to go round and
round, so overall it would be helix, but then this is how the cross section would look like
you know it will form bands and you can see that this has go to u z component, u r
component and a u θ component. So, all of them will come into the picture. Is that clear?
124
Now, this is not the end so in fact, when you look at the experiment you would see that
you know if this is the gap you will see that you know there are these bands kind of thing
that has formed. So, each of these band would represent these Taylor vortices. There will
be just you know arranged next to each other, just a representation each of them, so in a
one vortex band, next vortex band, next vortex band it forms. Now, if you increase the
Taylor number further even this becomes unstable ok.
125
So, then in addition ok; so, that is my inner and outer cylinder that is my r i, this was my r
o. So, this is, so this was the you know this is the one which you see on the right hand side
this was what the known as the Taylor vortices and the bands. These bands become
unstable and the bands would look something like this.
So, each of these bands will not just go in one direction like that, each of the band will
start doing like this go around. So, that is even more complicated flow profile, that is what
is going to happen. I can further increase my Taylor number because it is essentially a
property it is a number that you know consists only of geometrical parameters and a fluid
properties. If I increase it further that is actually going to give you a turbulent flow between
the plates..
This is not turbulent yet. Turbulent flow would mean that you know everywhere you are
going to see huge fluctuations. This is still laminar ok, the flow is still laminar, but the
flow has got all three components and there is nothing that is fluctuating it is just that you
know there is a you know some kind of a wave that you would see. In fact, each fluid
element would act like a wave which is traveling, so that is what.
So, all these three states the one which we derived yesterday which is the primary state
that is happens is a you know simple flow which would mean that it has just u θ going
round and round, that is also laminar flow. With further increase in Taylor number you
will see start seeing Taylor vortices which means u r, u θ, u z all of them are there it will
form bands. The flow is still laminar and then further increase ,the bands become unstable
and will form some waves it is the flow is still laminar and then it forms turbulent flow.
So, there are 2-3 points that I want to drive from here ok. One is that we keep making
assumptions right we say that oh I will take del by del θ to be 0, I will take my u r to be 0
ok. Are they always true? So, in this case it clearly says that it is not right with our
assumptions we could get only one of the flow states, if we had we if we were able to solve
the equations fully without making those assumptions, then we would have been able to
you know get any of these flow states ok, but we are not going to get that. So, you can
always question the assumptions that you make. Is that clear? Ok.
Second is that the you know it is the same governing equation right they are ordinary the
differential equation that you started with, it is always same. It is just that you have got a
now four different solutions and that is because the equation is a non-linear equation and
126
a non-linear equation is expected to have multiple solutions. So, you are ending up with
multiple solutions for the same governing equation because the equation is non-linear. Is
that understood two points? Ok one is the existence of multiple solutions, the second is
actually questioning the assumptions that we make Clear. Any questions?
So, so you know the point is that you can do theory always, but you do need to see
experiments ok. Unless you if you were not able to do experiments you would not have
seen any of these states ok. So, it is really experiments that has actually given insight into
what these flow profiles are. How would you do experiment, if you want to see flow? You
can take two cylinders that is not a big deal, you can rotate one of them put fluid and in
order to see it just put some particles ok. Just the way you know when the particles are
there you can see that the particles are moving just like you know you can see or visualize
how smoke is going.
So, the same idea is what is followed in an experiment also, you can take fluid and put
particle and keep increasing the Taylor number. So, let us see what that looks like it is a
nice movie.
127
So, you have two cylinders and one of them is rotating, you may not see much because it
is just the only the u θ things are just going round and round. So, you are not really seeing
you can, if you look carefully you can see things moving right. Can you see things moving?
Can you see fluid moving? So, you can. So, now, so, you are not. So, this is just. Yes. So,
this is Taylor number is very small less than 1700.
Now, they will increase the speed of the rotating cylinder which would mean that you are
increasing the Reynolds number and the Taylor vortices start appearing. The square
vortices that I talked about they will look like bands that is what you are seeing now.
Now, with further increase in the Taylor number this bands beam will become unstable
and it will start generating waves. So, the bands become thinner and thinner first. There
you go, so the formation of the waves ok. These waves this is the third state that we
discussed. So, the bands are forming the waves and the fluid elements actually are they
become traveling wave in fact. You can see that clearly now and with further increase in
the Taylor number it is going to go to turbulence.
So, just remember this flow state is still laminar and now, it is getting into turbulence, so
there you go. So, that flow is turbulence you can see that there is lot of fluctuations in the
velocity, you can see that all that nice structure is going away and it becomes very noisy
ok. So, that is a turbulent flow. Now, what they will do is they are going to reduce the
128
speed and you can see that as you reduce the speed that is equivalent to reducing the Taylor
number and it will go to these transitions again. So, right now it is all fully turbulent flow.
Now, the speed is getting decreased; so, that means, the Taylor number is going down ok.
So, it has come to that state where you have waves of bands. So, yeah, so that is about
Taylor Couette flow.
129
Fluid and particle Mechanics
Prof. Sumesh P.Thampi
Lecture - 17
Taylor couette flow
So, we have seen examples of how we can simplify the Navier Stokes equations namely
the mass and momentum balances in a Cartesian coordinate system. We have also seen
why these equations would be different for other coordinate system. For example, in a
cylindrical coordinate system and the spherical coordinate system and so on and we
learned that the main difference is because the unit vectors that you have for non-Cartesian
coordinate systems is location specific and therefore, they give rise to non-trivial you know
terms in the equation and therefore the equations would indeed look different.
So, we can see an example of a cylindrical coordinate system and see how do we simplify
the equations and that flow is a very nice flow because it gives you a lot of other insights
into the system. So, this is flow between 2 concentric cylinders ok.
So, you have 2 cylinders 1 inner cylinder and 1 outer cylinder, I am giving the top view.
So, let us say that is here inner cylinder and that is here outer cylinder and let us say R i is
the radius of the inner cylinder and R o be the radius of the outer cylinder and the fluid is
between the inner cylinder and the outer cylinder and let us say that this inner cylinder is
130
rotating. It is rotating with an angular velocity omega and so the. So, you have a fluid now
between 2 cylinders; the inner cylinder is rotating because of which the fluid outside will
actually start moving, but because the outer cylinder is not rotating the velocity of the fluid
exactly on the outer surface or outer cylinder should be 0 and our idea is to find out how
does the velocity change from the inner to the outer cylinder ok.
So, the appropriate coordinate system that we want to consider here would be cylindrical
coordinate system. Let us assume that the cylinders that you are looking at are really long
ok; so, much long in the z direction. So, the z direction is going to be perpendicular to the
plane that we are considering and the cylinders are so long that we can neglect all
variations in the z directions. So, we will and also the velocity in the z direction. So,
because this inner cylinder is rotating you would expect that the fluid flow is in the θ
direction really.
So, if I am taking the cylindrical coordinate system I have an R which is in the direction
of the radius, then I will have a θ direction which is in the plane and then there is an z
direction and that is perpendicular to the plane you will assume that V z is equal to 0 and
∂ by ∂ z are also equal to 0 because the cylinder is infinitely long perpendicular to the
surface there is nothing that is going to change in that direction. If that is the case then let
us see how do the equations simplify ok.
So, the first thing that we need to write down is the mass conservation equation.
131
1𝜕 1 𝜕 𝜕
(𝑟𝑣𝑟 ) + (𝑣𝜃 ) + (𝑣𝑧 ) = 0
𝑟 𝜕𝑟 𝑟 𝜕𝜃 𝜕𝑧
And as I said a you can take it from any standard textbook because the equations would
look a little different from the one which we write in Cartesian coordinate system. So,
again you have r θ and z direction. So, there are 3 terms one depends upon v r the second
depends upon v θ and the third depends upon v z.
Now, we need to simplify this equation, we know that we are assuming that the flow is
purely azimuthal; that means, it is only in the direction of θ it is and the pipe is infinitely
long.
Now, also we are going to say that there is nothing that is going to change in the θ direction
because θ direction if you look at that is the flow is just continuous there is no θ that is
different. So, θ equal to 0 or θ equal to pi by 4 it is just same. So, you would not expect
anything that is changing in the θ direction and therefore, derivatives of any quantity in
the θ is also going to be 0.
𝜕
( )=0
𝜕𝑧
𝜕
( )=0
𝜕𝜃
1𝜕
(𝑟𝑣𝑟 ) + 0 + 0 = 0
𝑟 𝜕𝑟
On integration:
𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑣𝑟 =
𝑟
Now, in order to evaluate this constant we need to calculate what we need to have some
physical insight. So, what should be let us say the value of v r on the surface of the inner
cylinder. Let us imagine that is your inner cylinder and this inner cylinder is rotating you
know in a counterclockwise fashion with some angular velocity. What do you expect the
value of v r on the surface of the cylinder? You cannot have fluid that is going into the
cylinder because the cylinder is a solid cylinder. So, v r on the surface of the cylinder has
to be 0. So,
132
𝐴𝑡 𝑟 = 𝑅𝑖 , 𝑣𝑟 = 0 → 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 = 0
So, now we have said we have found that v r is 0 everywhere in the fluid domain v z is
equal to 0 because we have assumed an infinitely long pipe and there is no flow along the
length of the pipe and there is only fluid flow is in the θ direction and therefore, we need
to only worry about the θ component of the fluid velocity. We can write down only the θ
momentum equation, but before that it is worth looking at the radial component of velocity
or r momentum equation as well.
Because that is going to give you some insight into the problem; so, we can look at θ
momentum equation that is what is going to give you information about the about the flow
profile.
So, let us start by doing that, we start by writing down the θ momentum equation
𝜕𝑣𝜃 1
𝜃 − 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛: 𝜌 + 𝑣⃗. ∇𝑣𝜃 + 𝑣𝑟 𝑣𝜃
𝜕𝑡 𝑟
1 𝜕𝑃 𝑣𝜃 2 𝜕𝑣𝑟
=− + 𝜇 [∇2 𝑣𝜃 − 2 + 2 ]
𝜌 𝜕𝜃 𝑟 𝑟 𝜕𝜃
𝜕𝑣𝜃 1 𝜕𝑣𝜃 𝜕𝑣𝜃

𝑣⃗. ∇𝑣𝜃 = 𝑣𝑟 + + 𝑣𝑧 =0
133
1𝜕 𝜕 1 𝜕 2 𝑣𝜃 𝜕 2 𝑣𝜃
∇2 𝑣𝜃 = (𝑟 (𝑣𝜃 )) + 2 +
𝑟 𝜕𝑟 𝜕𝑟 𝑟 𝜕𝜃 2 𝜕𝑧 2
1𝜕 𝜕𝑣𝜃 𝑣𝜃
(𝑟 )− 2 = 0
𝑟 𝜕𝑟 𝜕𝑟 𝑟
So, that is what the θ momentum equation reduced to. We can look at what r momentum
equation reduced to:
𝜕𝑣𝑟 𝑣𝜃 𝜕𝑃 𝑣𝑟 2 𝜕𝑣𝜃
𝜌( + 𝑣⃗. ∇𝑣𝑟 − ) = − + 𝜇 (∇2 𝑣𝑟 − 2 − 2 )
𝜕𝑡 𝑟 𝜕𝑟 𝑟 𝑟 𝜕𝜃
𝜕𝑃 𝑣𝜃2
=
𝜕𝑟 𝑟
Now, if you look at this equation carefully the right hand side that is V square by r. So,
that is not force or it is the for centrifugal force density the centrifugal force acting per unit
mass ok. So, that is equal to the pressure gradient. So, in other words whenever the fluid
elements are rotating it experiences a centrifugal force and that force is actually balanced
by a pressure gradient in the r direction. So, something is rotating in the θ direction it
experiences a centrifugal force and that force is exactly balanced by the pressure gradient.
So, the radial pressure gradient balance the centrifugal force, radial pressure gradient
balances the centrifugal force.
134
So, that is the information that you get the r momentum equation. The θ momentum
equation of course, gave you a differential equation which we need to solve and find out
what is we solving it now.
1𝜕 𝜕𝑣𝜃 𝑣𝜃
(𝑟 )= 2
𝑟 𝜕𝑟 𝜕𝑟 𝑟
𝑑2 𝑣𝜃 1 𝑑𝑣𝜃 𝑣𝜃
+ − 2=0
𝑑𝑟 2 𝑟 𝑑𝑟 𝑟
So, that is the differential equation that we need to solve to calculate what is v θ. Now,
these kind of differential equations ok. So, you know d square y by d x square plus 1 by x
d y plus y by x square is equal to 0.
135
So, these equations are called Euler Cauchy equations and they admit a solution of a power
law ok. For example, here the solutions that these equations will be something like r to the
power of m. So, that is the kind of solution that we will have. So, we can actually try to
find out what that would be. So,
𝐼𝑓 , 𝑣𝜃 = 𝑟 𝑚 → 𝑆𝑢𝑏𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑛𝑔 𝑖𝑛 𝑡ℎ𝑒 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑡𝑖𝑎𝑙 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛
1 𝑟𝑚
𝑚(𝑚 − 1)𝑟 𝑚−2 + 𝑚𝑟 𝑚−1 − 2 = 0
𝑟 𝑟
𝑚(𝑚 − 1) + 𝑚 − 1 = 0
(𝑚 − 1)(𝑚 + 1) = 0
𝑚 = ±1
That means, my solution is going to be
𝐶2
𝑣𝜃 = 𝐶1 𝑟 +
𝑟
136
We have an inner cylinder that is rotating with an angular velocity omega. We have an
outer cylinder that is static. So, and the inner cylinder has a radius R i the outer cylinder
has a radius R o. So, we can say that
𝐴𝑡, 𝑟 = 𝑅𝑜 ; 𝑣𝜃 = 0
𝑟 = 𝑅𝑖 ; 𝑣𝜃 = 𝑅𝑖 Ω
So, we can substitute and find that
137
𝐶2 𝐶2
0 = 𝐶1 𝑅𝑜 + → 𝐶1 + 2 = 0
𝑅𝑜 𝑅𝑜
𝐶2 𝐶2
Ω𝑅𝑖 = 𝐶1 𝑅𝑖 + → Ω = 𝐶1 + 2
𝑅𝑖 𝑅𝑖
Ω
𝐸𝑙𝑖𝑚𝑖𝑛𝑎𝑡𝑖𝑛𝑔 𝐶1 , 𝑤𝑒 ℎ𝑎𝑣𝑒 𝐶2 =
1 1
( 2 − 2)
𝑅𝑖 𝑅𝑜
𝐶2 Ω
𝐶1 = − = −
𝑅𝑜2 1 1
𝑅𝑜2 ( 2 − 2 )
𝑅𝑖 𝑅𝑜
And therefore, you are ready to get your expression
𝛺𝑅 𝛺
𝑣𝜃 = − +
1 1 1 1
𝑅𝑜2 ( 2 − 2 ) ( 2 − 2 ) 𝑅
𝑅𝑖 𝑅𝑜 𝑅𝑖 𝑅𝑜
138
So, if you want you can rearrange that expression you can do it yourself and nicer way to
represent it would be:
𝑅 𝑅
( 𝑅𝑜 ) − (𝑅 )
𝑜
𝑣𝜃 = (Ω𝑅𝑖 ) ( )
𝑅𝑜 𝑅𝑖
( 𝑅 ) − (𝑅 )
𝑖 𝑜
Just rearranging that equation if you like. So, that is the expression for the velocity profile
between 2 concentric cylinders. In fact, if you plot it if that is the inner cylinder and that
was the outer cylinder and if the inner cylinder is rotating you will see that the velocity
profile would. So, the velocity at that point would be something of that sort and then it
should become 0 and it would decay like that nice to plot it and see.
But that is what the velocity profile is going to be. We just box that is the expression you
need to remember anything it is just that is the information that you get when you solve
the cylindrical problem. It also tells you how you should deal with the equations in
cylindrical coordinate system, how some of the non-trivial terms you know contribute and
the pressure centrifugal force balance is the other thing that nicely came out of this
analysis.
139
Fluid Mechanics
Lecture – 18
Viscosity and Momentum Transfer
Last few days you have seen flow in various contexts. We started with looking at flow
between two parallel planes; we have also seen flow in a cylindrical channel that is actually
a pipe or flow between two concentric cylinders. You know also sometimes you know the
flow was you know it was along the long axis of the channel along the axis of the channel,
you have seen flow you know in circular flows. And, in each width depending upon the
geometry, now you are familiar with how to go about it you know there are certain
assumptions that you can make then you can question those assumptions and so on ok.
So, in each of these cases there was one property of the fluid which came up again and
again. What was that property that always that, you know it was coming up? What fluid
property was there in most of the equations that we derived?
Student: Viscosity.
Viscosity ok. So, why is viscosity so important? Where does viscosity come from or can
we understand viscosity a little bit more in detail? Or, how does you know or what is the
molecular origin of viscosity? That is what we are going to actually look at today and then
you will see that viscosity is not the only property that you need, there will be many other
properties that you might need in certain situations. So, we will see a few of them in the
next couple of classes ok. So, that is the idea for the next two three classes. So, we are
going to look at what is viscosity.
140
So, how does one define viscosity? What is the impression that one has? That it is a
resistance right; it is a resistance to the fluid flow a little more precise is it is a resistance
for the deformation ok. So, if a fluid element is undergoing deformation, then you know
whether the fluid resists that deformation or not that is basically the idea of viscosity right.
And, if you go back to your older notes what you would have seen is that you take let us
say fluid between two plates right and let us say that you apply some force F on the top
plate there is fluid in between. And, let us say that this that area of the top plate is A because
you have applied the force the top plate would start moving with some velocity which let
us call it capital U and let us say if h is the distance between the plates.
You would say that F by A is essentially proportional to U by h and you would have made
it an equal sign by defining a coefficient of viscosity μ that is what is done right and this
left hand side is what is called as the stress and U by h is the velocity gradient.
𝐹 𝑈
𝑆𝑡𝑟𝑒𝑠𝑠: =𝜇
𝐴 ℎ
So, this is what Isaac Newton has postulated and that is what is known as the Newton’s
law of viscosity. Now, so, if you let us say select this as the x-direction and that as the z-
direction we can write our shear stress let us say as τ x z the force that is acting in x
direction acting on an area which has a direction in along the x z direction. So, to get why
this viscosity comes into the picture let us actually replace the fluid. So, the fluid that is
141
between the plates let us imagine that fluid is a gas ok. So, we say that now we have a gas
that is between two plates. So, when I say fluid it could be either gases or liquids generally.
So, we take a gas and then let us say that we start by a situation where U is equal to 0.
So, we will put U is equal to 0. So, the moment you put U is equal to 0; that means, the
gas that you have is essentially a static thing right it is not moving hm. And, you start you
know you know zooming into the gas that is you know filled up in between the plates.
What would you see? You zoom in and zoom in till you start seeing the molecules that is
making the gas ok.
So, you go to that level where you start seeing the gas as molecules. Would those
molecules be having a 0 velocity or they would have some velocity? They have some
velocity right. How do you characterize that velocity? Some root mean square velocity of
a gas that you would have learned, does anybody remember the expression the average
velocity of molecules in an ideal gas? What is that?
𝑣𝑎𝑣𝑔 = √3𝑅𝑇/𝑁
So, that is the average. So, what is R?
The ideal gas constant; T is the temperature and N is the number of molecules that you
have. Now, so, that means, that each of the molecule is moving, but the net motion is 0,
right. So, molecules are moving in all different directions it just so happens that oh a net
142
is 0. So, if one looks at the individual molecules and look at their velocities then in fact, it
would look something like this. So, what I will draw now is on x-axis I will plot let us say
v x ok. So, that is the velocity of the molecule in x-direction and on y-axis I will plot
probability of finding molecules with velocity v x and the picture would look something
like that ok.
So, the molecules which has no velocity the 0 velocity there is a high probability and then
there will be few which are moving in positive x-direction with some velocity there will
be few which are moving in negative x-direction. And, if you say look at the number of
molecules which have got large velocities that fraction is going to be really small ok. So,
there is so, what I am trying to say is that the molecules are not moving with uniform
velocity, they are random in all direction. It also has different speeds ok; some of them
will have 0 speed, some of them will have a large speed and what you had calculated or
what you knew is some average speed of all the molecules, that is clear? Ok.
So, that is the picture that you have when you have contained this gas between two solid
plates which are not moving ok. Now, so, the only thing that happens is that suppose you
start moving the you know moving the plate, so; that means, that the fluid start moving the
gas is moving ok. Then, this picture essentially tilts a little bit tilt. So, it is not really tilting
what I meant to say is that here the mean was at 0 velocity that is because there is no net
velocity.
143
Instead of that you would see that this entire picture is something like that. So, here I have
got v x. So, if I am looking at gas which is moving at a velocity U it is just that the entire
distribution gets shifted ok. So, most of the molecules have a velocity U some of them
have got velocity larger than U, some of them have got velocity smaller than U and on an
average you will see that you know there it has got a velocity of capital U. So, that is what
is it if you say that a fluid element or a gas element is moving with a velocity U. So, that
is really the picture that you have ok.
So, in that sense when you defined your Newton’s law of viscosity which you defined us
τ x is z is equal to μ into ok. So, we had defined it as U by h right U by h which is nothing,
but μ into d v x by dz correct. So, that v x that we have defined is the average velocity of
many molecules that you are considering right. So, U so, what we defined as U by h is
essentially the velocity gradient ok.
So, the velocity at every point now we have realized that it is actually not just a single
value, it is just that the average comes out to be a single value there are things which are
changing from there. So, I should have really defined this v x as an averaged v x. So, the
right way to represent it is that for a gas you should have said it is really:
𝑑 < 𝑣𝑥 >
𝜏𝑥𝑧 = 𝜇
𝑑𝑧
144
Now, in this case what is going to be average v z? There is no fluid net motion
perpendicular to the plate. So, it is definitely 0, but that does not mean that the molecules
are not moving. The molecules are still moving right in particular if you plot v z as a
function of you know sorry the probability of finding a molecule with velocity v z as a
function of v z I am still going to see that right. So, there are molecules which are going
up you know perpendicular to the plate there are molecules that are coming down right
perpendicular to the plate.
Now, that is actually a crucial thing and that is what leads to the definition of viscosity.
So, why do I say that?
So, let us have the plates that is moving with a velocity U. I will draw the velocity gradient
ok. So, if I look at let us say two layers ok. So, this is I do not know some z is equal to
some number some z at some z. We are looking at a plane there are molecules here, there
are molecules there. These molecules have a x-directional velocity, the molecules here
have got an x-directional velocity. It just so happens that because what this profile indicates
is that the molecules that you see at the bottom have a slower average velocity compared
to the molecules above right.
But, these molecules not only are fluctuating when they are moving in this x-direction they
are also fluctuating in the z-direction. So, that means, that some of the molecules that you
are seeing here; so, let me give two different colors, the red ones and the blue ones. They
145
are identical. It is just for convenience that I have given different colors. The blue ones
have lower v x, some of them when they are you know executing their random walk they
will go to the reach above the that plate right. So, not above the plate this region; so, some
of the red molecules will actually intrude into the region of the blue molecules, some of
the blue molecules will come from the region into the red molecules.
What would be the consequence of that? The v the red molecules actually have got a lower
momentum compared to the blue molecules right because on an average the red molecules
are at a lower velocity right. So, because these are moving up and down, some of the red
molecules that have gone towards the region of the blue molecules is essentially
transporting x-directional momentum in z-direction. Sorry, they are. So, I should do the
other way. The blue molecules which have got a large x directional momentum rather than
the red molecules; when they come down they will essentially transport the momentum in
the negative z-direction.
So, the blue molecules have got a larger x-directional momentum because they happen to
move in z-direction or rather minus z-direction. So, we have defined z up. Some of the
momentum gets transported from the blue region to the red region and that is what you
call as momentum transport right. So, that is the definition of momentum transport and
you simply say that momentum gets transported across the layers and it happens because
of this molecular motion ok.
Even though there was no net velocity in the z-direction the random you know components
or the where fluctuating velocity that the molecules have it is sufficient to carry the
momentum from one layer to another layer. So, that is the molecular mechanism of
momentum transport and if you think about red to blue it will be essentially in the other
direction because when the red molecules are go into region of blue it is going to be taking
a lesser momentum in that sense ok. Because this from blue to red it will be a larger
momentum from red to blue it will be a slower momentum or you could say the momentum
in the opposite direction if you were in a center of mass frame ok.
So, the momentum is getting transported from regions of larger v x velocity to regions of
lower v x velocity. So, the direction of momentum transport is in the minus z-direction. Is
that clear? Doubts? So, if you look at momentum transport which is typically a written as
π x z ok. So, that is the amount of momentum that is getting transported:
146
𝑑 < 𝑣𝑧 >
𝜋𝑥𝑧 = −𝜇
𝑑𝑧
So, why would you write that because this is our x-direction, that is our z-direction dv x
by dz is so, vx is increasing in the positive z-direction, but momentum is getting
transported in the minus z-direction.
So, you have to recognize that fact and when you recognize that fact you basically say that
the momentum that is getting transported is really proportional to the gradient in the
velocity and it will really be in the negative direction of the gradient and velocity. So, that
is why momentum transport is defined like this. It is just the way you would have done
your heat transfer. So, in your heat transfer you have learned Fourier’s law of heat
conduction? You have learned yes, no? Heat flux is proportional to temperature gradient
right you would define it as some heat flux is equal to some diffusivity times the
temperature gradient and the heat basically flows in the direction of decreasing the
temperature. So, similarly here the momentum gets transported in the direction of the
smaller velocities ok. So, that is why you have put this negative sign to recognize that fact.
Now, if you look at this the momentum transport and our original you know equation of
what we have defined as the shear stress they are essentially same with just a negative sign
ok. So, in another word of another way of saying is that you would define your τ x z has
just negative of π x z ok. So, shear stress is typically defined in the other way. Now, you
can also see that it is just a matter of convention.
𝜏𝑥𝑧 = −𝜋𝑥𝑧
Doubts? Go ahead.
Student: Sir, how does it saying momentum transport is proportional to velocity gradient?
It is just like this. So, so here you have red here you have red molecules and you have blue
molecules right. So, when the blue molecules and red molecules let us say are at the same
velocity then there is no reason there will not be any net change in the momentum right.
So, whatever momentum that is going to be transported is simply going to be proportional
to the difference in the velocity at two different points. So, when you take a small element
it essentially becomes the gradient. So, that is all.
147
So, you are essentially assuming that the momentum transport is linearly proportional to
the velocity gradient that is a simplest relation that you could think about and that is what
actually Isaac Newton has postulated and most of the fluids do obey that. You had some
doubt?
So, we are right now talking about only fully developed flow, but this momentum transport
mechanism has really nothing to do with it. See, what you are really looking at is you are
looking at a region in the molecular scale you have zoomed in to that scale and we are
seeing that if there are two regions with different average velocities because you know the
molecules you know go you know go in some random direction and mix there is a way in
which the momentum transport happens.
So, what happens is that see the blue molecules right some of them will go this way, but
then we have green molecules here ok. So, some of the blue molecules are going to come
here some of the green molecules are also going to come up there. So, the green molecules
which are going to come up will have a lower momentum ok. So, on an average nothing
happens. On an average the velocity of the layer would still remain like that and that would
happen all the way down here. So, for example, if you look at the molecules which are
very near to the wall they would not be moving at all while the region the molecules which
are very close to the top wall will be moving with a large velocity ok.
So, since we have assumed an incompressible fluid there is not going to be any change in
the density. So, that means, the number of blue molecules that will come towards the red
region will be same as the number of red molecules that are going to go to the blue region
and that will be through true throughout any z that you take.
Student: Sir how will we realize that τ x z is minus of π x z?
No. So, we define τ x z as just μ times velocity gradient. Now, for the momentum transport
we have just said that π x z because it makes sense to define it with a negative sign because
we have defined it as momentum transport or momentum flux is proportional to negative
of the velocity gradient. So, we rather define our shear stresses negative of momentum
transport or momentum flux. See, this negative sign is just to account for the fact that
fluxes always in the direction of the decreasing gradient.
148
It is just a definition, that is all, yeah. Yeah. So, in one case see in one case you defined it
as a force in the other case you have defined it as a momentum flux and you have connected
the two now. In fact, you know why would you define this moment and transport as a
force? Actually it is easy to see. Look the blue molecules there right, the blue molecules
when they actually intrude into the regions of the red molecules, what is it really trying to
do? It is trying to increase the momentum of the red region right. So, that momentum
transport is as if this; so, let me color the regions ok.
So, that is what I am trying to say. There will be an equal number of molecules from the
green region which will also come up there ok. So, there is no reason to expect that the
number of the velocity would change. Now, what is the consequence of that? See, the blue
molecules which have got larger velocity when they are coming into the region of red you
know the red region what it is trying to do is it is really trying to speed up your red region
right. So, it is as if the blue region is exerting some force on the blue region that is why it
is interpreted as a force ok. So, that is why momentum and transport is exactly equivalent
to shear stress.
Or thinking of the other way, if the red molecules are going towards the region of the blue
molecules, then the red molecules are really trying to bring down the velocity of the blue
region. So, that is as if the red region is exerting a force in the opposite direction onto the
blue region. So, that is why it is a shear stress and that is why shear stress and momentum
flux are equivalent quantities.
So, the topmost layer for example, these green ones they are getting continuously collided
with a wall a solid wall which is moving. So, every time a molecule goes and hits that solid
wall it will gain momentum from the top wall. Similarly, the molecules that are there at
the bottom whenever it is hitting that bottom wall it is actually losing momentum because
it was originally moving some velocity, but since it has hit a static wall it will lose some
momentum. So, there will be a continuous transport of momentum from the top wall to the
bottom wall. Anything else? Yeah.
Π is in. So, we have just defined π as a momentum flux ok. So, that is momentum is getting
transported in some direction ok. So, here momentum is getting transported from the top
wall to the bottom wall so, that is getting transported in the minus z-direction. d v x by dz
we have defined it as a positive quantity and therefore, we just put a negative sign.
149
No, no. I am saying that the effect of the effect of the momentum transport is as if to reduce
the speed of it. If you look at it is not really reducing it because there is an equivalent thing
that is coming from the top layer ok, but these so, if you want to think about it as a problem
in which you start everything static and then you are you know moving the top layer then
you can see that a layer of molecules which are going to come down to the next one is
really speeding up the next layer. And, that would continue to happen till it reaches the
fully developed flow and we are really looking at right now as a fully developed flow. But,
even otherwise this is the picture. Let us try to write this thing mathematically ok, that is
going to give us an expression for viscosity in fact.
So, let us look at some region let us call that as z is equal to 0. There are molecules here,
there are molecules above. Let us say some of these molecules are going up and let us call
j as the flux of molecule crossing a plane at z is equal to 0 and we do know that the flux of
molecules which are crossing downward would also be equal to j correct. So, π x z the
momentum that is getting transported is going to be equal to what the net momentum in
that direction to the upward direction minus the net the momentum that is getting
transported in the other direction right. So, let us write that.
So, we need to know, so, let us define J number flux. So, that means, the number of
molecules that are crossing per unit area ok. So, if that is the case then it is equal to J times
m these two layers ok. So, the molecules; so, the molecules that are crossing z is equal to
150
0, why do they do that? The picture is that the molecules are hitting each other they
undergo collisions and then they you know move in a random fashion right. So, that means,
that if I look at z is equal to 0 plane, the molecules that are going to cross z is equal to 0
plane would have undergone some collision before crossing that way and do you know
how often the collisions happen? You have heard about mean free path right? What is
mean free path?
So, that means, I assume that all of you know. So, the mean free path is essentially the
distance between average distance between two consecutive collisions. So, we can actually
imagine that the things which are going to cross z is equal to 0 would have undergone a
collision somewhere at a distance you know approximately of the order of λ from z is equal
to 0 or in other words this could be we can define it as let us say approximately minus λ
and this one is going to be approximately plus or plus λ ok.
So, that is the typical distance with which these things these things would have moved
between collisions. So, all the molecules that are crossing z is equal to 0, would have
undergone collision at some distance close to λ before it is crossing it and so, the other
way. So, in other words one could say that we write momentum flux as:
𝜋𝑥𝑧 = 𝐽𝑚𝑈𝑥 |𝑧=−𝜆 − 𝐽𝑚𝑈𝑥 |𝑧=𝜆
Now, what would be an approximate value for J? We will correct it ok. So, right now I am
just assuming that it is going to be of the order of λ. We can write that as:
𝑑𝑈𝑥 𝑑𝑈𝑥 𝑑𝑈𝑥

𝜋𝑥𝑧 = 𝐽𝑚 {[𝑈𝑥 |𝑧=0 − 𝜆 |𝑧=0 + ⋯ ] − [𝑈𝑥 |𝑧=0 + 𝜆 |𝑧=0 + ⋯ ]} = −2𝐽𝑚𝜆
𝑑𝑧 𝑑𝑧 𝑑𝑧
151
So, J is number flux and some so, a reasonable way of telling J is the J is equal to the molar
concentration. So, essentially the number of molecules per unit volume into let us say some
average velocity. So, this is the number of molecules per unit volume into some average
velocity will give you some average flux at various points and therefore, this quantity is :
𝑑𝑈𝑥 𝑑𝑈𝑥
𝜋𝑥𝑧 = −2𝑛 < 𝑣 > 𝑚𝜆 =𝜇
𝑑𝑧 𝑑𝑧
Now, there is a factor that should really come in because I have just said that you know it
goes from minus λ to plus λ and so on which is not really correct. But, the picture that
comes out is going to be correct because it says that my viscosity my viscosity is going to
be proportional to the number of molecules that you have its average velocity, the mass,
the mean free path and so on. So, let me repeat ok.
So, what I am trying to say is that now you want to. So, you have understood what is
momentum transport and you want to write down a mathematical picture based on that ok.
Now, it is very hard to write down an exact picture you do not know at any moment where
the molecules are. But, you can sort of hand wave and write down something which might
give you some decent picture and that is what we have really tried to do here by saying
that let us look at any you know a plane let us say this z is equal to 0.
And, let us say that the layer above it is moving with a larger velocity ok. So, that is your
really profile velocity profile the molecules above z is equal to 0, are moving with a larger
152
velocity; the molecules that are below are moving with a lower velocity because of the z-
component they are going into each other’s regions. You are trying to write down what is
the momentum that is getting transported because of this you know molecules going up
and down ok.
So, what is that momentum going to be? It is nothing, but the number of molecules that
are getting transported times the velocity of those molecules and we are worried about now
x-directional momentum getting transported. So, we are writing down everything that is.
So, we need mass number of molecules that are going per unit area and then the velocity
in one direction versus that in the other direction except that we really do not know whether
they are coming from.
So, I am just saying that I will assume that the ones which are going you know through
this z is equal to 0 plane they have come from a distance of minus λ and the ones which
are coming from plus is they are coming from a distance plus λ because those are some
numbers that I have some idea about. So, I just if I use that then I basically end up seeing
that my momentum transport or momentum flux is n into v into m into λ times d U x by
dz. So, that is what comes out.
So, that means, viscosity is basically coming from this picture and it is dependent upon
these quantities such as mean free path, mass and number density and so on.
153
So, that is the idea that I want to in fact, if you do the proper derivation what you find is
that viscosity is:
1
𝜇= 𝑚𝜆 < 𝑣 >
3
So, it is just that you know a complete derivation will give you this picture where the only
that pre factor changes. There is a 2 that will comes is it because you are adding two things
yeah anyway it is a wrong pre factor. Yes it is for a 3-dimensional case. In fact, you can
do calculations very rigorously which is probably beyond the scope of what we can do at
the moment.
Student: It is a kinetic theory?
It is kinetic theory. So, we are really using the results of the kinetic theory and that means,
we also know what is λ. Do you know what is λ from kinetic theory? The expression is:
1
𝜆=
√2𝜋𝑑 2 𝑛
You can calculate the mean free path based on the molecular size and the number
concentration that will help you what is the approximate value of viscosity that a fluid
would exhibit. And, that is why viscosity is there and that is why whatever you know
equations that we have derived is in that particular form that is the take home message.
So, yeah because that is what I am saying I do not know way up to what distance I should
really consider, but the typical distance that you will have would be of the order of what
you call λ because, if it has if it is hitting another molecule then this when it is going to
come back. So, the average distance before it is undergoing to go a coalition is going to be
of the order of λ. You could do it as λ by 2 if you like, but then you will see that that pre
factor is going to change, but we know that that pre factors at the moment just hand waving
thing. So, this equation is important in two aspects ok: one is that v average velocity how
does it scale with temperature, how does it change with temperature.
Proportional to root T; so, that means, that viscosity is going to be proportional to root T.
So, that means, as you increase the temperature of a gas the viscosity is going to go up not
decrease ok. So, for most of the liquids what you find is that when you increase the
154
temperature you will see that things are becoming loose and then viscosity would go down,
but viscosity of a gas actually increases with temperature and that is an important thing
because many times in factories and so on you would want to you know heat gas and do
things and increasing the temperature is only going to increase the viscosity of the gases
and now decrease it, but for liquids is the other way.
So, that also means that the picture that we talked about is not really or the derivation that
we have done is not really true for liquids. The mechanism is similar there also. There is a
momentum transport the only difference is that the liquid molecules do not undergo these
so much fluctuations. They do not collide each other and go away somewhere else. So,
that picture is not really correct and that is why we are not getting the right expression for
viscosity, but the momentum transport is still there and the connection between momentum
transport and sheer stress all those things is still true for liquids.
We assume that the molecules are transporting they are undergoing collisions you know
basically transporting momentum. Even in case of liquids what happens is that you have
large number of molecules and those larger because of the large number of molecules they
do not really go somewhere else. They basically have a very vibrational kind of motion.
So, that is the dominating mechanism which you will have to take into account if you want
to really derive an expression for viscosity of liquids which people have not been able to
do so very clearly, but there are some approximate theories which can now give you that;
so, the fact.
Student: There are another model for that?
There is another model for liquids yes, but the fact is that the momentum getting
transported because of the molecular mechanism is still true and that is the picture that is
the take home message for us. The other thing is that this viscosity so, viscosity comes out
to be proportional to root T. In fact, if you substitute this λ here μ is independent of n, you
can see that right. So, μ is independent of n.
𝜇 ∝ √𝑇 ; 𝜇 𝑖𝑠 𝑖𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑜𝑓 𝑛
Another way of saying is this μ is independent of pressure once the temperature is fixed.
So, assume you have fixed the temperature; you are only changing the pressure ok. You
would change both number density and mean free path simultaneously. So, that viscosity
155
will have no effect. So, the pressure so, the viscosity of a gas is typically independent of
pressure that is other and these things have been seen experimentally.
In fact, if you look at when you do calculations and if you are actually the fluid that is
under consideration is a gas then you do not have to really worry about what is the pressure
that you are working with ok, unless the pressure is extremely small or extremely large.
So, that is another important thing that you can keep in mind while doing calculations. So,
yeah I think that is what essentially I had to say. Any questions? So, then we will meet on
Friday.
156
Fluid Mechanics
Prof. Madivala G. Basavaraj
Lecture – 19
Device for measuring fluid viscosity and flow behavior
If you look at any textbooks in fluid mechanics right, you will see that you know
irrespective of who has written it ok. You will always see these different class of problems
that have been discussed ok. Now, I think Sumesh has already discussed about you know
your pipe flow right. Plus we have also done; you know, this is a concentric cylinders right.
You have two cylinders one inside the other. So, that problem also has been done ok. And
then you will also see a problems where, there are two plates; that means, you know you
have one plate at the bottom one plate at the top ok. And then the top plate is rotating. So,
there are cases, where you kind of come across problems like this ok.
In the case of constructive cylinders what happened, whether you had a cylinder got an
inner cylinder. And there was a way of rotating your inner or outer right depends on what
kind of devices you want to deal with and. So, this geometry is called a parallel plate
geometry. What you have is you have two plates ok.
157
And I actually have an example of something like that. So, this is if you are interested in
know a little bit more about it you can come to our lab. So, we have these devices. So, this
is actually a kind of a shaft to which at the bottom you have a plate right. And then that is
there is going to be a bottom plate, I am just going to keep you know this configuration.
And your fluid is going to be filled between the plates ok; the bottom plate and the top
plate ok. And there is a way of rotating this ok. That is you know this parallel plate
geometry ok. Yeah ok, that is your parallel plate geometry ok.
Another example is something called as a cone plate geometry. So, again you know there
is a; again a similar device like this instead of a flat plate here, it will be a tapered cone ok.
And again there is a bottom plate ok. So, these are four class of standard problems that you
do in different books ok. pipe flow, plate plate, cone plate and concentric cylinders ok.
Now, when you do these problems ok, I mean all of you know the classic way to approach
these problems right. You right you know momentum balance you know all of that right.
So, you kind of know how to get the flow fields you know all of that right.
Now, when you do problems like this, what is your thought? Do you think these are very
academic oriented problems or do you think there are some applications these things, there
is some applications ok. What would those applications be? How do you think we can use
some of these geometries that, you know you guys learn you know in every course you
know or every course in fluid mechanics how do we use them ok. That is going to be the
158
remainder of you know the today’s you know lecture ok. So, I am going to show you some
examples of, how we can actually exploit you know some of these things for measuring a
few things ok.
So, that comes to you know the measurement of viscosity ok. So, of course, there are
several devices for you know measuring several ways of measuring viscosity of fluids ok.
in my opinion one of the most versatile, you know instrument which can be used for
measuring viscosity. Not only viscosity a lot of flow properties of materials ok. Because
if I take for example, like say syrup ok, versus ointment right. They are very different right
ok, one easily flows other one does not flow right.
Now, when you have complex materials like that ok you will you know you will see, that
you know I cannot use something like say devices that are typically meant for measuring
viscosity of fluids cannot be used of course, for the measurement of you know flow
properties of the ointment right. So, in that context rheometers are one of the most versatile
devices, which can be used for a range of samples all the way from liquid to solid like you
know for example, toothpastes and ointments and stuff like that.
So, and of course, there are lot of different flow geometries that people use, but three most
commonly used geometries are a kind of you know mentioned here, that is your parallel
plate, cone plate and your concentric cylinders ok. Again if you want to know how to use
159
it if you want to see the instrument you can of course, drop by the lab we can show you a
few things as well.
Let us just look at now. So, when I say that you know these geometries are kind of used
for measuring flow properties, there is some you have to do something you know what I
mean by that is that. So, Sumesh talked about, he showed some videos right. You know
the videos were, if you had like say concentric cylinders there is fluid in between them
right. You know when you are rotating you know you talk about Taylor number right if
when you are rotating one of the cylinder at a very low you know rotation speeds you
know it looked like the flow was all laminar right and at some point there were some
instabilities and stuff like that right.
Now, and of course, when you do flow problems related to this ok. we said that you know
you kind of assume some conditions that you know whether your v r is 0 or v 𝜃 is 0 you
know you have to figure out right. Now, when you work with these geometries under
specific conditions ok what I mean by specific conditions is that, if you choose geometries
of appropriate dimensions ok.
Then it turns out that, the flow field that you generate in the fluid that is in the gap ok.
Between these geometries is very well defined ok. And for one to use rheometers that is
the most important criteria ok. What I mean by that is, if you are trying to use concentric
cylinders ok you are; if you talk about R i which is the radius of the inner cylinder and R
o which is the radius of the outer cylinder. It turns out that know all these well defined
flow fields are going to be you know generated, if the ratio of Ri you know versus Ro is
very close to 1.
That means that is at the limit of a very very narrow gap ok. So, therefore, the point that I
want to make is that you would have to work with geometries of appropriate dimensions
ok. Only then you develop a flow field ok, which can be described by some of the math
that you have done in the class ok, that is most important thing ok.
160
So, and typically these measurements are done as I said right, there is a basically a shaft
ok.
And so, only thing that you control during your measurement is that, you basically control
the rotation speed that is all you do ok. There are some ways by which I can tell the
instrument and I can say; hey you rotate it at a particular rotation speed ok. That is the
control that I have ok. And in terms of the measurement ok, what the instrument basically
measures is it measures something called as a torque ok. There is something called as a
161
torque sensors ok. Whenever some things are rotating right, you can think about you know
the generation of the torque ok. This instrument has something called as a torque sensor
and it basically accurately picks up what the torque that the instrument is applying ok. And
this is common for any geometry, whether you use a parallel plate you know or a cone
plate or a quad geometry ok.
So, the configuration you know is clear to you guys right. So, that is you know your
rotation speed right. And M is basically the torque that you are you know; that is being
measured. And of course, Ri is the inner you know radii in inner cylinder radii and Ro is
outer cylinder and that is your length ok. So, this is a. So, basically this is a geometry of
well defined dimensions ok.
That is all I wanted to say now. So, now, when you have a some cylindrical object ok,
when it is rotating ok.
162
Now, what I can actually obtain is, what is the shear stress ok. What is the shear stress that
the wall of the rotating cylinder is facing right or you know what is the shear stress? That
this your rotating object is feeling right and so now, if I know. So, you know that you know
this is your torque ok. Torque divided by distance will give you the force right ok.
Now, your τ R θ is the shear stress that is being applied onto the your cylindrically rotating
geometry right ok. Because it is a function of R and θ right it is independent of Z right. No
matter you know, because that is your Z axis right and it is rotating like that ok. So, what
you are basically measuring is at the wall of the rotating cylinder ok. Your shear stress
multiplied by you know the dimensions ok. It is going to give you your force ok. Therefore,
if I basically know if I because I am measuring Mi which is the torque that I am applying
I have a way of calculating, what is the shear stress right. Because I know what I am
applying, I know the dimensions, I can get your τ R θ.
163
Now, you worked out the; you know your in the previous couple of class ago right you
worked out you know expression for v θ right. Yes or no you did right and you kind of got
this expression that you know it depends on you know there is a mistake here right. So, it
has to be capital R so right. So, basically you know it depends on your rotation speed you
know our inner you know radii and outer radii and you know your radius right is a function
of R right.
Now, you can go back and actually find out that you know. So, whenever you do a fluid
flow problems you are going to think about shear stresses right. And shear stress is a tensor
there are nine components and depending upon the problem that you are dealing with there
are some are going to be 0 some are going to be nonzero ok.
So, what I have done is I have taken an expression for τ R θ you know from one of the
standard books ok and because I know v θ ok. I know that this is 0 right. So, I can actually
substitute for v θ, I can actually get τ R θ ok. I am going to give these slides. So, you do
not have to write it down So, I can substitute for v θ and I can actually calculate τ R θ ok.
And τ R θ right it goes as mu times you know all these parameters.
Now, I can evaluate this τ R θ either Ri or Ro depending upon inner one you know whether
I am rotating the inner one or a outer cylinder ok. I can actually evaluate τ R θ right. Now,
because I have already know, what τ R θ is in terms of the torque, I can substitute that into
this equation. I can actually measure what is your viscosity ok.
164
So, all that is done in these devices is that there is some way of correlating the torque that
the instrument measures ok, into the shear stresses ok. And if I know the shear stresses, I
can actually back calculate what my viscosity is.
And similarly you can also again, you know your shear rate would also depend on you
know the your average radius you know the rotation speed and you know inner and outer
cylinders.
So, basically what I can do with these devices is, I can take a fluid ok. I can put them in
the gap between you know the cylinders are the bottom plate on the top plate, I can just
change your you know the rotation speed. I can change both the shear stress and I can also
change the shear rate. I can measure viscosity yet you know every value of shear rate and
shear stress. I will get information about how does the viscosity is changing as a function
of shear rate or shear stress ok. That is what is done in a in rheometer ok. Any questions?
165
So, there are other devices right, I mean rheometer as I said is one of the most versatile
devices for doing you know, what is called as a rheological studies or flow behavior of
materials ok. And there is also something called as a falling ball viscometer and something
called as a capillary or a U-tube viscometer.
In this capillary viscometer, what is actually exploited is the fact that you have a thin
capillary. And through which a liquid is flowing and that is a classic case of flow through
pipes right. So, basically whatever equation that you develop in flow through pipes, is
going to be kind of useful here.
166
In the falling ball viscometer what you basically have is a liquid columns. So, that is you
know; that is you know a spherical object ok typically a steel ball or some other material.
And that is there is a column of liquid, that is the liquid is filled in this column and of
course, it has to be a constant temperature bath, because you know that in a viscosity is a
function of temperature as well right. So, right and then what you do is, you make this
object fall in the fluid.
167
And of course, you are going to learn a little bit more about it, whenever you have a fluid
that is going to that is falling in a you know a column of, whenever you have a an object
which is falling in a column of fluid and if the fluid is stationary ok.
So, you can think about writing a simple force balance ok. And there is a net gravitational
force acting on the particle, which is your gravity minus buoyancy and there is a drag force
ok. So, you can put up expression for each of these things and. So, therefore, you can
actually ultimately get an expression for something called as a terminal settling velocity
ok.
That means. So, the moment you drop this you know object in the fluid initially there is
some acceleration ok. At some point that ball is going to move at a constant velocity. And
that constant velocity is basically governed by the balance between the net gravitational
force and your drag force. And we know that you know your drag force depends on your
viscosity right. As it falls down the presence of a liquid, basically slows down the ball right
And therefore so what you can do is I can get this terminal velocity and if I have some way
of measuring terminal velocity. If I know what is the radius of the or the diameter of the
ball I am using ok. If I know what is the density of the particle density of the fluid; if I
have some way of measuring u t I can actually get the viscosity ok. So, it is a simple
concept ok. But people still use, I mean there are commercial instruments available which
kind of exploit this technique ok.
168
So, and it turns out that are this particular you know approach you can only use for cases
where, you know the expression that I took for drag force ok. That is what is called a
Stoke’s drag ok. And that is only true you know if your Reynolds number is less than 1
ok.
So, therefore, this only works, if you are working with you know very small particles or
very small objects ok. If I take like say meter sized you know spherical objects if I make
it fall it is not going to work, because you know you have your diameter of the object in
the numerator. Its going be really large number ok, you are this condition is not going to
be satisfied therefore, you would have to work with appropriately small you know
dimension objects ok. And then you would have to ensure that you know, it is in the what
is called as a laminar flow conditions ok. And that is when you know you can use this.
And how do people measure. So, what they will typically the instrument would look
something like this right, I think I have shown this here. So, there is basically the central
column that you see here right, that is your fluid column.
Now, there are some graduations in the column ok. Now, it turns out that when you want
to measure you are going to have this experiment in the lab next semester ok. The objective
would be you drop a ball ok and you look at the velocity of the ball at different sections in
the column. It turns out to you know if you plot the velocity as a function of time ok, it is
169
going to increase initially and its going to remain constant right that is when your terminal
velocity is reached right.
So, therefore, I would have to; have some way of measuring velocity of the ball that is
falling at some successive positions ok. So, they are basically there are markers ok. there
are markers like that, that is a you know graduation that you see in the ball ok. Now if you
are what you do is you have basically identify at, what is the instant of time at which the
ball intersects this ring ok.
And what instant it basically travels and you know goes outside this ring ok. Therefore, in
that sensor, it will travel the distance that is equal to the diameter of the particle in some
time right. So, therefore, I have some way of measuring, what is the distance it travels ok.
And what is the time that it has taken for you know a traveling that distance right.
So, basically you know this expression ok. And you know I can measure what is the
distance is travels and that is your length, time right. And because I learn these are all
constants ok. So, everything is constant I can actually write it as a K times rho p minus rho
times t ok. Now, your rho p is it is kind of kept here, because you know depending upon
the kind of fluid that you going to work with ok. You may be working with you know
objects of different materials you can have a steel ball, glass ball and things like that ok.
So, in that sense you know you can. So, basically your viscosity you know is actually some
constant K that is something called as a ball constant ok. It depends on the properties of
the ball and things like that and ok. So, you measure the time and you basically get your
viscosity ok. That is how the falling ball viscometers work.
170
So, in the capillary viscometers, what you do is you have some blobs right, something like
that here alright. So, what you do is, you basically fill these blobs with the fluid of known
volume and then you basically find out, what is the time that it takes for draining out ok.
So, again the working equation is something like this again you can derive this. So, your
V by t, which is the volumetric flow rate. Can anyone recognize this equation does it look
familiar to you?
Student: (Refer Time: 21:25).
171
Was that?
Student: Poisson equation.
Correct right; Poisson equation; now I do not know if. So, we did this in the class. you
kind of derived the velocity profile for a flow through pipe right. Now, if I know what is
the velocity profile, I can there is a way of getting your flow rates right. You would have
to you know integrate your integrate the velocity profile multiplied by you know your area
right. So, you can basically get your flow rate ok. That is your V by t ok.
So, volumetric flow rate. So, now, if you look at this equation there are actually two things
right one is your Δ p plus your rho gl that is there are two contributions to the pressure
drop right. One is your, because you know you typically these experiments are done in a
vertical column therefore, I would have to also worry about your hydrostatic pressure as
well, that is why you have this term ok.
Now, when you are working with you know capillaries ok. Your pressure drop across the
pipe becomes important, when you are working with pipe that are really large in size right.
When the length of the pipe is really large your Δ p becomes appreciable, but when you
are working with a capillary, when you are working with small distances actually I can
neglect that ok. And I can say that you know the pressure drop purely comes from you
know your hydrostatic you know fx therefore, your V by t basically goes as you know rho
gl times you know the pi R c power four that is the R c is your raised to the capillary and
you know your viscosity in the length ok.
If I know the length of the capillary, if I know the diameter of the capillary, if I you know
I have a way of measuring, you know your flow rates ok. So, basically I can substitute all
of this and I can basically get your viscosity.
So, what people typically do is, you can express this as a some instrument constant times
rho times t ok. Where instrument constant, what you do is if I have two fluids ok, I can
calibrate my instrument, I can actually get the you know the instrument constant ok. So,
yeah that is what I wanted to say today ok.
So, basically the point is that there are different fluids different and you know depending
upon, what kind of fluids are you dealing with, they exhibit different fluid properties ok.
172
And there are devices for measuring you know viscosities ok. And you should be aware
of some of these things and some of these things you are going to use it in your next
semesters lab ok. So, with that I will end, if you have any questions, we can take them.
173
Fluid Mechanics
Lecture – 20
Fluids properties and its behavior
So, Sumesh is not here today, so I am just substituting for him. So, we are going to talk
about I mean I kind of titled this you know this particular lecture as something about
different fluids viscosity and its measurement. A lot of contents that are you know kind of
compiled from different sources, the references you can see that ok. If you want to know
more about each of some of the things that are going be discussed today you would have
to go to those sources and then you know a little bit ok.
We will just begin by just defining what is viscosity ok. I think Sumesh in the last class
about the viscosity of the gases right and you know talked about their T power one and
half relationship that you get for gases ah. In a Layman’s term you know you could just
say that you know viscosity is def you know you can think about it as a resistance to flow
right that is a Laymanish definition of viscosity. And, in SI units, what is the unit of
viscosity? Yes.
What is that?
Student: Pascal second
It is Pascal second right. It is Pascal second and so, you see a list here basically there are
different fluids and the viscosity of each of these fluids are kind of listed there.
174
You know think about water has a viscosity of like 1 10 power minus 3 Pascal second ok.
And, so now, there are two columns here right. There is one is viscosity another one is
density if you take a closer look at these in the numbers for you know different fluids it
turns out that you know it is not necessary that something that is highly dense need not be
more viscous right.
I hope all of you know this fact that, for example, if I take a look at say mercury for
example, the density is really high right because its viscosity it is just about that of water
right whereas, the density of water is about 1000 right. So, therefore, your density and
viscosity you know they do not go you know hand in hand ok. So, it is need not be that
you know something that is more dense is more viscous that is really not true.
So, if you want to tell somebody that when people say that you know viscosity is a kind
of a quantitative measure of resistance have you kind of experienced you know if you want
to tell you in a in a practical way, how would you tell somebody what is viscosity? You
could think of like if like say a simple case of running right. If I take two cases, one is I
am running in water other one is running in air right the fact that you know it is much more
difficult for you to run in water ok you can think about this in terms of the viscosity right,
the resistance that the fluid offers.
And, actually that is kind of quite clear if you look at the viscosity values if we take water
is 1E-3, if you take air it is 1.8E-5 ok; the ratio of viscosity is about 50 ok. The fact that
175
you know the water is definitely 50 times more viscous than you know air basically you
know gives you this additional resistance which because of which you kind of run much
much slower in spite of putting the same effort right. So, that is you know you could think
of you know arguments like that to really you know think about the resistance that you
know that a fluid offers right.
Now, there are different types of fluids I am sure all of you have kind of a learned this at
some point for sure. So, there is actually a plot of you know your shear stress as a function
of a shear rate and then there is another plot next to it which is your viscosity is a function
of you know shear rate right. Do you know what these you know do you know typical
class of fluids that kind of obey such behavior your what is that?
Student: Newtonian fluids.
Newtonian fluids right. All of you have heard of Newtonian fluids and you know you know
your stress is proportional to you know shear rate plus if you measure viscosity is a
function of you know shear rate it turns out that you know the viscosity should be a
constant right. So, all of you know about this. If you are not familiar with the terms you
will see that you know a lot of people use you know your γ dot that you see here right γ
dot you can actually you can call it as rate of shear shear rate sorry, shear rate strain right.
Sorry, you can call it as a shear rate or you can also call it as you know a shear strain rate
that kind of comes from the simple argument right.
176
If you have this dv by you know dy which is the velocity gradient ok. I can express this dv
as a as dx by dt right, distance by time ok. The change in length by original length, I can
actually express it in terms of strain and that strain divided by time is going to give you
the strain rate ok. So, your γ dot people either call it as strain rate or you can also call it as
a shear strain rate as well ok.
𝐹 𝑑𝑣
𝜏= =𝜂 = 𝜂𝛾̇
𝐴 𝑑𝑦
You will you will see there are different books use different kind of terminology and you
would have to be a little bit aware of these terms and as you know this is a classic case of
Newtonian fluids ok.
And, it turns out that you know a lot of things that we use they are not Newtonian type
there are different.
177
So, before you go to different kinds of fluids I just want to make a point about something
called as a shear rates that are typically kind of exist in different processes ok. So, what
you see is the first column it is a it lists processes. The second one is what is called as a
typical range of shear rates and the third one is application ok.
So, it turns out that you know depending upon the processing conditions your shear rates
are going to be very very different ok. What I mean by that is if you have like say a cream
that you are trying to apply to your you know your body for example and when you put
your cream onto your you know and the body and then if you are trying to rub it the kind
of shear rate the material feels is very very different ok, then maybe putting an oil onto
your hair for example, ok. So, therefore, so for example, a typical you know there is a list
called rubbing you know the typical shear rate would be 10 power 4 to 10 power of 5
second inverse, that is really fast ok. That is really doing a you know very fast shearing
ok.
And, there are also applications, for example; if you look at sedimentation of you know
fine powders you have a column of liquid under you know the particles is settling ok. In
such cases you know the shear rates are very very low, 10 power minus 4 to 10 power
minus 6 ok, that is too low.
178
So, therefore, there is also need to look at different fluids and then have an idea as to how
does the material respond to different shear rates ok. So, and again it depends on the kind
of fluids that you are dealing with and stuff like that ok
Now, again I think I am sure all of you know this particular plot that there are typical
material flow behavior that fluids exhibit. The top two plots are basically plots of you
know shear stress was a shear rate both of them basically taken from different sources and
I am sure all of you are familiar with these terms right; dilatant fluid, pseudo plastic, you
know yield stress fluids and stuff like that ok. I hope all of you know these terms right.
No? Do not know? ok These are fairly simple ok.
So, now let us think about toothpaste ok. Every morning you just go on take a toothpaste
and then you know if you open the lid and if you to want it to fall by gravity it is never
going to fall right and of course, if you have you know if I take like say again there are
different types of toothpaste, but typically most toothpaste if you want it to come by itself
you know it would not right under the action of gravity it would not. So, for that for that
to come out you would have to apply some kind of pressure right ok. Now, such materials
what happens you know they would not come out until you apply a particular force or you
can think about it in terms of a until you apply a certain stress they would not come out
ok.
179
So, if you look at you know the class of flow is represented by these two fluids you clearly
see that you know I would have to apply a nonzero stress for the materials to start flowing
and such fluids are called as yield fluids or it is also called as a pseudo plastic fluids and
stuff like that ok. So, therefore, if you want them to flow they would have you would have
to apply some minimum stress for them to start flowing ok, those are some class of fluids.
Of course, you have this you know the Newtonian fluid right. Again, it is very similar to
what we saw in the previous plot your stress versus shear rate is a linear variation that is
your you know Newtonian fluid. And, you can think about when you whenever you have
this stress versus shear rate plot you can think about slope of these plots ok. At any instant
of you know shear rate you can think about it you know it as giving a resistance and as I
mentioned that resistance is you know you can think about that in terms of viscosity ok.
Therefore, if I take you know a Newtonian fluid like here you know your slope is constant
therefore, if I were to measure viscosity the viscosity of my fluid is going to be constant
right because you know my slope is constant ok. Now, if you take other type of fluids your
viscosity could be changing as a function of shear rate for most for all non-Newtonian
fluids your viscosity is going to be a function of your shear rate ok.
Now, depending upon whether the viscosity of the fluid is decreasing if you increase the
shear rate if that is the case such fluids are what is called a shear thinning fluids ok. These
are the fluids where if you take these fluids and if you start shearing them the viscosity
goes down with shear rate such fluids are what are called as shear thinning fluid. On the
contrary, there are fluids where the viscosity could increase as a function of shear rate such
fluids are what are called as a shear thickening fluids that is. So, basically your dilatant
you know is also same as shear thinning and pseudo plastic is same also same as shear
thickening.
I think it is the reverse; I think it is a reverse, so I am sorry about that. It is the reverse right
because if I take a slope of these plots, now if I take say pseudo plastic for example, right
slope is initially constant. Now, if I increase the shear rate strain my slope is going to
decrease right that is it is a shear thinning fluid. Sorry about that, that is your shear thinning
ok. And, if I take a dilatant fluid if I take slope at different points with shear rate my slope
is increasing therefore, my viscosity is going to be increasing if I increase the shear right
there what is called a shear thickening.
180
So, just to you know I just want to make a point that you know there are different class of
fluids and their viscosity need not be constant you know. It could be varying as a function
of shear rate and depending upon whether it increases, decreases or remaining the constant
you can basically think about fluids in different you know and the most of the fluids that
you come across in daily life or you know and or in industrial processes they kind of you
know kind of fall into some of these categories ok.
So, now, ok. So, the question would be can you think of; can you think of a fluid that could
exhibit all different behavior, one fluid which can be both shear thinning, shear thickening,
a Newtonian? So, can you think can you name some example of a Newtonian fluids? What
is that? A lot of fluid that you use right; water, you know honey for example, you know
all of that right. So, for them you know if you measure viscosity as a function of shear rate
it is going to be a constant value they are all Newtonian fluids.
Any example of some fluid where you know the viscosity need not be constant? Have you
thought of some example? Should go back and take a look at some nice videos on YouTube
ok. There is a nice video of you take a take water ok, then you mix what is called as a
cornstarch ok. If you go on adding some cornstarch at some point it starts behaving you
know very some funny way ok.
So, there are there are nice videos where people are made you know like say bathtub you
know made of this fluid and you can actually walk through the fluid that is because you
know at the instant of if you are running through this fluid at the instant of contact between
your foot and the you know bathtub the viscosity goes are very high and that can actually
support the whole weight of the person who is walking on the who is running on the on
the back ok.
181
So, a classic case of fluids which exhibit all this you know properties in one you know in
the same fluid is an example of something called as a colloidal dispersions or nano particles
dispersions ok. Simple way to think about this would be these are slurries; so, liquid with
some particles. The particles could be very small in size nano particles it could be micron
in size or it could be even larger macroscopic particles ok. So, what you are seeing in this
plot is actually it is a plot of viscosity as a function of shear stress. You could also have a
instead of shear stress I could have also plotted a shear rate as well ok.
What you should note is that for each plot that you have there is a parameter that is defined
that is called phi and that phi is what is called as a volume fraction of particles in the fluid
ok. Phi is equal to 0 corresponds to a pure fluid without any particles if I say like say phi
is equal to 0.09 that basically corresponds to a case where I have 9 percentage of particles
in the fluid. Similarly, if I say phi is equal to 50, .50 you know your 50 percent fluid and
50 percent particles right.
So, all that has been done here is they have subjected this fluid to a similar test of basically
measuring viscosity as a function of you know your shear stress and as a function of
different concentration what you know is a is a result right and if you look at the low
concentration data that is you know from about 0 to 0.18 you can see that you know the
viscosity versus you know your shear rate it is a flat line ok. So, basically that is your
Newtonian behavior right that is your Newtonian behavior.
182
Now, if you go a little bit higher in concentration you know 0.34 and 0.28 ok what you see
is initially you know your viscosity is constant and that is basically followed by a very
small decrease and that small decrease it is an example of a case where you know there is
a shear thinning behavior right. Now, if I go a little bit higher up in the concentration it
turns out that there is of course, a decrease followed by increase at a higher shear stress
that is an example of a case where you have both shear thinning in this region and shear
thickening in this region ok.
So, and this phenomena of shear thickening actually has been exploited in a lot of
applications. So, I kind of you know given you some applications that one common
example is a body armor ok. So, whenever people have this bulletproof you know jackets
what is there are recent developments where in between different layers of cloth what they
put is you know these fluids which when you know a bullet penetrates at a very high you
know force, it basically in locally solidifies the material and thereby stopping you know
the you know the bullet from penetrating into the vest ok.
So, there are applications of such material, so I just wanted to you know give you a little
ok. So, now, there is a there is an understanding as to why this happens yeah ok. So, the
common you know the picture that people kind of think about is that the case where you
have your shear you know your Newtonian behavior is a case where you know you have
kind of a randomly distributed particles in your fluid and you are subjecting them to you
know your shear flow right.
And, then if you go for a higher shear rate or shear stress, it turns out that there is some
kind of a ordering and kind of particle from these lanes lines of you know particles and
these lines of particles essentially you can think about them as offering less resistance for
the flow ok. One way to think about this would be you know a randomly moving traffic
versus traffic in a lane ok.
So, now, if you go higher up in the shear rate or shear stress what happens is you know
you are basically forming what are called as particulate clusters when you are trying to
you know kind of shear them at a very high shear rate all the particles are kind of brought
together and you kind of form locally these clusters and these clusters kind of interlock
and this interlocking of these clusters in a way leads to enhanced resistance there by the
183
viscosity goes up ok. So, this is a kind of understanding that people have about such fluids
ok.
So, that is about different types of you know fluids you know just to give you a picture of
you know different class of fluids ok. So, all that we have done so far is we have just said
something about what is viscosity and then you have kind of discuss different class of
fluids and I just took one example of a particular class of fluid where you see the
Newtonian, non-Newtonian you know a behavior in the same fluid right. That is what we
have done.
184
Lecture – 21
Tutorial - 04
Hello everyone; today we will solve the problem on calculating pressure drop through an
inclined pipe. So, first we will see the problem clearly. The problem states water flows
downward in a pipe at 45 degrees and we will draw the schematic first.
So, we have an inclined pipe and we have a point 1 here and point 2 here. To calculate the
pressure drop between point 1 and 2 we are using an YouTube manometer, ok. And this
angle is 45 degrees. So, this is what is given in the problem. And we are supposed to
calculate the pressure drop between point 1 and point 2, point 1 and point 2.
So, the manometric fluid; we have is the mercury. So, the dash to region represents the
manometric fluid; which is mercury and it is certain the reading in the manometric fluid
shows us 6 inch, ok. So, yeah, so let me put this way. The point 2 is here. We know that
manometric, we know that the YouTube manometer works on the principle of hydrostatic
equilibrium, using that; first we will calculate the pressure at the point 1 and at the point,
sorry, at the point 2. From the principle of hydrostatic equilibrium; we know that, the
pressure at point capital I and capital II are same.
185
We will make use of, we will make use of that fact and we will calculate the pressure
difference between the points 1 and 2. For that; first we will calculate, we will write the
pressure at point I. So, the water is flowing in this inclined pipe, so for point I. So, sorry
for point 1 if we calculate the pressure it would be P 1 plus the pressure whatever we have
from.
So, the pressure exerted by the fluid; which is inbuilt in this region, and to do that region
to calculate the pressure in that region; first we will make an assumption that, the distance
between this point and this point here is h and this as x. So, I repeat that the distance
between the point 2 and the manometric reading in the right limb is h. And the distance
between the manometric limb reading in right side and the left side. The difference
between the two manometric limbs heights is given as 6 inches in the problem and we
considered that; the distance between the vertical distance between the point 1 and point 2
is x.
So, if we look at the left hand side; the fluid, which is above the manometric limb will give
us a pressure on the point capital I here. So, P 1 is the pressure at point 1. And we have the
water as a fluid from point 1 to in this entire region we have the water as a fluid.
So, for that; we have to calculate x and we should know the this distance ok. So, to
calculate x; first we will see, we make a right angle triangle, so that x. And again in the
question it is given that; the distance between point 1 and 2 is 5 feet. So, I just remade the
part of this channel here which is a 5 feet distance that is between point 1 and point 2. And
we know that the angle is 45 degrees. So, if we take sin 45, it would be side opposite to
theta by hypotenuse that implies x equal to 5 times sin 45, sorry x equal to 5 times sin 45.
So, now, we know x and we already know this distance because it is just it is nothing but,
6 inch plus h. We will just keep this as h. We do not have to calculate h, we will see that
y.
So, the pressure at point 1 is P equal to the pressure at point exactly, sorry the here it is not
the point 1 its capital I. Pressure at point capital I equal to pressure at point 1 plus the
pressure exerted by the fluid in the region with an height of x which would be 62.4 times
32.2 times x. Where 62.4 is density of water and 32.2 is the g in fps units. And we know
that P equal to rho gh.
186
So, this is the pressure at 62.4 into 32.2 into x is the pressure exerted by the fluid in this
region; which is in the by the fluid, which is at an height of x. And then, we will calculate
the pressure in the remaining section in from the left hand side and it is 62.4 times 32.2
times 6 inch. We will convert that to feets; now 1 feet equal to 12 inch.
So, it is a 6 by 12 feet plus h. So, it is P 1 plus 62.4 times 32.2 times x plus 6 by 12 plus h;
where x equal that implies P 1 plus 62.4 times 32.2 times. x we have calculated it as 5
times sin 45 plus 6 by 12 plus h. Similarly we can calculate pressure at point capital II.
. So, the point capital II does have two fluids above it; water, we will write that. So,
pressure at point II equal to capital II equal to; pressure at point 2 plus pressure exerted by
the fluid with an height of small h that is water. So, we will take the density of water which
is 62.4 in fps units times 32.2 which is the acceleration due to gravity in fps units times h
plus; we have an we have the manometric fluid the mercury for an height of 6 inch.
We will write the pressure exerted by the mercury of height 6 inches. For that, we need to
use the properties of the mercury; which is the density. So, the density of mercury is 846
times. So, we know that the acceleration due to gravity is 32.2 times and 6 inch. The height
is 6 inch; which is converted to feets and that is 6 by 12.
So, this is the pressure at point II. And from the principle of hydrostatic equilibrium; we
know that, the pressure at point capital I is equal to pressure at capital point II. Pressure at
point capital I equal to pressure at point capital II and.
187
So, what we know is P 1 plus 62.4 times 32.2 times 5 times sin 45 plus h plus 6 by 12.
This is the pressure exerted that pressure exerted by the fluid at point 1; which is equal to
P 2 plus 62.4 times 32.2 times of h plus 846 times 32.2 times 6 point 12.
The LHS corresponds to the pressure exerted by fluid at point capital I. And RHS
represents the pressure exerted by the fluid, water plus the pressure exerted by the mercury
at point 2. So, now, we need to calculate P 1 minus P 2; which is equal to, so yes. 32.2
times 6 by 12 times 946 minus 62.4 times. And this term gets cancelled with this term and
we will left with 62.4 times 32.2 times 5 times sin 45, 45 degrees.
This is the total pressure that we are supposed to calculate. And we have two parts here.
This part of pressure drop is coming from the fact from friction. So, this is called as the
friction head and this is coming from gravity head, ok. And now, for the question if you
see the question clearly; it is asked to calculate the P 1 minus P 2 in terms of lb force per
in square. If you look at the; now we will just look at the units of this P 1 minus P 2. So,
we have 32 minus 2; which is feet ok, I will write it down here. 32 minus, 32.2 is feet per
second square times 6 by 12 is actually it is, it was in inches and we have connect converted
that to feet and 846 minus 624.
These corresponds to the densities of mercury and the densities of the water. So, the density
is in fps units; it returned as lb for feet cube. So, finally, we have ended up with lb per phi
times second square. But, we were we are asked to calculate the pressure difference in
188
terms of lb force per in square. To get lb force per in square from lb per phi time second
square.
We will multiply this with g; gravitational acceleration due to gravity which is feet per
second square. And finally, we will end up with lb force per sorry, so we will divide it
more, we will divide it with feet per second square. So, we will end up with lb force per
feet square, yes. Because, 1 sorry, because 1 lb force equal to 1 lb times g. So, we already
have lb here, so we divided it with g; that is a unit the units of g are feet per second square.
So, we have got lb force per feet square.
So, in this, if we in the above equation I; if we divide I with g in fps units. We will get P 1
minus P 2 in terms of lb force per feet square. We already, we have said that 32.2 is a value
of gravitational due to, acceleration due to gravity in fps units. So, we will have P 1 minus
P 2 equal to 6 by 12 times 846 minus 62.4 minus 62.4 times 5 sin 45 degrees. This comes
out to be 171 lb force per feet square yes.
So, we have got P 1 minus P 2 as 171 lb force per feet square. But, in the question; we are
asked to get the pressure difference in terms of lb force per in square and, yes. 1 feet equal
to 12 inches. So, lb force per 12 square 144 times in square which is equivalent, which is
1.1875 lb force per in square. So, the pressure difference P 1 minus P 2; that we are
supposed to calculate is 1.1875 lb force per in square. So, this is the final answer; P 1
minus P 2 equal to 1.1875 lb force per in square.
189
And later in this same question; we are asked like, what is the pressure drop due to friction
only between 1 and 2 in lb for force per in square? So, the yeah, here we said; we have the
pressure difference in the system, because of friction, which is termed as friction drop.
This part corresponds to the pressure drop due to friction between 1 and 2. Does the
manner, and next we are ask; does the manometer reading correspond only to friction drop?
So, so P 1 minus P 2; due to only friction drop will be this term 6 by 12 6 by 12 times 846
minus 62.4 yes, that is lb force per feat square. Does the manometer reading correspond
only to friction drop, why? You know, the manometer reading corresponds to both friction
drop as well as the gravitational height. So, that is about the problem.
190
Fluid Mechanics
Lecture - 22
Choice of Scaling parameter
First let us just do it and then see what it means ok. So, I am going to suggest so how many
parameters are there in; how many variables are there in this problem?
S, ok S0, u, t, g there are 5 things right, suppose you did not know this equation or you
want to represent this equation, but as a data set. Not as an equation what would you do?
Your interest is to prescribe the location of the particle as a function of time. So, if you
were to plot a graph, you will actually plot is s versus t right.
You will plot s versus t as a graph, but then you have to tell for what value is s 0 what
value of u and what value of g you would be plotting it? The moment you say change s 0
u and g the plots are going to be different agreed. So, you can say
1
𝑆 = 𝑆0 + 𝑢𝑡 + 𝑔𝑡 2
2
𝑆 → 𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒
191
𝑡 → 𝐼𝑛𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒
𝑆0 , 𝑢, 𝑔 → 𝑃𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟𝑠
The moment you change the parameters you are going to get different different plots. So,
this may be for a given set of parameters s 0, u and g ok, let us say 0, 1 u 1, g 1 you might
get it for a 0 2 u 2, g 2 and so on for each of them you are going to get different different
plots ok.
Or if you were to do experiments, without you knowing this relation, you would have
change s 0 g and u multiple times get this plots ok. So, you have got in this particular
problem 5 you know really variables out of which you have identified a dependent variable
and an independent variable and 3 parameters ok.
Out of which I will say we will select two scaling parameters select two scaling parameters.
Now, what is meant by scaling parameters? Is essentially you going to use the scaling
parameter to non dimensionalize your other variables.
So, remember when we were doing the pipe flow, we non-dimensionalized our equation
in some fashion we chose a characteristic length we chose a characteristic velocity ok.
Similarly we are going to select certain things and then you are going to use that to non
dimensionalize your equation. So, here we do not know how many we should have actually
chosen at the moment. Let us do not worry I am just telling you that let us select two of
192
them. So, in this case I have a s 0, u and g has the parameters we have to select two of
them, so let us say in case 1, select s 0 and we use u as the scaling parameters ok.
So that means, I will define:
𝑆 ∗ 𝑡
𝑆∗ = ;𝑡 =
𝑆0 𝑆0 /𝑢
So, our original equation was:
1
𝑆 = 𝑆0 + 𝑢𝑡 + 𝑔𝑡 2
2
∗
𝑆0∗
1 𝑆 2
∗ 0
𝑆0 𝑆 = 𝑆0 + 𝑢𝑡 ( ) + 𝑔 (𝑡 )
𝑢 2 𝑢
1 𝑔𝑆0
𝑆∗ = 1 + 𝑡∗ + ( 2 ) 𝑡∗2
2 𝑢
1
𝑆 ∗ = 1 + 𝑡 ∗ + 𝛼𝑡 ∗ 2
2
That is equation that you are ending up with right. So, α has come out has a non-
dimensional parameters, so it is as α dimensional or non dimensional it is g times s 0
divided by u square ok.
193
So, that has non-dimension, so that is the non dimensional parameter that is actually
characterizing this equation; in which case you would have been able to just plot s * as a
function of t * for just one parameter α ok. That tells you what the curves curve is you do
not have to worry about 3 different parameters now. So, you have actually simplified your
description agreed ok. Now, we have been able to do this because we selected two scaling
parameters right.
So, what did we do we selected two scaling parameters one was s 0, other was u and then
we then we non dimensionalize the equation. Let us say you are going to select a second
case, Case 2:
𝑆 𝑡
𝑆 ∗∗ = ; 𝑡 ∗∗
=
𝑢2 /𝑔 𝑢/𝑔
194
On non-dimensionalization we get:
1
𝑆 ∗∗ = 𝛼 + 𝑡 ∗∗ + 𝑡 ∗∗ 2
2
Correct ok. So, this parameter is what we had originally defined, identified as α right. So,
we have really gotten s double * is equal to α plus, t double * plus half t * square ok. Again
we have got a non-dimensional equation ok, which is again parameterized only by their
parameter α ok; though it appeared differently because we have now scaled it differently,
we have chosen different set of scaling parameters ok. We will do one more and then we
will conclude case 3:
𝑆 ∗∗∗ 𝑡
𝑆 ∗∗∗ = ;𝑡 =
𝑆0 𝑆
√ 0
𝑔
𝑢 1 1 ∗∗∗ 1 ∗∗∗ 2
𝑆 ∗∗∗ = 1 + 𝑡 ∗∗∗ + 𝑡 ∗∗∗ 2 = 1 + 𝑡 + 𝑡
√𝑔𝑆0 2 √𝛼 2
195
So, again we have been able to get it, in all cases
𝑆 ∗ = 𝑓1 (𝑡 ∗ , 𝛼)
𝑆 ∗∗ = 𝑓2 (𝑡 ∗∗ , 𝛼)
𝑆 ∗∗∗ = 𝑓2 (𝑡 ∗∗∗ , 𝛼)
So, α is the parameter that has always come out in some form and therefore, α is the only
non dimensional parameter that is in the problem. It does not matter which scaling
parameters you select you will always end up with that parameter the right non-
dimensional parameter it may just come out in a different form. In fact, if you look at in
the first case we had where did α appear in the first case.
In the first case it was really s * is equal to 1 plus t * plus half α g t * square. So, α was
coming in the place of gravity ok. So, you really can think about when you are changing
α there, you are really changing your gravity. So, you can really say that ok. So, there α
really represented how if I were to change gravity what would have happened? That is the
physical interpretation that would have come out. On the other hand in the second case α
came out at the place of the initial displacement. So, α was really playing the role of the
initial displacement.
196
In the third case it would have come out as the initial velocity u ok. So, u could interpret
that non dimensional parameter in any fashion you like because α always contained these
3 things; g s 0 and u ok. And you can actually interpret in multiple ways and that is because
its essentially a non-dimensional quantity. So, how did we define α? Α was defined as g s
0 divided by u square no? What is the other way that we defined this way right.
So, so this contains g s 0 and u square, so I could write this as g divided by s u square
divided by s 0; in that sense we are saying Oh its gravity to some other acceleration in the
system. Or I could have written it as g by u square divided by s 0 inverse, then I would say
that or maybe I should do 1 by α, so let us see here 1 by α I will write it as u square divided
by g s 0.
So, there it would carry the meaning of some velocity square by some other velocity
square. So, all non dimensional numbers you will be able to interpret it in multiple ways
ok, whether you want to interpret it as strength of gravity, whether you want to interpret it
as strength of the displacement anything ok. And what this exercise has shown you is that
it does not matter what the scaling parameter that you select you will always end up with
the right expression. In all these cases we ended up with right expression, even though the
precise form of the equation was different in each cases ok.
So, non dimensionalization will not able to tell you what the precise form is it will only
tell you what are the relevant parameters involved; is that clear? Any questions, should I
repeat anything?
Yeah.
Depending upon what you want to do it right, if you worried about flow here then you do
not have to care about. But, if you have worried about flow between what is happening at
equator and what is happening at pole you might or you want to talk about what is
happening in you know moon and what is happening here you want to talk about right?
So, that is why so you want to call it a dimensional consider a dimensional variable
depending upon your problems.
So, you look at this; this simple equation has now given actually told you everything that
you have to know really about non dimensional parameter and non dimensional analysis
ok; except that we just do not know we got a situation how to do that ok, so that is our next
197
goal. I will just say few things and then, we will stop and will do the rest in the next class;
power law expression.
Given any physical quantity ok, any physical quantity if you look at its dimension, so we
only variable let us say you know fluids and fluid flow any parameter that you are going
to look at let us say x, its dimensions will have the form:
[𝑋] = 𝐿𝛼 𝑀𝛽 𝑇 𝛾
𝐿 → 𝐿𝑒𝑛𝑔𝑡ℎ, 𝑀 → 𝑀𝑎𝑠𝑠, 𝑇 → 𝑇𝑖𝑚𝑒
Is that what is true for any of the properties or any of the variables any of the physical
quantities that you are familiar with?
You are agreeing or disagreeing I am confused; you want to say something go ahead.
So, let us so if you want temperature you want to include that I am going to worry about
things that relevant for flow heat transfer he will keep it aside always variable constant
temperature right. But how do we know whether it is always correct, we know that this is
what we have always observed right velocity, length, mass, force, power, work right. So,
we are going to prove that this is always going to be the case form which we will derive
something called Buckingham Pi theorem. Have you done Buckingham Pi theorem? No
198
ok. So, we will derive that and that is the theorem that we are going to use to get non
dimensional equations, which we will do next class.
199
Fluid and particle mechanics
Lecture – 23
Non dimensional analysis
So, last week you found or you were exposed to what is meant by fluid properties? What
are the various fluid properties that could be important in fluid mechanics? Right; so, if
you look at what we have been doing we took simply situations and then we try to calculate
what are the; how does the flow look like right. The tube flow the flow between two
parallel plates. Then, we talked about how the fluid properties are important and how it is
important in determining the velocity profile.
Now, you know that the equation that you used to calculate the flow profiles they were
really complicated and they were complicated because on one hand they were partial
differential equations then, they are also non-linear equations right. And therefore, you
have to make a whole lot of assumptions every time to simplify the equations and then,
you did get some profiles and you also found that could be reasonable in certain cases, but
they may not be reasonable always. And the way therefore, you should be attacking a
problem you just cannot dependent theory always ok. You need to look at other options
ok.
There are two other options that come frequently and one is to do a numerical simulation.
In other words, you have got a partial differential equation. You have done numerical
methods last semester right. So, you basically use any of those methods and try to solve
these equations ok. So that field is actually known as computational fluid dynamics. So,
that is one of the option.
200
So, one option that we have seen so far was essentially looking at the theory in which we
try to get analytical solutions, but we may not be always able to do that and then, we could
try doing CFD ok; so, the computational fluid dynamics. The other option of course, is to
see what is going to happen directly and that is by doing experiments ok. So, these are the
three options that you will have when, you want to look at a flow problem.
Now, this is straight forward because you have already seen that these two there is an issue;
the issue is that you know let us see you doing an experiment you do it under certain
conditions. Let us say you keep the temperature something you fix take some particular
liquid and do it and then, you find some solution you find some flow fluid, but you do not
know whether that is always going to be the case.
You do an experiment with water. Now, you change that to some oil; you do not know
whether that is what should be expected. While in case of theory you did not have a
problem because the entire problem is defined in terms of certain parameters ok. So, is the
case with CFD ok? When you are doing a numerical solution, you will be doing it for a
certain given numbers, even not actually solving the equations; you will be doing it for
number.
So, from both from CFD and from experiments you are actually going to get numbers not
equations and therefore, this numbers are for a given set of conditions and then you would
know how to interpolate or extrapolate for that matter; that is clear? So, when you are
201
doing CFD or experiments you are you actually going to get lot of data. Now so, when you
get lot of data ok. So, let us look at a simple example and see what that means. Let say,
you have some complicated shape object, some object ok.
And there is a fluid that is going to come with some velocity U ok. So, the fluid will come;
it will go pass this object right and your interest is in finding out what is the force that this
fluid is going to exert on the particle or you could ask the other question how much drag
the fluid will experience because of the presence of the particle right. The object is there,
it is going to put some restriction on the fluid motion which the fluid will experience as a
drag force or you could see that the particle is going to get you know force because of the
moving fluid ok.
Either way you are interested in finding of what is that force ok. So, you are interested in
finding out, what is the force? Let us say exerted by the fluid on this particle that you have
put there ok. You would expect that it would definitely depend upon what is the velocity
with which the fluid is coming. So, you would say maybe that is a function of U, the fluid
properties right. What are the two fluid properties that would be relevant? Viscosity and
density is the property that comes into the picture and then, of course, you know how big
the object is it does not have any shape. Let us say that you know some particular length
which might be some characteristic length of this object ok.
𝐹 = 𝑓(𝑈, 𝜇, 𝜌, 𝐿)
202
Let us say these are. So, we expect that these are the things may be that would be important
and you want to determine the force. What would you do, if you are doing either you know
a numerical solution of equations or if you doing an experiment? You would choose a
particular value of U, a particular value of mu, a particular value of rho and a particular
object to run the simulation or run the experiment and calculate the value of force, right.
Then, you would do that for various combinations right. So, for example, you could keep
U fixed, mu fixed, l fixed; I will change sorry, rho fixed and you will change L. Let us say
you will change L let say ten time; so, that you can get it for large range of length ok. Then,
you would keep mu sorry U mu and L fixed and change rho and let us say you would
change that for another ten times and you would keep doing that ok. So, you will get huge
you have to do huge number of experiments or huge number of simulation, to find what is
really the dependence of force on something. So, if like say for each of them; if you do ten
times, 10, 100, 1000, 10000 experiments or 10000 simulations you will have to do, if you
really want to find out what is the force that is going to look like agreed that is clear.
So, you are; so and the problem is that one is that you have you need huge amount of
resources large amount of time and then, end up with the huge amount of data which you
do not know how to really even represent because you have things which are dependent
on four things and then you basically end up with huge number of tables. So, that is what
is going to happen. So, the question is, is there a better way of doing this and that is what
we are going to do and that is what is you know for the domain or non dimensional analysis
ok. So, we are going to look at non dimensional analysis.
So, we will see how to do non dimensional analysis, but let me tell you what the result is
going to be for this particular problem that we just talked about. For this particular
problem, if you were to do non dimensional analysis, you would find that this equation
can be recast into:
𝐹 𝐿𝑈𝜌
= 𝑔(𝑅𝑒); 𝑅𝑒 =
𝜌𝑈 2 𝐿2 𝜇
𝑅𝑒 → 𝑅𝑒𝑦𝑛𝑜𝑙𝑑𝑠 𝑛𝑢𝑚𝑏𝑒𝑟
So, this is going to be a non dimensional number. What will be the units of F by rho U
square L square? It is non dimensional. This is another non dimensional number ok. So,
203
you get one non-dimensional number is equal to some function of another non dimensional
number ok. So that you can easily represent, if you were to; you know have let us say
graph.
You will just say that ok I will just plot F by rho U square L square as a function of
Reynolds number and then, you will get some data, some curve and thats all needed for
this particular problem. You dont have to worry about 10000 points and so on representing
space because the only quantities that are really going to matter in this case would have
been Reynolds number and force this you know force divided with this particular
quantities; so, a particular form of force becomes a function of Reynolds number alone.
You change fluid doesn’t matter; it is only what matters is only this particular curve which
you would have represented as a particular function ok. This is a nice way of presenting
things; it is very compact and then, you do not need to worry about how to do the
extrapolation or interpolation; clear. So, the question is how does one end up with such a
relation? We may not know, what exactly g is?
The non dimension analysis not going to tell you whether this is you know this particular
form is Reynolds square or is it exponential of Reynolds or this is sinusoidal of Reynolds;
we will not get it. We will know the exactly what it looks like, but we will know what are
the number that are going to be connected by a particular form of a function. And that is
what we care about because that is enough in most of these cases because things basically
204
get into a compact form. Doubts? Few things which you are very familiar with; I will just
till tell principle of dimensional homogeneity ok; that means, given an equation each term
in that equation will have the same dimensions, you all know that very well:
1
𝑆 = 𝑆0 + 𝑢𝑡 + 𝑎𝑡 2
2
The equation describing the motion or the trajectory of a particle right; S 0 is the initial
displacement, u is the initial velocity and a is the constant acceleration and this equation
is going to tell you where the particle is going to be at a given point in time. What would
be the dimension of each of them? S will have the dimensions of length; S S 0will have
the dimensions of length; u t will have the dimensions of length, metre per second into
second; a t square will have the dimensions of length ok. So, all terms basically have the
same dimensions and that is known as the principle of dimensional homogeneity.
How about Bernoulli’s equation
𝑃 1 2
+ 𝑉 + 𝑔𝑧 = 𝑐𝑜𝑛𝑠𝑡.
𝜌 2
So, therefore, each of the terms have the same dimensions and that is one of the principle
of homogeneity. Few other things. So, these are the other things that you will come up in
any equation. Dimensional variables mean some kind of variable that would have
205
dimensions. For example in Bernoullis equation it would have been pressure, density,
velocity any of them ok.
Dimensional constant are constants, which may also have dimensions; for example,
gravity, gravitational constant and so on. Pure constant like for example, the half we had
there; things like that ok. So, these are the three things the typically come in any equation
and pure constant are something which we won’t really have to worry about throughout
the dimensional analysis because we would never worry about that. We would be worrying
about dimensional variables and dimensional constants things which have got dimensions
ok. So, now before doing it ok; we need to worry about one another thing that is choice of
variables.
Or choice of rather scaling parameters; choice of scaling parameters; so, see we are going
to have few variables ok. In this problem of which where, we talked about flow ok. Past
some object; we said it is basically a function of rho, mu, U, L. We need to identify some
parameters to non dimensionalize it ok.
Now, it would not be obvious, what are the parameters that we should choose. And it is
basically, I mean there is no way to really do it. But in order to see what it means let us
take a simple exercise where, we already know the functional flow which is again will go
back to this equation :
206
1
𝑆 = 𝑆0 + 𝑢𝑡 + 𝑔𝑡 2
2
207
Fluid Mechanics
Indian institute of Technology, Madras
Lecture - 24
Non-Dimensional analysis - 2
So the last class right, you we started discussing Non-dimensionalization we said you
know given a number of variables we do not want to deal with all of them. So, what is the
best way to reduce it? Right. So, we took some simple equation and then we started non-
dimensionalizing it and then we found that you know it could indeed be reduced. So, the
expressions the simple expression that we had was that of a you know the trajectory of a
falling particle.
And, if that was actually given as s equal to s 0 plus u t plus half g t square depending upon
what method you adopted to non-dimensionalize it, you could write it typically as a non-
dimensional distance as a function of a non-dimensional time and some additional
parameter, that is all what we had done right.
1
𝑆 = 𝑆0 + 𝑢𝑡 + 𝑔𝑡 2
2
𝑆 ∗ = 𝑆 ∗ (𝑡 ∗ , 𝛼 ∗ )
208
Then I said and in this particular case ok; in this particular case we knew the equation and
therefore, we could go ahead, but typically that would not be the case ok. So, what is the
general way of doing it? Before we discuss that let us look at non-dimensionalization of
the governing equations that we already know ok.
So, we are going to look first incompressible Newtonian fluid and we will do the same
thing that we did for the trajectory of the particle and see what non-dimensional parameters
are going to come out ok. So, that is going to tell you how you know what is the most
important parameter that determines the fluid flow and then we will look at when we do
not know the equations how to proceed. So, we are looking at a very general system , we
have you know some fluid which is flowing and we want to describe the dynamics. We
will write down the governing equations which would include the continuity equation and
the momentum conservation equation.
So, let us just write it down:
Continuity: ∇. 𝑢
⃗ =0
𝜕𝑢
Momentum: 𝜌 [ 𝜕𝑡 + 𝑢 ⃗ ] = −∇𝑃 + 𝜇∇2 𝑢
⃗ . ∇𝑢 ⃗
209
And, you will additionally have boundary conditions which would be you know some
velocities that you must either u is specified which must be 0 or something else depending
upon the problem. So, this is typically what the problem will look like right.
So, now let us so, there are difference between the previous case that we considered and
this is that in the previous case, it was just an algebraic equation, this is a partial differential
equation that is the only difference ok, but otherwise there is no difference. And therefore,
there is no difference in the way you want to non-dimensionalize it also to non-
dimensionalize we need to select some scales. We do not know how many scales we should
choose, but in this case I am going to tell you how many we will select ok. We are going
to say that we will have a characteristic velocity we will have a characteristic length and
we will have some characteristic pressure ok. So, that is what we are going to do.
We will select let L be a characteristic length in the problem, let U be a characteristic

velocity in the problem right. So that means:
𝐿 → 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝑙𝑒𝑛𝑔𝑡ℎ
𝑈 → 𝐶ℎ𝑎𝑟𝑎𝑐𝑡𝑒𝑟𝑖𝑠𝑡𝑖𝑐 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦
𝑥 ∗ 𝑦 ∗ 𝑧 ∗ ∇ 𝑢
⃗ 𝑡
𝑥∗ = ; 𝑦 = ; 𝑧 = ; ∇ = −1 ; ⃗⃗⃗⃗
𝑢∗ = ; 𝑡 ∗ =
𝐿 𝐿 𝐿 𝐿 𝑈 𝐿/𝑈
210
Pressure now there are various ways to do this and we have seen this and it. So, we
discussed it in the last class that it really does not matter what exactly we choose right, we
want exactly we select to non-dimensionalize. I am going to suggest that we will non-
dimensionalize pressure by a quantity ρ U square ok. One way of thinking about it is that
if you look at Bernoulli’s equation; Bernoulli’s equation is really p by ρ plus half U square
plus g z right is a constant. So, you can see that p essentially would be like ρ U square or
p could be like ρ g z; if actually hydrostatic pressure was more important ok.
So, by really selecting p is like ρ U square we are saying that in the problem it is likely
that my pressure will be you know acting like my ρ U square that is the only recently have
selected it ok. So, those are my non-dimensional variables that list so, we can go ahead.
So, this is so, what I what you see on the right hand side is my non-dimensionalization. So
now, we can go ahead and substitute and see what happens ok, how many reduced
variables we are going to get.
211
So, the first equation again is continuity equation:
∇. 𝑢
⃗ =0
⃗⃗⃗⃗∗ ) = 0
(𝐿−1 ∇∗ )(𝑈𝑢
[𝐿𝑈][∇∗ . ⃗⃗⃗⃗
𝑢∗ ] = 0
[∇∗ . ⃗⃗⃗⃗
𝑢∗ ] = 0
All the variables remain as such except that they have become now non-dimensional, there
was no non-dimensional parameter that came up during the non-dimensionalization. Now,
if you remember we had done something very similar when we derived the equation for a
flow between a plates or tube flow one of them where we did the non-dimensionalization.
Does anybody remember? We had a differential equation, we actually non-
dimensionalized it with a characteristic length, a characteristic velocity and we ended up
with something very similar. There again we did not have any non-dimensional parameter
so, this is something very similar.
In other words the continuity equation basically you know whether its dimensional form,
non-dimensional form whatever it is there is no parameter that is going to be prescribing
what you know the mass conservation is. The mass conservation will be just given by that
irrespective of what the flow parameters are, that is all. There is no particular parameter
212
that you can identify which could actually change the way the solutions are going to be,
just based on the mass conservation.
So, now the momentum conservation so, we have several parts we will take one by one.
So, we have an unsteady term then we have a non-linear term, then we have a pressure
term and then we have a viscous term; let us non-dimensionalize each of them and
substitute.
𝜕𝑢
⃗ 𝜕(𝑈𝑢⃗⃗⃗⃗∗ ) ⃗⃗⃗⃗∗
𝜌𝑈 2 𝜕𝑢
𝜌 =𝜌 =
𝜕𝑡 𝐿 𝐿 𝜕𝑡 ∗
𝜕 (𝑈 𝑡 ∗ )
𝑈2 ∗ ∗ ∗
𝜌𝑢
⃗ . ∇𝑢 ⃗⃗⃗⃗∗ 𝑈). (𝐿−1 ∇∗ )(𝑢
⃗ = 𝜌(𝑢 ⃗⃗⃗⃗∗ 𝑈) = 𝜌 ⃗⃗⃗⃗
𝑢 . ∇ ⃗⃗⃗⃗
𝑢
𝐿
𝑈2
−∇𝑃 = (∇∗ −1 ∗
𝐿 𝑃 𝜌𝑈 2)
=𝜌 (−∇∗ 𝑃∗ )
𝐿
𝜇𝑈 ∗ 2 ∗
⃗ = 𝜇𝐿−2 ∇∗ 2 (𝑢∗ 𝑈) =
𝜇∇2 𝑢 ∇ ⃗⃗⃗⃗
𝑢
𝐿2
213
Let us go back and substitute in the equation now. We get:
𝜕𝑢∗ 1 ∗2 ∗
∗
+ ⃗⃗⃗⃗
𝑢∗ . ∇∗ ⃗⃗⃗⃗
𝑢∗ = −∇∗ 𝑃∗ + ∇ ⃗⃗⃗⃗
𝑢
𝜕𝑡 𝑅𝑒
So, what non-dimensional number did we end up with? Reynolds number ok.
214
That is the non-dimensional form of the momentum conservation equation ok.
You can see that there are now non-dimensional variables which are u * t * and p * and
there is only one parameter and that parameter is nothing, but the Reynolds number ok.
Depending upon what is the Reynolds number you can have equation, if form of the
equation will remain the same, but the answer will come out to be a function of Reynolds
number ok; that is why Reynolds number actually plays a very important role.
You know that for example, Reynolds number is very small you get laminar flow when it
is very large you get turbulent flow and so on ok. So, you are actually ending up with
different different solutions based on one non-dimensional parameter. It does not matter
what is the size of the pipe you are using, what is the fluid that you are using. It only
matters what the Reynolds number is, that is what the equation or the non-
dimensionalization of the equation that just said clear. Should I repeat? I am just saying
that we have you know really non-dimensionalized our governing equation, when we non-
dimensionalize the continuity equation ok, it did not give you any non-dimensional
parameter.
We have non-dimensionalized the momentum conservation equation and we have ended

up with just one non-dimensional parameter and that parameter is Reynolds number.
Remember when we had our original problem ok, we had s * as a function of t * comma
alpha right. So, that trajectory of the particle we said alpha determines what is the relation
215
between s * and t *, where s * was some non-dimensional distance, t * was some non-
dimensional time. And, based on the value of alpha alone we could say what where the
trajectory is going to be. So, alpha is the only non-dimensional parameter that determines
the solution.
Similarly, here this is a partial differential equation, the solution is determined only by R
e ok. When you solve it for example, will come out to be a function of R e along with time
and space and the value of R e is that something going to be different right. You can take
one pipe let us say case I and another pipe case II. In each case you can select some
Reynolds number I, this is another Reynolds number II ok. And, you can let us say let a
fluid go in and out and let us say you will select you know different fluids.
So, if you by the way you had selected different fluids you are going to get two different
Reynolds numbers. The flow is going to be different and it will be different simply based
on the Reynolds number nothing else, that is all we are trying to say and that is all what
the equation has said ok. Again any doubts? Yeah.
Yes. So, the so, in this particular case I have kept it same, but otherwise you could, you
could take a bigger pipe you could take a smaller pipe. So, that is why see we are talking
about for example, t * or length when you take out the solution in terms of an x *, it does
not matter what the actual dimension is.
216
Let us say if you are talking about a; you are talking about a small pipe and then you are
talking about a large pipe. Let us say in the small pipe the diameter was R 1, in the large
pipe let us say that sorry the radius is R 2. In the small thing we would have non-
dementialated the non-dimensional R * would be R divided by R 1, here R * would be R
divided by R 2. So, this R * would vary between 0 and 1, here the R * would be vary
between again 0 and 1 ok. So, in the scale of that 0 and 1 whatever velocity that you get
ok, will remain same; here you mean on the one on the left hand side and the one on the
right hand side.
So, let us say you are looking at a point R * is equal to 0.5, here also you are looking at a
point R * is equal to 0.5. Your equations would give you a u * here; here it will give you
a u *. This u * the u * on the left hand side and the u * on the right hand side is going to
be same. But, the absolute magnitudes will be different because in this case you would
have non-dimensionalize u * with some characteristic velocity which you would have
defined it as small u divided by some capital U 1.
In the right hand side you would have defined your u * as some velocity divided by sorry
you do not need that divided by u 2. That means, this u that you are going to get in the
bigger pipe and this u that you are going to get in the smaller pipe, the numbers would be
different ok, but this u * would be same in both the cases.
No oh sorry I so, I am talking about here as a Reynolds number being different and
therefore, Reynolds number is the parameter. If you want to get this Reynolds number in
problem I should be same as a Reynolds number in problem II, then only you get this
match. And, that is the beauty of talking about it, we do not care about what are the actual
dimensions or the fluids ok. In fact, this is a very important principle that is used when you
have to make you know models in a lab instead of testing out in a field, which we will
discuss next week and this is a principle that we would use ok.
So, what we have done so, far two cases one was the trajectory of the particle, second is
the governing equation and these two cases are when we knew the form of the equation
and therefore, we have gone ahead and done the non-dimensionalization. Now, in most of
the cases we would not know the equation because, if we knew the equation we could have
solved it in some way ok. So, we would not know the equation, we would probably know
what are the relevant variables and how would we simplify that situation that is what we
217
are going to look at ok. And, there are two methods for doing it, the first method is what
you call the Buckingham pi theorem. And then there is an Ipsen’s method which we will
see if we have time today.
So, the situation that we are going to consider is the situation that we actually *ted talking
about when we *ted non-dimensional analysis which is flow past a body ok. We have some
object you know of some characteristic size L and then there is a fluid that is going to go
past it ok. And, one of the questions that we post was what is the force that this object
would exert on the fluid because we know that this we are keeping the object stationary;
that means, the fluid is going to slow down.
So that means, the you are go the object is exerting some force on the fluid flow and you
want to know what is that force, we even hypothesized a form for that right. So, let us say
that if the fluid is coming with a velocity U and the fluid has a density ρ and viscosity mu.
218
Then we said the force will be a function of the you know the velocity with which the fluid
is coming, the size of the object, the density of the fluid, the viscosity of the fluid and so
on right.
𝐹 = 𝑓(𝑈, 𝐿, 𝜌, 𝜇)
And, this is the example that we chose where we said we have like we have force 1 2 3 4
5 different variables and if you want to do an experiment we will do huge number of
combinations. And therefore, then it becomes you know a large data analysis problem
rather than making sense out of it ok. So, we want we said it can be further simplified ok.
In this case now we do not know what is the force that is going to be exerted, but we can
still calculate some form of the force ok.
And, Buckingham theorem tells you how to do that, we will do the proof on Monday we
will just follow the procedure first and see what the procedure is and then we will prove
that what we have done is right. So, right so, the steps involved step I is list the variables
involved in the problem. Now, this is a very crucial step because you are *ting from
nowhere ok, you do not know what are the variables involved. But, you could do let us say
some preliminary experiments, vary something and see that things are different which will
give you some idea about what might be the variables involved ok.
219
And, when you *t working on a problem let us say you are trying to analyze a reactor you
will have some idea maybe the size is important, some of the fluid properties are important
the you know things like that. So, it is a its the selecting the variables it is actually a skill
and there is no real way of doing it. So, in this particular problem I said we know that you
know the fluid the speed with which the fluid is coming, it can be an important parameter.
The size of the object, it does look like you know it can be an important parameter. I did
not say for example, the what the thermal conductivity of the fluid is important because,
one would not expect that. I did not notice a the surface tension is important because, there
is no really surface involved ok.
So, it is some so, you should not select too many parameters because, then it is not going
to give you any simplification. The thing that you should really look for us what could be
the relevant parameters, you could keep adding it to this later and see whether you know
things are different or not. So, in this particular problem we have taken now five different
variables right; force, length, velocity, density and viscosity of the fluid. These are the five
things that we have and our objective is to know derive an equation which will be a non-
dimensional equation. So, first thing is done list the variables involved in the problem, step
II write down the dimensions of each variable.
So, we have a force, we have velocity, we have a length, we have a density, we have a
viscosity unit. So,
[𝐹] = 𝑀𝐿𝑇 −1
[𝑈] = 𝐿𝑇 −1
[𝐿] = 𝐿1
[𝜌] = 𝑀𝐿−3
[𝜇] = 𝑀𝐿−1 𝑇 −1
220
So, this is the next step. So, what we need to do now is we have got five variables, we have
got the dimensions of them. Now, we have to select a number from it which I am going to
call j and I will select j to be the number of dimensions involved. So, in the problem we
have length, mass and time three dimensions involved. So, I will select j to be 3 to *t with
ok, I will select j equal to 3 and then I will select j equal to 3 and; that means, I have to
select now three variables from five variables that I know ok.
So, the three variables that I will select, I could select any three variable. So, right now I
am going to select length, velocity and density as the three variables. Now, when I select
them what I need to ensure is that they do not form a non-dimensional number by
themselves. So, let us see if I select L U ρ so, L has the dimension of length, U has the
dimensions of L T inverse, and ρ has the dimensions of M L to the power minus 3. Can
you write ρ, L and U in any combination that you like so, that it can become non-
dimensional? No, right that is because one thing is that ρ has an M in it, there is there are
no other quantities which has got mass in it. So, whatever way you write that M is not
going to go away.
Similarly, U has a T in it so, that T is anyway not going to go away. So, you can ensure,
so now, you are assured that L U and ρ just cannot form a non-dimensional number ok.
So, these are then this is these are the variables which we call are the repeating variables,
which will be part of each non-dimensional group that you are going to find ok. You will
221
find actually some non-dimensional groups, each of them will have L U and ρ as part of
the non-dimensional group along with something else and that along with something else
is something we need to find.
For example: if you are not able to find you know see we have selected 3 and we are not
able to find a you know variables which are not forming non-dimensional group then we
should reduce it. We should reduce the value of j by 1 which basically says instead of
looking for three variables select two variables as the repeating variables and proceed. Is
that clear? Ok. So, we have got L U ρ yeah.
Now, it is easy step IV add the additional variables to form pi groups. So, we have L U
and ρ, the variables that we have not considered yet are force and viscosity; so, we can
form two pi groups one with F and one with mu. So, our first pi group which I will call pi
1 is going to be just F times L to the power a U to the power b ρ to the power c. So, L, U
and ρ are my repeating variables, I raise each of them with a certain power, I multiply it
with some force ok. And, that is what I am going to call my first non-dimensional group
to be.
𝜋1 = 𝐹𝐿𝑎1 𝑈 𝑏1 𝜌𝑐1
222
I do not know what a b c are, but it is likely that there might be an a b c which might exist
which can give you a non-dimensional group ok. Ask no ok, I will take a pause and I will
take questions ok. Let me repeat what we have done so, far maybe then that will help.
What is our problem right now? We have a object, we have a fluid that is flowing past it
and we want to know what is the force that this object is going to exert on the fluid ok.
You are clear that the fluid is going to slow down right because, you have kept an object
there so, the fluid is going to go slow down. Fluid was actually coming with a velocity U
and it is going to go down to some other velocity which we do not know, but our interest
right now is to find out what is this force ok.
Now, this is a very important problem as you will see many applications of this. So, that
is the that is the objective that we have. Now we, but we do not know anything; one
possibility was that we solve the governing equations. So, which were complicated
equations and you knew I mean you have actually seen that only in certain situations, you
can solve it ok. This particular problem is a very difficult problem ok, you do not know
exactly at the moment how to solve it. So, you are actually trying to do an alternate
approach, that alternate approach is what we are going to we are trying to learn right now
ok.
So, it involves certain number of steps and we are just going to follow those steps and see
what we are going to end up with ok. And, that result is what is one calls as Buckingham
pi theorem ok, basically it is a theorem that tells you that how many non-dimensional
variables you can identify and there exists a relation between those non-dimensional
variables. So, I am telling you that we will do the proof on Monday we will just follow so,
that we are familiar with the you know the idea of dimension. So, that you know the proof
becomes simpler. So, the problem now is to calculate the force we are yeah. Yeah.
It will depend on the shape of the body.
So, how would you include the shape? What is that?
Angle does not have dimension so, I do not care. What else would you include? Ok. So,
you are right so, for example, if it was a rectangle then we could have chosen two different
lengths right or if it was a * I could have chosen multiple lengths ok. So, right now I am
assuming that there is only one particular length that is relevant for my problem. If you
223
wanted to actually do a rectangle you could define an a, you could define a b and you can
say it is a function of a and b. So, if you like you can assume right now that it is a sphere
of diameter L.
How do you know b does not have any effect? See fluid is going to go like that right so,
the regions of b it is coming in contact a lot. There you are applying the no slip boundary
condition, the fluid is indeed going to go down right, whether it is a smaller one or whether
it is a longer one the fluid will slow down up to different extent.
Which surface area?
Of the object. So, you could so, that you are saying that a into b is your non-dimensional
parameter, sorry it is a dimensional parameter that you are worried about; I do not know
at the moment whether a into b is the right parameter or is it something else. So, I am going
to keep my a and b separate, I am not even caring about right now in the problem I am just
saying L is the only parameter that I will worry ok. So, I will continue. So, this is so, we
have written down a expression not really an expression just a functional form, where we
have said that F is dependent on the other four variables in the problem which we have
identified as velocity, length, density and viscosity ok.
So, we have got five variables and we know the dimensions of each of them. We have
identified the dimensions of each of them, that was step number II. Simply writing down
what is it in terms of mass, length and time that is a simple thing to do right. Now, I am
saying that really how to the next step really is to you know tell how many non-
dimensional groups are there ok, that is why we are coming up with this variable j ok.
So, let me state the no its ok. So, we are now out of these five we have to select so, it is
like selecting the scaling parameters, when we did the trajectory right we selected two
scaling parameters ok. So, here we are actually going to select three of them and these
three are L, U and ρ at the moment. We have taken L U ρ and I am saying that we can
form two groups which we will call pi groups containing L U and ρ pi 1 and pi 2.
224
So, L to the power a U to the power b ρ to the power c so, these are the two pi groups that
I have that I am writing down. In the first pi group I have select put all the three variables,
in the second pi group also I have put all the three variables ok; L U ρ L U and ρ, in the
first one I have added my force, in the second one I have added viscosity.
𝜋2 = 𝜇𝐿𝑎2 𝑈 𝑏2 𝜌𝑐2
So, now you can see that I have covered all the five variables I am not saying so, this we
let us call it a 1 b 1 c 1 and let us call this a 2 b 2 and c 2. I do not know whether what are
the values of a 1 b 1 and c 1 ok, but they have all of them have got some dimensions. So,
I may be able to find out some a 1 b 1 c 1 such that says that it forms a non-dimensional
number. So, let us try to find out a 1 b 1 c 1 ok. So, how would we do that? If we talk about
pi 1 that is F times L times U times ρ.
(𝑀𝐿𝑇 −2 )(𝐿)𝑎 (𝐿𝑇 −1 )𝑏 (𝑀𝐿−3 )𝑐 = 𝑀0 𝐿0 𝑇 0
𝑀 → 𝑐 = −1
𝑇 → 𝑏 = −2
𝐿 → 𝑎 = −2
𝐹
𝜋1 =
𝜌𝑈 2 𝐿2
225
226
Fluid Mechanics
Lecture - 25
Non-dimensional analysis - 3 Buckingham Pi Theorem
So, we were looking at Buckingham Pi Theorem in the last class. Let us just go through it
once again so, that you get it straight.
So, we really did not complete the theorem that day. So, let me just state the theorem. So,
we actually what we did is we actually try to do an example. So, what it says is that if a
physical process involves n variables which contain j dimensions, then relationship
containing k non dimensional variables can be written down and k will be greater than or
equal to n minus j ok.
So, suppose you have a physical process, which contains let us say n variables and it
involves j dimensions then you can find a k which is n minus j or greater than n minus j
and you will be able to find k non dimensional groups and you will be able to write down
a relation between them.
227
So, the problem that we had was some object with a characteristic dimension L and a fluid
approaching it with a velocity U and has parameters sorry the properties density rho and
viscosity mu and you were interested in calculating what is the force exerted by the object
on the fluid. And you said that is actually going to be a function of you know U, rho and
mu.
𝐹 = 𝑓(𝑈, 𝐿, 𝜌, 𝜇)
And we said we have identified now, 5 different variables and we will write down the
dimensions of each of them. So, So that means, we have identified n is equal to 5 j can be
3 because its mass length and time and therefore, we should find k which is equal to 5
minus 3 2 two non dimensional variables that we should be able to identify and then we
said that we have to select the repeating variables first.
[𝐹] = 𝑀𝐿𝑇 −1
[𝑈] = 𝐿𝑇 −1
[𝐿] = 𝐿1
[𝜌] = 𝑀𝐿−3
[𝜇] = 𝑀𝐿−1 𝑇 −1
228
So, j repeating variables which we chose as length U and rho and therefore, we wrote down
our first combination 𝜋1 And we said that has to be equal to a non dimensional number;
that means
(𝑀𝐿𝑇 −2 )(𝐿)𝑎 (𝐿𝑇 −1 )𝑏 (𝑀𝐿−3 )𝑐 = 𝑀0 𝐿0 𝑇 0
𝐹
𝜋1 =
𝜌𝑈 2 𝐿2
So, then we can talk about the next parameter.
229
So, the second non-dimensional number pi 2 can be generated with 𝜇;
𝜋2 = 𝜇𝐿𝑎 𝑈 𝑏 𝜌𝑐
[𝜋2 ] = [𝑀𝐿−1 𝑇 −1 ][𝐿]𝑎 [𝐿𝑇 −1 ]𝑏 [𝑀𝐿−3 ]𝑐 = 𝑀0 𝐿0 𝑇 0
Therefore,
1 + 𝑐 = 0 → 𝑐 = −1
−1 − 𝑏 = 0 → 𝑏 = −1
−1 + 𝑎 + 𝑏 − 3𝑐 = 0 → 𝑎 = −1
𝜇
𝜋2 = 𝜇𝐿−1 𝑈 −1 𝜌−1 =
𝜌𝑈𝐿
230
Buckingham pi theorem says that there exists a relationship between 𝜋1 𝑎𝑛𝑑 𝜋2
𝐹
𝜋1 =
𝜌𝑈 2 𝐿2
𝜇
𝜋2 =
𝜌𝑈𝐿
𝐹
= 𝑔(𝑅𝑒)
𝜌𝑈 2 𝐿2
So, this is one way of doing it there is another way of doing it which is relatively simpler
and it is known as Ipsen’s method.
231
So, in this case so, we will look at the same problem and try to derive the same relationship,
but in a slightly different procedure. We will write we want to get force as a function of
what all things L U rho mu that is our intention at the moment. The first thing that you do
is again write down the dimensions of each of them:
𝐹 = 𝑓(𝑈, 𝐿, 𝜌, 𝜇)
[𝐹] = 𝑀𝐿𝑇 −1
[𝑈] = 𝐿𝑇 −1
[𝐿] = 𝐿1
[𝜌] = 𝑀𝐿−3
[𝜇] = 𝑀𝐿−1 𝑇 −1
So, in this method what you have to do is you have to systematically eliminate dimensions
ok. So, what you should do Ipsen’s method says that systematically eliminate dimensions
that is all it says which means this eliminate we have got mass length and time let us say
we decide to eliminate mass first.
232
Eliminating Mass:
𝐹 𝜇
= 𝑓 (𝐿, 𝑈, 1, )
𝜌 𝜌
[𝐿4 𝑇 −2 ] = 𝑓([𝐿], [𝐿𝑇 −1 ], [1], [𝐿2 𝑇 −1 ])
So, now, this expression if you look at this expression has only length and time unit I can
choose to eliminate length or time ok. So, let us say we do time first.
233
Eliminating time:
𝐹 𝜇
2
= 𝑓 (𝐿, )
𝜌𝑈 𝜌𝑈
[𝐿2 ] = 𝑓([𝐿], [𝐿])
So, now the entire expression only has length in it, I can now eliminate L:
𝐹 𝜇
2 2
= 𝑓( )
𝜌𝑈 𝐿 𝜌𝑈𝐿
So, the good thing is that here you do not need to solve you know the algebraic equations
you do not need to really identify you know how many; you know you do not need to
figure out what are the repeating variables and so on it sort of comes out naturally yeah.
Huh. I did not understand can you explain again. Yes. No. So, j is 3. So, you will get two
non dimensional variables n minus j non dimensional variables.
Oh! then you will get three non dimensional variables since see this we have. So, for
example, here you have F by rho U square L square is only a function of another non
dimensional variable right. So, similarly if you have 3 of them you will find pi 1 is a
function of pi 2 comma pi 3 you do not know what f is, but this is the best you will be able
to write.
So, originally if your problem contains six non dimensional variables; let us say let us take
the same situation.
234
Let us say so, you only asked last time probably that if the shape was different. Let us say
that it has a length L and let us say a width w let us say the third dimension is also w, then
our F is going to be a function of L comma w comma rho mu U.
So, here n is going to be 6, but in this case again j is going to be 3 ok. So, you can expect
to find 3 non dimensional variables. Let us see you will find out them as pi 1, pi 2, pi 3;
then you can say that this is what the relationship going to be and so on. What would be
the extra non dimension parameter that will come in this case can anybody guess? It’ll be
just w by L everything will remaining same right you only have an extra w k. So, w by L
is the extra parameter that is going to come there which is basically the aspect ratio of the
body which is a non dimensional parameter. And that is so, simple in this case because we
have added just something which is a dimension of length and something physical ok. So,
that is fine.
So, we have now learned how again a given problem if you are familiar with the number
of variables or the number of variables that could be relevant, then you can write down a
simple relationship instead of investigating the dependence of each of the variables ok. So,
that so now, if you look at the flow what did we do? Initially we said if we knew the
relation how to non dimensionalize the variable. So, we did it using an algebraic relation,
we also did how to deal with the differential equations. Then now what we have done is
235
we did not know anything we only knew that these are possibly the variables that are
affecting and then we are we have written down a relation.
So, one of the crucial things that you have seen in this step is that you can write down the
dimensions of any physical variable as a power law expression.
So, I mentioned it a few classes before that we have written down the variable any variables
the dimensions as a power law expression ok. So, is this always going to be the case? Can
we always write the dimensions as a power law expression? And we are going to prove
that yes it is possible and therefore, all the things that we have said so far we will continue
to be hold for any problem that you pick up is the objective clear?
Consider a variable let us say X which I will write as x U when I have chosen unit system
U. So, what is that I have chosen? A unit system U in which x is some variable and this
small x is therefore, the magnitude of that quantity; x is some physical variable, but x some
let us like f is equal to 10 Newton force is equal to 10 Newton that is all what I have
written, but in a general form or I could write this as x prime U prime if I select a unit
system U prime or x double prime U double prime if I select a unit system U double prime.
So, what I am saying maybe I will take my SI system or CGS system or MKS system I
could always write it right. So, for example if its length I can write it as 1 meter 100
centimeter anything that I like. So, that is what I have written now and we know that all
236
this should be related right all these will be related by some factors. For example, x prime
by x will be some number right x prime is the magnitude in a unit system x double x is a
magnitude in some other unit system and they will be related by some number.
𝑋 = 𝑥𝑈
𝑋 = 𝑥′𝑈′
𝑋 = 𝑥 ′′ 𝑈 ′′
So, similarly x double prime by x for example. These are called reduction ratios. Reduction
ratios are nothing, but the numbers that you need to multiply when you are moving from
1 system of units to another system of units.
𝑥 ′ 𝑥 ′′
, → 𝑟𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑖𝑜𝑠
𝑥 𝑥
𝑟𝑙′ , 𝑟𝑚′ , 𝑟𝑡′ → 𝑟𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑟𝑎𝑡𝑖𝑜𝑠 𝑓𝑜𝑟 𝑙𝑒𝑛𝑔𝑡ℎ, 𝑚𝑎𝑠𝑠 𝑎𝑛𝑑 𝑡𝑖𝑚𝑒
So, when I define my x I did not say what it is its some variable some physical variable
this vector and you know r l prime is for example, the reduction ratio for the length k.
When I am changing from oh unit system U prime to unit system U what is the what is the
number that I should be multiplying with that is r l prime. If it is I am talking about length
or if it is mass, let it let us call it r m prime if it is time let us call it r t prime ok.
237
Similarly, so, within I know that I can write x prime by x is going to be sum function of r
l prime r m prime r t prime ok. So, the reduction ratio for length reduction ratio of mass
reduction ratio of time each of them if I combined in some form I should be able to get the
reduction ratio for any other quantity that I am dealing with remember I am only worried
about 3 dimensions now 3 particular mass length and time only. As you expand the
problem you could have other things coming into the picture.
Similarly I could write
𝑥′
= 𝑓(𝑟𝑙′ , 𝑟𝑚′ , 𝑟𝑡′ )
𝑥
𝑥 ′′
= 𝑓(𝑟𝑙′′ , 𝑟𝑚′′ , 𝑟𝑡′′ )
𝑥
𝑥 ′′ 𝑓(𝑟𝑙′′ , 𝑟𝑚′′ , 𝑟𝑡′′ )

=
𝑥′ 𝑓(𝑟𝑙′ , 𝑟𝑚′ , 𝑟𝑡′ )
Now,
𝑥 ′′ 𝑓(𝑟𝑙′′ , 𝑟𝑚′′ , 𝑟𝑡′′ ) 𝑟𝑙′′ 𝑟𝑚′′ 𝑟𝑡′′

= = 𝑓( ′ , ′ , ′)
𝑥′ 𝑓(𝑟𝑙′ , 𝑟𝑚′ , 𝑟𝑡′ ) 𝑟𝑙 𝑟𝑚 𝑟𝑡
238
So, what have I written really? See if you look at the right hand side I take the reduction
ratio for length mass and time separately and combine them or if I if you look at the
expression in the in the middle this part I calculate the quantity in 2 different units and
then take the ratio then I am going to get the same thing. For example, if I am talking about
force I calculate. So, this part let us say the double prime is I have calculated it in some
units let us say mass into acceleration I have done let us say I have done it as kilograms
times meter per second square or this quantity the one in the denominator I will do it in
CGS unit let us say I will do grams per grams in 2 centimeter per second square if I do the
force ratio or I do the mass ratio and acceleration ratio and multiply them the answers are
going to be same.
Essentially this part is just that fact that whether I do the forces and take the ratio or if I
take the each of the components mass, length and time separately and then take the ratio
and do combine appropriately I am going to get the same expression your same ratio same
number that is what we have any doubts? What we will do is we will fix U double prime
and change U prime arbitrarily. So, so far I talked about fixing an SI system and saying
that one of the system I am going to fix the other 1 I am going to consider to be a variable
so, that I can do differentiation and integrations and so, on.
So, I will say that in this quantity in this box expression, I am going to keep my double
primed variables as fixed quantities and my primed variables as really variables so, that I
can do differentiation. In fact, what I want to do is I want to take partial derivative of let
us call this expression 3 with respect to let us say r l prime ok. So, I am keeping my primed
variables as variable really. So, if I differentiate it. So, they will do the other way I will
keep the U prime I will keep U prime as the variable as the fixed quantity and U double
prime as arbitrary so, that I can differentiate the primed variable. So, you could do it either
way you will see that its symbols another.
𝜕𝑓 ′′ ′′ ′′ 𝜕𝑓 𝑟𝑙′′ 𝑟𝑚′′ 𝑟𝑡′′

(𝑟 , 𝑟 , 𝑟 ) ( , , )
𝜕𝑟𝑙′′ 𝑙 𝑚 𝑡 𝜕𝑟𝑙′′ 𝑟𝑙′ 𝑟𝑚′ 𝑟𝑡′
=
𝑓(𝑟𝑙′ , 𝑟𝑚′ , 𝑟𝑡′ ) 𝑟𝑙′
239
Now, having done that, let r l double prime approach r l prime ok. 𝑟𝑙′′ → 𝑟𝑙′
𝜕𝑓 ′ ′ ′ 𝜕𝑓
′ (𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) (1,1,1)
𝜕𝑟𝑙 𝜕𝑟𝑙′′
=
𝑓(𝑟𝑙′ , 𝑟𝑚′ , 𝑟𝑡′ ) 𝑟𝑙′
I will just call that alpha for the convenience also this is really a function of primed
variables I will just drop the primes.
240
𝜕𝑓
(𝑟 , 𝑟 , 𝑟 ) 𝛼
𝜕𝑟𝑙 𝑙 𝑚 𝑡
=
𝑓(𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) 𝑟𝑙
Now, I am going to integrate this quantity
𝑑𝑓(𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) 𝛼
∫ = ∫ 𝑑𝑟𝑙
𝑓(𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) 𝑟𝑙
ln 𝑓(𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) = 𝛼 ln 𝑟𝑙 + ln 𝑔(𝑟𝑚 , 𝑟𝑡 )
𝑓(𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) = 𝑟𝑙𝛼 𝑔(𝑟𝑚 , 𝑟𝑡 )
𝛽 𝛾
𝑓(𝑟𝑙 , 𝑟𝑚 , 𝑟𝑡 ) = 𝑟𝑙𝛼 𝑟𝑚 , 𝑟𝑡
You can do it and you will find that you are going there is another thing that is going to
change L was just an arbitrary choice what have you really done? We found that f of r l r
m r t some function of reduction ratios is just going to be a power law expression that is
what we found. Reduction ratios are the ones which you will use when you are changing
from one system of units to another system of units. We started by saying that we are going
to look at a quantity x which we will write it as either x U or x prime U prime or x double
prime U double prime and so, on and we said we are looking at quantities like x prime by
x x double prime by x and so, on ok.
241
Now, if you want your if the reduction ratios are going to be power law expressions, then
the only way x can be X will X has to be having a form L to the power of alpha M to the
power of beta T to the power of gamma if x has the form of L to the power of alpha M to
the power of beta T to the power of gamma then only I can have my reduction ratio having
you know my reduction the reduction form will basically have this power law expression
right if my x had any other form not just this power law form anything else then my
reduction ratio cannot be like this because the reduction ratio is by definition when I am
going from one system of units to another system of units. So, any X that I choose if the
dimensions do not have this form then I cannot have this form of reduction ratio.
So, now we have found that the only way you know in an arbitrary case, the reduction
ratios will always have a power law expression; that means, dimensions will always be
power law expressions for any physical variable that we choose and you have seen that
right any system any unit that you write will always be something to the power of
something its never in any other form that is what we have proven now that that has to be
the case always doubts yeah. Sure here.
Oh i. So, this is. So, this was my equation right dou f by dou r l prime. So, the only variable
that I have is r l prime. So, I just drop that r primes or you could just continue writing
everything and you will end up with an expression saying that f of r l prime r m prime r t
prime is r l prime to the power alpha. So, there is no physical this thing there yeah. Which
one. Ha r l prime should I go up no its general huh. Yes. Can you be bit loud?
No it cannot no because by definition it contains only length, it is like 100 it could be. So,
when you are going from cgs to si r l prime is just 100 like that it does not contain by
definition mass or time mass and time are fundamentally different dimensions. So,
therefore, r l prime cannot be a function of rm or rt; however, x prime could be I mean we
wrote it as x prime by x which may which for example, could be a reduction ratio for force
which means it will contain both length mass and time in it.
So, what we have done is either we write it as a combination or we write it separately that
is what we did and that is what exactly that box the expression was telling this red box the
expression that on the left hand on the right hand side I do it as a function of I will change
all the lengths I will change all the masses I will change all the times and appropriately
combine it or on the left hand side I am saying that I will take the force in 1 units I will
242
take the force and other units and I will take the ratio of it I will come I will get the same
number go down.
So,. So, far we. So, till we got this expression we kept these two things as two different
units r l prime and rl double prime or rather U prime and U double prime has two different
units now I am saying that one system is up and when we kept one of them fixed and other
changing. So, that we could do a differentiation and we are saying that there was the one
which we are changing is actually approaching the 1 which we have kept fixed ok. So,
therefore, each of these ratios r l double prime by rl prime is essentially just one number I
mean its basically same quantity because U double prime and U prime are the same system
of units now ok.
So, there is no reduction ratio really that is why I just wrote it as 1 1 or simply you say that
I am going to say my rl double prime is equal to rl prime and so, on for all primed quantities
and that is why its a constant and you can see that that constant is was comes out to be as
a factor which says that you know the quantity is that you know that comes as powers is
essentially the derivative you know with respect to that unit system that is what that power
means ok.
So, you go back you revise all these things the proof is not very important the proof is
there because I thought its a nice thing to do you can just learn it and more important is to
learn how to do the problems we will do one more proof and that is we are going to prove
the Buckingham theorem. So, we I just said that if you have n variables and j dimensions,
then we will have n minus j non dimensional parameters right. So, we will again do a
similar proof like that and then we will wind up this session this topic ok. I will see you
tomorrow.
243
Lecture – 26
Non-Dimensional Analysis - 4
Trinity Test
Let us start by saying that we have some physical process and we know some variables
involved in that physical process and we are trying to find out a relation between those
variables, without really solving anything about the system.
Let us say that a is a function of a 1, a 2 etcetera a k, a k plus 1 etcetera, a n where so a i

are all the independent variables in the system and a is the dependent variable. So, this is
analogous to if you think about the fluid flow pass the object that has been the problem
that we have considered.
𝑎 = 𝑓(𝑎1 , 𝑎2 , … , 𝑎𝑘 , 𝑎𝑘+1 , … , 𝑎𝑛 )
So for example, the small a is basically the force is a function of all other variables like
length, velocity, viscosity and so on. So, the everything that you see on the right hand side
are your independent variables and they decide some dependent variable in the problem
244
and let us say this function represents some physical process that is where we have started
with is that clear.
Now, one thing that we could check is there let us say that f actually has a multiplicative
form. Now once you do the proof you will find that this has nothing to do with this
assumption is not really important, this is just to determine some number what do you
mean by say that saying the let us say a be given by a 1 to the power of x 1, a 2 to the
power of x 2 and so on ak to the power of x k, a k plus 1 to the power of x k plus 1 etcetera
a n to the power of x n. So, each of them are raised to some powers and then multiplied
and let us say that if that is the functional form.
𝑥 𝑥 𝑥 𝑥
𝑘+1 𝑥
𝑎 = 𝑎1 1 . 𝑎2 2 … 𝑎𝑘 𝑘 𝑎𝑘+1 … 𝑎𝑛𝑛
This particular dependent variable has on the independent variable and if I write down the
dimensions again I will only worry about mass, length and time. So, let us say so the they
dimensions on the left hand side should match with the dimensions on the right hand side.
So, let us say the dimensions on the right hand side is L to the power of alpha, M to the
power of beta, T to the power of gamma is equal to a 1, let us say L to the power of alpha
1, M to the power of beta 1, T to the power of gamma 1, raise to the power x 1 times L to
the power alpha 2 M to the power beta 2, T to the power gamma 2 raise to x 2 and so on L
to the power of alpha k M to the power beta k T to the power gamma k raise to x k and so
245
on, L to the power of alpha n, M to the power of beta n, T to the power of gamma n raise
to x n.
𝑎𝑖 = 𝐿𝛼𝑖 𝑀𝛽𝑖 𝑇 𝛾𝑖 ; 𝑖 = 1,2, … , 𝑘, … , 𝑛
What have I done I have just written down the dimensions for each of this variables, of
course I do not know what those way dimensions are for a I said it basically is some length
to the power of something, mass to the power of something, time to the power of
something. For a 1 I said it is L to the power of some other x power other power L to the
power of alpha 1, M to the power of beta1, T to the power of gamma 1 and the whole thing
is raised by x 1 and so on for each of the independent variables that is ok. And the thing
that I am going to do is I am just going to equate the powers for each of the dimensions.
If I equate for alpha I will find sorry for length I will find that alpha is equal to alpha 1 x
1 plus alpha 2 x 2 and so on alpha k x k alpha n x n. If I equate the powers of mass I am
going to get beta equal to beta 1 x 1 plus beta 2 x 2 plus etcetera beta k x k x n gamma is
equal to gamma 1 x 1 plus gamma 2 x 2 plus etcetera right just 3 equations where we do
not know what are x 1 x 2 and so on.
𝛼 = 𝛼1 𝑥1 + 𝛼2 𝑥2 + ⋯ + 𝛼𝑘 𝑥𝑘 + ⋯ + 𝛼𝑛 𝑥𝑛
𝛽 = 𝛽1 𝑥1 + 𝛽2 𝑥2 + ⋯ + 𝛽𝑘 𝑥𝑘 + ⋯ + 𝛽𝑛 𝑥𝑛
246
𝛾 = 𝛾1 𝑥1 + 𝛾2 𝑥2 + ⋯ + 𝛾𝑘 𝑥𝑘 + ⋯ + 𝛾𝑛 𝑥𝑛
So, if you look at it what is this so we have n variables which are x 1 x 2 and so on and we
have got 3 equations right. So that means, I can select n minus 3 variables independently
and then that will fix the rest of the three variables right. I have got linear equations right
which has got n variables in it, but I have got only 3 equations so that means I can select
n minus 3 variable and the rest of them basically get fixed correct.
So that means, I would be able to select n minus 3 variables this is because I selected three
fundamental dimensions or if you generalize by saying that if you have k fundamental
dimensions then you can select n minus k variables which will fix let us say k other
variables, this is what you would have done if you were solving the simple system of linear
equations.
Let us x 1 x 2 x etcetera x k are the variables that are fixed. So, let me just try to tell again
what are we trying to do we have got a dependent variable a which is a function of lots of
independent variables, by writing down the dimensions I have found that if I cannot have
any arbitrary number of powers that I can assign to any of the variables. If I talk about
three fundamental dimensions then I am allowed to take only n minus 3 I can
independently choose n minus three variables and that means I will be fixing three of them
right three of them will be uniquely determined.
247
So, in general case if I have k dimensions I am going to be basically fixing k quantities as
fixed variables and n minus k will be independently chosen. So, in other words what I am
going to do is that I am going to say that the first k variables are the ones that are going to
be fixed and n minus k variables are the ones which can be independently chosen.
So and if you choose that if the first k variables x 1 x 2 etcetera up to x k are the powers
of the first three variables, remember the functional form is a equal to f of a 1 a 2 etcetera
up to a n. I am going to be fixing these and these from k plus 1 onwards I am going to be
independently choosing. So other variables that are fixed which belong to a 1 a 2 etcetera
up to a k. So, a 1 a 2 etcetera up to a k are the ones which I am going to fix and let x 1 x 2
x k be their powers which we have already taken and let a 1 a 2 etcetera up to a k are their
dimensions. But simply what I am saying is that a 1 is nothing but a 1 a k is same as a k.
So, for the first k variables let us say capital A 1 capital A 2 and so on are their dimensions
and A k to a n or a k plus 1 to an are the independent variables that I have and I am going
to now represent the dimensions of these independent variables in terms of dimensions of
the fixed variables. What am I trying to do I am expressing dimensions of independent
variables a k plus 1 etcetera up to a n in terms of fixed variables a 1 etcetera up to a k.
So, the first k th variables you have fixed the dimensions, so let us take the problem of
force is given by let us say mu rho L V. So, there are or I will write it in a different way L
V rho mu, what I am saying I will say L V and rho I am choose going to select as the
fundamental dimensions. So, that I can represent the dimensions of mu in terms of L V
and rho and I will be able to do that right.
[𝑎1 ] = 𝐴1 ; [𝑎𝑘 ] = 𝐴𝑘
𝐹 = 𝐹(𝐿, 𝑉, 𝜌, 𝜇)
So, similarly I am saying that if I have you know M variables I will skip the first k variables
I will select this for dimensions of the first k variable and then the n minus k th variables
can be represented in terms of first k variables. This is the way you can represent the
viscosity in terms in the terms in dimen in the in terms of dimensions of L V and rho.
248
So that means, if I have a k plus one dimensions of a k plus 1 can be written in terms of
dimensions of the first k variable let us say A 1 is the dimension of small A 1 which will
be raise to some arbitrary power let us say p 1 A 2 to the power p 2 and so on A k to the
power p k.
[𝑎𝑘+1 ] = 𝐴1𝑃1 𝐴𝑃22 … 𝐴𝑃𝑘 𝑘
So, I have A 1 this capital A 1 capital A 2 etcetera up to capital A k are the dimensions of
the first k variables and you are representing the k plus 1 th variable it is dimensions as a
power of other variable other k variables that you have already chosen and so on you could
do it. All the way up to let us say a n minus sorry a n which will be another some A 1 to
the power of r 1 A 2 to the power of r 2 so on A k to the power r k.
Similarly, you will also be able to write your dependent variable again, so let us say F you
want to write F in terms of L V and rho. So, small a which is the dimension of the
dependent variable again in terms of other dimensions let us say A 1 to the power n 1, A
2 to the power M 2 and so on A k to the power M k. I will take a pause and ask the questions
if you have.
[𝑎𝑛 ] = 𝐴1𝑟1 𝐴𝑟22 … 𝐴𝑟𝑘𝑘
𝑚𝑘
[𝑎] = 𝐴1𝑚1 𝐴𝑚2
2 … 𝐴𝑘
249
Should I go through the procedure once again or you are clear or you are lost either say
yes or no should I go through it again. So, we are trying to prove Buckingham pi theorem
and we want to do it in a general form, we do not want to take five eq five independent
variables three dependent variables and so on. So, I have taken a general function a with
that so I have taken a general function a which is a function of a 1 a 2 etcetera up to a n,
so that is a form that I have considered. Now the first thing that I want to do is to find out
how many dependent and how many independent variables that I can select.
To do that I am going to say then let us say that this function is actually of a multiplicative
form, that is I am saying that if small a the dependent variable is expressed as a power as
a multiples of powers of each of the variables. So, a 1 to the power x 1, a 2 to the power
of x 2 and so on and if I do that and if I write down the dimensions of all the variables let
us say the dimension of small a is in terms of L M and T raise to some arbitrary powers a
1 raise to the power.
So, again you know each of them in terms of L M and T raise to some arbitrary powers to
the power x 1 because, I have assumed that it is the power of it is the function itself is a 1
to the power of x 1. So, I have just written each of them as you know some L M T to the
power of something and if I equate the dimensions of each of them, then I basically end
up with three linear equations.
Of course we did not know what x 1 x 2 etcetera to x n where and that is our objective we
want to find out x 1 x 2 and so on and we are now we cannot do that because we basically
have three equations and we will have n unknowns and we are ending up with n unknowns
because we had n variables to start with right. So, we have n variables three equations, that
means that we have to select n minus 3 variables independently and that will fix three
variables as you make solutions right.
Because these are linear equations you will be able to solve them and you will be able to
find three of them. So that means, in this problem if I take this the if I take this as the form
I have to say that what is x k plus 1 x k plus 2 etcetera to x n, if I independently fix them
then I will be able to tell what is x 1 x 2 etcetera up to x k right.
If I fix x k pl if I choose x k plus 1 x k plus 2 x a etcetera to x n then I will be able to tell

what is x 1 x 2 etcetera up to x k right. If I have five variables and if I have three dimensions
250
two of them I will independently select which will fix the you know the powers of the
other three.
So, so that is what I am so we so the that means, we can select some of them independently
and some of them as a fixed value. Now the idea is that we are going to represent all the
fixed variables all the variables that we are going to select as the fixed ones, which means
this particular set in terms of the independent sorry the independent ones that we are going
to represent in terms of the fixed ones. So, the dimensions of k plus 1 we are going to write
in in dimensions of a 1 a 2 and so on. So, that is what I was telling you this example that
if you have force as a function of L, V, rho and mu. So, this is 1 this is 2 etcetera and let
us say this is the k plus 1 th variable the k plus 1 th variable I will represent in terms of
first k variables or to be precise the dimensions of k plus 1 th variable will be represented
in terms of k other dimensions of other k variables.
So, because I can represent the dimension of mu in terms of dimensions of L V and rho, I
can keep on doing that I can add any other variable to this system and I will be able to
represent the dimension of the variable that I am adding in terms of dimensions of the first
three variable I have chosen right. So, I have so in this system I have; I have taken a k
fundamental dimensions.
So, all the variables that I am going to have from k plus 1 onwards will be represented in
terms of dimensions of first k variables. So, this is the dimension of the k plus 1 th variable
I have represented it as dimensions of all the first k th variables and the dependent variable
a, I can again represent it as a in terms of dimensions of the first k th variables.
251
If that is a case then so what is our form right now a is f of a 1, a 2 etcetera up to a k, a k

plus 1 etcetera up to a n, this is the functional form that we have chosen and we have said
that we are going to represent the dimensions of all these the dimensions of these in terms
of first k variables that is all I have said so far. Now let us say we select a unit system as
or as a 1, a 2 etcetera up to a k. So, we might deal with the unit systems like SI you we
will deal with the unit systems like mk mks instead of that let us say we select a unit system
which is given by the first k variables itself. If I do that then that means each of the variable
that I have can be expressed in terms of a 1, a 2 and so on.
So, if I do this then I can write let us say a represented in terms of a 1 a 2 and so on is a 1
to the power m 1, a 2 to the power of m 2 etcetera up to a k to the power of m k. So, that
is a dependent variable is what have I done now a is written in terms of the k variables that
I have chosen. So, if you like to think about let us say; let us say you have a system in
which let us say you have you are dealing with acceleration.
Let us say acceleration is given as 980 meter per Second Square. So, if you were chosen
an SI system of unit and you say that your acceleration is 980 meter per Second Square.
But if you choose a unit system in which g is a fundamental quantity, then what would be
acceleration if g is the fundamental dimension that you have chosen then the acceleration
would be just 100 right because it is 100 g.
252
So, if you are choosing SI as the coordinate system then you would have said 980 meter
per Second Square is your acceleration. But if you select g you know as a fundamental
quantity in your unit system then you will say oh my acceleration is 100 that will tell you
what the acceleration is right. So, similarly if you select a 1, a 2 etcetera up to a k as the;
as the fundamental variables or fundamental quantities in your unit system, then a has to
be divided by each of those variables to get the number that you are looking for.
Just the way if a is equal to 980 meter per second square, if you select g as a fundamental
quantity then your a is nothing but 100 because 100 g is going to be your ah you know
your acceleration some number. Similarly if I have a any variable I can just divide it with
whatever independent variables that I have chosen you know raise to powers appropriately
and then I will get you I will. So, this is like this small a measured in a unit system which
is panned by a 1 a 2 etcetera up to a k.
I have selected a unit system in which my fundamental quantities are a 1, a 2 etcetera up

to a k. In this problem f which is a function of L V rho mu ok. I want to represent force in
terms of a length, velocity and density and then I would write F. If I want to measure F let
us say some 10 newton I want to measure 10 newton, but not as newton but in terms of
things that are contributed by length, velocity and density. Then I would have written it is
L to the power of alpha, V to the power of beta, rho to the power of gamma, of course L
beta sorry alpha, beta, gamma are to be chosen in such a way that it will basically will have
the dimensions of force, so that is all what I am doing here.
I am saying that I am selecting a 1 a 2 etcetera up to a k as my cor as my units. So, that I

can measure any variable in terms of other k variables that I have chosen, so that means
that is going to be equal to I will just write down the expression we will talk about it again.
So, a 1 measured in terms of a 1 will be just one a 2 measured in terms of a 2 will be just
1 and so on, a k plus 1 measured in terms of a 1 a 2 etcetera up to a k will be again some
a 1 to the power of p 1 which we do not know a 2 to the power of p 2 etcetera a k to the
power of p k and so on.
All the way up to variables a n which will be measured as a 1 to the power of m 1, a 2 the
power of m 2 some arbitrary powers, a k to the power of m k that is what we basically
would get. If we were you know taking up unit system span by the first k variables let us
253
stop there we will talk about it once again tomorrow. That is all actually Buckingham pi
theorem it already says that what is that we need to do.
𝑎 𝑎𝑘+1 𝑎𝑚
𝑚 𝑚 𝑚 = 𝑓 (1,1, … , 𝑃1 𝑃2 𝑃 ,…, 𝑚1 𝑚2 𝑚 )
𝑎1 1 𝑎2 2 … 𝑎𝑘 𝑘 𝑎1 𝑎2 … 𝑎22 𝑎1 𝑎2 … 𝑎2 2
254
Fluid Mechanics
Lecture - 27
Non-dimensional analysis - 5-Concept of similarity
So, to remind you what we were looking at the; so, we said we have some physical process
let us say which contains a dependent variable a which is a function of several independent
variables
𝑎 = 𝑓(𝑎1 , 𝑎2 , … , 𝑎𝑘 , 𝑎𝑘+1 , … , 𝑎𝑛 )
𝑎 𝑎𝑘 𝑎𝑛
𝑚 𝑚 𝑚 = 𝑓 (1,1, … 1, 𝑃 𝑃 𝑃 … 𝑟 𝑟 𝑟 )
𝑎1 1 𝑎2 2 … 𝑎𝑘 𝑘 𝑎11 𝑎22 … 𝑎𝑘 𝑘 𝑎11 𝑎22 … 𝑎𝑘𝑘
So, this is where we ended up yesterday. So, what did we have we have generally had a as
a function of several independent variable and we selected a system of units which is given
by a few of them. So, we had k fundamental units.
So, we have taken k variable and we expressed all other variables in terms of that k
variables and that is all the proof is because each of these quantities now you have ended
up are non dimensional numbers ok.
255
𝜋 = 𝑓(𝜋1 , 𝜋2 , … , 𝜋𝑛−𝑘 )
𝑓(𝑎1 , 𝑎2 , … , 𝑎3 ) = 0 → 𝑓(𝜋1 , 𝜋2 , … , 𝜋𝑛−𝑘 ) = 0
So, if you have n variables and k dimensions you will find out n minus k independent way
in the n minus k dimensionless groups. Then I want to tell you something else and that is
actually a story; it is a story that typically goes with Buckingham pi theorem actually
dimension analysis. The story that we may see its almost true I think over a time it has got
lots of spices into it, but still I think its overall a good story to see. So, let us see ok.
256
So, it is a here you go ok. So, trinity test actually it is a code name it was code name of the
first nuclear bomb that was tested. So, now, I mean you theory act and everything is very
common, but long time back that is 1945. When actually this test was held this was held
in right now, it is in Mexico in a place this is in fact, this all things are taken from Wikipedia
you can read more about there its says that it was in on 16th of July at 5.29 AM this
happened. Actually it was held or it was suppose to happen at 4 o clock because of rain
then you known it was moved to 5.29.
Now, all this details are precisely given and that actually shows how important this test
was. We will see why this test important was and right hand side that you are seeing is just
you know about that you will see if you go there at the moment it is a protected side.
257
Let us see video of what this test look like.
So, this is shot from a distance of almost 10 kilometre away from where that test was
actually happening; 10 kilometres is quite far you can imagine. Did you see that explosion?
You saw something right.
258
So, there I think its better now. So, imagine you are seeing it 10 kilometre away from
where you are sitting and is that huge the huge fire ball that rises. It is a near to the ground
here, you will see the top portion.
So, what is the big deal about the test? So, the point is that before this test, the world had
never seen something called a nuclear bomb ok. So, people had developed nuclear
reactions and so on some theory was available. Then theoretically, it was found that it is
possible to make a nuclear bomb ok, but how strong is the theory will it actually happen?.
259
So, this was the time of Second World War so, now why this story I told suddenly there is
a famous name associated with it and that is G I Taylor. We have already heard this name
right when we talked about Taylor Couette flow this guy was a fluid dynamicist he is
famous for several things that is in fluid mechanics. So, you heard about Taylor number
Taylor Couette flow there are lots of other things like Taylor’s instability, Taylors Vortices
Taylors scales and so, on ok. So, lot of things he has contributed.
260
So, what he said is this look what is bomb it is basically a release of energy. A lot of energy
that was concentrated at a particular place essentially gets expanded to a large amount
there is a wave of energy that is going to come spherically in all direction that is what for
him the bomb was. So, he said if that is the case let us write down what might be the
relevant variable involved and that is a first step of doing non dimensional analysis.
So, let us see what are the various that he considered.
He said energy because energy is very concentrated it is going to be released in a

spherically symmetric form. So, let us say E the total amount of energy involved is 1 of
the variables that we want to worry about. So, that is E then R the radius of the fireball
which is getting expanded and he definitely wanted to know how fast it is going to grow
and how much it is going to grow ok. So, R is a the radius of the fireball t is the time over
which this is happening.
Then he said there is a huge pressure that is going to be develop in the fireball right which
is what is going to expand the whole thing and its going to work against the external
pressure. So, he said let us consider the pressure inside the ball and the pressure outside
the ball as two relevant parameters and he also said we will also worry about the density
of air.
261
So, the density of the air is going to be a. So, it is will be a super heated air ok. So, you are
going to consider the density of that and you are going to consider the density of the outside
fluid against which it is expanding ok. So, this is because it is really that the high pressure
is working against the atmospheric pressure to expand it. So, that must be the mechanism
that is controlling how fast it is going to expand. So, let us say if these are the variables
involved then what could be the relation that exist between these variables.
Now, you know how to do that because you know Buckingham pi theorem or you can use
Ipsen’s method find out what is the non dimensional number what are the non dimensional
numbers that are going to come up.
So, that is the way Taylor chose to represent, but you might probably get some
combinations of this you can actually verify that some of. So, we as we said it can be
represented in many ways depending upon what it uses here repeating variables or what is
select as your fundamental dimensions, we will get different different combinations, but
they would represent the same thing. So, if you are getting let us say pi 1 and pi 2; if you
combine pi 1 and pi 2 in a some particular fashion, you are going to get a same non
dimensional number that he has written there.
So, he got these four of them pi 1 at some Et square by rho 0 to the power minus 1 by 5
then there was a pi 2 a pi 3 and a pi 4. And of course, as we have been seen there exist a
262
relation between these four and that is what he wrote and that is not surprising to anyone
more.
Now, he since he found that these four then he had to make some conclusions out of it. So,
he thought what could be the best thing that he can do and he said let us look at some of
those non dimensional numbers. There involved a non dimensional number that was ratio
of the densities ok.
So, if you basically say look at it is density of air outside air by density of that fireball that
is going to be extremely small because you have large amount of gas in a small amount of
space. So, is presser ok. So, 2 non dimensional groups are going to be extremely small and
he also said look at this pi 2 one of the non dimensional group that he go which actually
had this particular combination and he said he would look at things, when time is very
small; that means, at the very beginning of the explosion ok.
So, this is t to the power 6; that means, it is going to be extremely small quantity. So, in
other words what he said he said I will look at that situation in which pi 3 pi 4 and pi 2 are
very small ok. If you do that then what you so, originally you had pi 1 written as a function
of pi 2 pi 3 and pi 4 and now you are saying that you are going to make pi 2 pi 3 and pi 4
very small or in other words that would come out to say that pi 1 is some function which
is not dependent upon anything else. And therefore, pi 1 is a constant or some number
whatever it is.
263
So, pi 1 was this expression which contained radius E t square and so on. See he wrote it
in a nice form then that E is some constant times rho 0 times R to the power of 5 t to the
power of minus 2 alright ok. So, where K would be some constant which we do not know
what is going to come out or if you rewrite this equation as R as a function of t to which
of course, you can do you will see that R is a into t to the power of 2 by 5 ok. So, he said
that is how basically the radius of the fireball is going to grow it will grow with time as a
function of 2 by 5 at least at the very initial movement wherever he could neglect other
non-dimensional numbers.
𝑅 = 𝑎𝑡 2/5
264
Now, the test happened exactly witness the test also, but he did not get the details he had
no idea about the numbers. So, he could not actually verify any of the calculation. And
then, in 47 there was a magazine called this magazine life magazine which carry an article
about this trinity test or the pictures that may release were of this form ok. So, basically
you see the explosion and that was enough for Taylor because there was a length scale and
a time scale present in that photograph you see the ok.
265
In fact, the magazine actually contain a series of them in which its see at 0.10 milliseconds,
this was the how we look like and then it had information up to 1.93 milliseconds.
So, he got basically time and radius and that is all he needed he went back to his equation
which so, you had derived that R is a function of a sorry some constant which is basically
dependent upon other quantities t to the power of 2 by 5 where this constant a will now
contain what is the amount of energy that is involves right. So, he fitted r versus t calculated
a and calculated t the energy of the bomb which was actually classified information there
ok.
Nobody knew, but he calculated and then. So, the graph actually because of this simple
pattern during come well, but he also we can see the data points it came out to be a very
nice straight line falling on t to the power of 2 by 5 ok. And then he back calculated what
is the energy and this is the number that he came up with and in fact, it happened to be
very close to what was actually used ok. So, that is the strength of non dimensional analysis
without knowing anything you could still calculate some very relevant information.
266
Yeah. So, this is I thought I will show you some pictures of the classic problem that we
have been considering I have been telling that let us see we have a fluid you know that is
flowing pass some object and we have been thinking about you know how to calculate the
force on that object and so on these are from some experiments. So, these some
experimental pictures and the way people do it I think, I mentioned sometime you just put
some particles into the fluid and look at the motion of the particles that would tell you how
the fluid flow is going to look like here you have this circle is actually cylinder.
So, you have a cylindrical object fluid is flowing from one side and it is going to the other
side ok. In this picture it is not obvious, but here the fluid is coming from the left side it
basically you know deflects and when it goes back ok. So, these lines would be what lines
this would be? Yeah these lines are stream lines in this particular case this is a streamline
and this is this is a set a Reynolds number of 0.038 very small Reynolds numbers this is
how you would see flow past a cylinder.
267
This is even at a slightly higher Reynolds number and you would see something very
similar Reynolds number is actually 1.
This is Reynolds number is equal to 19 when Reynolds number increases what happens is
that the fluid is actually coming from the left side left side of your screen s and then it goes
and then this circulations will develop at the rear end or the back side ok. These are called
vortices and there is a lot of you know there is a strong fluid flow ok; that means, you
actually lose a lot of energy because of the formation of this let us see that later.
268
So, that is what would look like. It is another picture which is even at a higher Reynolds
number. So, it basically comes and then it goes and this it goes, then this vortices actually
become bigger this circulations.
And if you increase the Reynolds number further which is 55, this vortices will actually
dislodge from the cylinder and then it actually start going away ok. So, this is a some
picture of that then you can see that initially when flow was very neat there this nothing
that you were losing ok, but know you basically have a large region with lots of you know
269
vortices or recirculating region and you are actually going to lose a lot of energy because
of this particular process yeah.
So, that’s what flow pass an object would look like and we know that you know you can
actually calculate the force exerted by that object simply as a function of Reynolds number
and what you see is essentially that graph on the left hand side its call the drag coefficient
its the force then I should going to write yeah.
So, it is a drag force that he calculated over rho U square L square goes Reynolds number.
That how you take this plot and this is actually plot that you get when you whatever
experiential observation whatever relation you do, you can actually put it into the single
line ok. So, there are 2 thing that you see one is that of the sphere and one is that of a
cylinder; cylinder and sphere are 2 different objects. So, that curves would be different,
but you can see that it is giving you a nice plot.
We still do not know how to calculate this theoretically there is no way you can calculate
it except in some limits, but. So, this is the very important curve and you can see that as
you change the object you are going to get different different curves like this. So, this is a
way of representing.
So, here the flow is laminar, here the flow is laminar here also actually the flow will be
laminar it would just be that there will be lot of circulation there will be lot of vortices like
270
this, but that does not mean that the flow is turbulent ok. Here also I think the flow will
remain turbulent by the time this reaches there is chances that some turbulent you know
flows will develop in between the outside fluid will still continue to be laminar and
remember that 2,100 has nothing to do with this particular case ok.
Here you can see that the Reynolds numbers are much smaller oh no this is different
transition this is a transition called turbulent boundary layer which is not the Reynolds
number at which it happen.
So, what happens is which we will see this region this region which is very close to the
surface the we would expect the flow to be laminar right because it is very close to the
surface there will be lot of friction coming from the surface the velocity will be all, but
beyond some Reynolds numbers actually when that region will become turbulent. So, the
region which is very close to the surface will become turbulent and that happens at that
particular Reynolds number and this is just formation of process.
271
Fluid Mechanics
Lecture - 28
Characterization of particles 1
Fluid Mechanics aspects so, far right. So, Sumesh has been teaching about flow through
pipes you know flow between cylinders and things like that and he ended up with talking
about friction factor. So, he is going to come back in about 5 weeks and then he is going
to do a little more of a friction factor and few other things like boundary layer flows and
turbulence and things like that ok. I am going to you are going to see me for the next 5
weeks and we are going to talk a little bit about particles ok.
So, what happens if you have particles in the fluid that is the portion that I am going to
cover. So, for today’s lecture I just want to do a brief introduction that is the purpose of
today’s lecture.
So, the object is two fold I just want to introduce you to fluid particle systems that is
objective 1 for today’s class. The second thing is to I am going to talk a little bit about
fluid particle operations. So, at least the these operations I mean I think its also there in the
if you look at the course contents, if you have gone through I do not know how many have
you gone through the course content, but if you have gone through you will see that you
272
know I am going to discuss some of the terms that are there in the course content in terms
of you know what we are going to look at in terms of the fluid particle operations ok.
Just to begin with how do I mean particles are kind of everywhere right you look at any
chemical industry or any industry in general, you will always have particles right I mean
you have particles are kind of dealt with in different different forms as listed here you can
either work with particles in the dry form there is a lot of interest in looking at powders,
you know processing of powders how do you handle them you know aspects of that sort.
You also have cases where the particles are kind of processed in a you know wet form ok.
For example, paste you know things like that and more often people work with the particles
dispersed in a fluid that is the last two things you will come across something called as
slurries ok, which is basically dispersion of solid particles in a fluid only thing is when the
particle size is coarse larger size particles maybe tens of microns or several or millimeter
sized objects in a fluid that is when you kind of think about something called as a slurries.
You also have something called dispersions which typically refers to fine particles
dispersed in the fluid more of buzzwords if you want to call it as you can think about
something called as colloidal or nano particle dispersions that is ok.
273
So, what we are going to do in the rest of the you know 5 weeks is that, we are going to
look at you know how do people handle you know particles when you are going to put
them in a liquid that is going to be the goal of the next few lectures ok.
Any examples of fluid particle systems you know of? Some example that you I just want
to list different fluid particle system that you kind of know.
Some names, I am sure all of you know this let us write some examples right how about
paint right. Paint is typically you have some solid particles dispersed in a fluid of course,
there will be additives like pigments and stuff like that, but definitely there are some fillers
and these fillers are going to be in the time in the form of particles and therefore, paint is
an example of a fluid particle system any other examples? What is that?
Aerosol is a; aerosol is a kind of fluid particle system yes how about milk right milk
essentially has about you know a lot of water plus there are some case in my cells there
are some proteins and stuff like that is an example of a fluid particle system how about
blood right you have RBCs there is plasma right. So, I can really think of lot of examples
where particles coexist along with the fluid right what is common between paint, milk and
blood? What is that?
Student: Colloid; colloid.
274
I do not know if you can really call colloid as a blood as a colloid because typically typical
definition of colloids is that, you should have a particle size ranging from 1 nanometer to
about 1 micrometer. If the size of the particles are in that range that is when they are called
as a colloidal dispersions or nanoparticle dispersion, but if you look at blood the size of
the red blood cells is larger its about 8 microns in size. So, its really not a colloidal
dispersion, but still there is something that is common between 1 3 and 4 that is paint milk
and blood. What is that?
You know emulsions no I mean if you like look at blood its not an emulsion we will come
back to some of the classification little later on, but what is common between the three is
that the dispersing medium ok. They are all particles in a fluid right and in this case fluid
is a liquid if you look at all these things these are particulate dispersions in a liquid right
that is a common thing between 1 3 4 right.
Now, depending upon you know the fluid particle combination that you are going to work
with, there are several ways that people classify fluid particle systems that kind of
classification is given in this. What you are seeing you know the picture is a cartoon to say
that there is something called as a dispersed phase it could be your is your particulate
matter and there is something called as a continuous fluid and this continuous fluid if you
want to call it as you can call it as a carrier fluid or solvent you know people use different
275
different terms. So, there is a table here. So, if you look at the continuous phase right the
continuous phase can be gas and your continuous phase can be a liquid ok.
So, and your dispersed phase can again be liquid, solid or gas depending upon you know
what kind of system that you are working with you can have if you have a liquid droplets
dispersed in gas or if you have a solid particles dispersed in gas you have what is called as
a aerosols right. And again you know aerosol research is again a lot of importance now
because a lot of pollution you know stuff like that right. So, examples of aerosols would
be your fog, smog, smoke ok. So, all of these are examples of aerosols.
Now, when you have a liquid as a continuous phase and if your dispersed phase is gas you
have what is called as a foam right all of you are familiar with foam right I mean if you
take a bucket of water you know with some surfactant or some soap, you know only if you
open your tap water you know falls in and you know generation of bubbles right. So, that
is an example where you know you have gas or air bubbles in this case dispersed in a
continuous liquid medium that is an example of a foam. You have what are called as liquid
in liquid dispersions examples are I mean they are commonly called as emulsions as
somebody said milk is an example of an emulsion, there are lot of skin cream there are lot
of two in one shampoos these are all examples of you know liquid in liquid dispersions
also called emulsions. So, you also have cases where you have solid particles in a liquid
ok.
They are called typically they are called as a Sol and examples would be your ink, muddy
water you know paint and things like that ok. So, now that we know a little bit about you
know fluid particle systems and their you know different examples.
276
Let us think about I have a question to ask ok. Can you think of an example where the
formation of particle occurs during a processing operation or unit operations you know
whatever that you can think of any examples? I am I want an answer. So, far you know we
looked at you know milk and you know paint and stuff like that right.
So, when people of course, milk and blood is naturally occurring. So, there are particles
inherently in the fluid right and if you look at like say paint; paint is a formulation wherein
what you do is you there are particles and you put these particles into a fluid and you
basically make a formulation to get a paint of whatever desired quality right. Now these
are examples where the particles are already either there or they are kind of dispersed
externally right you take it from you know some source and put them into this fluid ok.
I want an example where you know do you think there are operations where the particles
are formed during an operation or the cases where the particles are formed during an
operation or it could be any chemical processes.
Student: Grinding.
Grinding ok. So, he is using a term called grinding; grinding is typically I hope all of you
know grinding right I mean it is basically what you do is you take a solid particle and you
basically reduce the size ok. There are different ways of reducing the size, but if you know
chop off particles and makes smaller stuff right that is what is called as a grinding ok. That
277
is an operation where you start with a larger particle and you break it down into smaller
sized particles right. I am I want an example for a case where I have a fluid and the particles
are formed during its processing.
Student: (Refer Time: 11:23) precipitation.
Precipitation right precipitation is one example. So, one would be ok. So, all of you have
heard this word called precipitation right. This typically happens you know when you have
you know if a fluid becomes you know in a loose term if the fluid becomes insoluble for
the particle right, if I have like say a solution for example, water plus some particles right
or say there you know I have some salt or something you know dissolved in a fluid right
and if I have a way of changing the solvent quality that you know that let me take an
example ok.
Precipitation actually is again one of the areas where there is a lot of research, I am going
to give an example where precipitation of precipitation is used as a method for making
particles ok. There is a class of technique called nano precipitation what is done is this
what you do is, you take a container your container has water now what you do is you take
another solution where you have some solvent and in the solvent, you dissolve a polymer
this solvent is a good solvent for the polymer this polymer you know completely dissolves
in the fluid and it gives you a clear homogeneous polymer solution.
Now, when I add this into water if the solvent and water they mix they like each other and
if the polymer does not like to be in contact with water as soon as I add the solution into
water what happens is you know this polymer kind of separates out and it separates out in
the form of a particle ok. This technique is what is called as a nano precipitation again one
of the widely used techniques for making drug particles and things like that ok. So, one of
the examples where the particles can be formed during your you know operations could
be nano precipitation. Any other example?
Chemical reactions as well right. So, you can have chemical reactions right you can have
a reactant a plus b giving rise to some product and the formation of the product could be
in terms of you know chemical reactions I know you doing the reactions any other thing?
Tell me at all of you know crystallization right crystallization again under the process
where you know whenever there is a you know homogeneous fluid you know if you create
278
some sort of super saturation right. So, there is going to be nucleation of particles and stuff
like that right.
Now, I am going give tell you an example where particle formation occurs during flowing
of a fluid ok. I am going to take a case of a transportation of crude oil, what can you say
from the images that you see? That is actually a cross section of a pipe. So, the in this pipe
there was a crude oil that was flowing and at some point you know they have cut the pipe
and they have taken a picture what can you say from this image? So, that is your pipe wall
alright this is your wall what is in the center any guess?
What is that?
279
Any guess? That is your wax deposition ok. So, all of you know that its crude oil it is a
complex mixture right it has you know hydrocarbons paraffins you know there is aromatic
stuff there is asphaltenes and things like that there are organic compounds of different
molecular weight present in the fluid and what happens when you transporting these fluids
is that, depending upon you know several conditions this crude oil could be subject to
different conditions you know there could be cases where the temperature during its
transport temperature around the pipe could reduce you know typically right it depends on
weather conditions right there are some like say cases where the weather conditions are
hot cold and stuff like that right.
Now, because you have a molecules of different molecular weight right long chain
molecules in the crude oil, what happens is that you know when you cool such fluid there
is going to be some sort of crystallization.
280
So, what you are actually looking at is images, which are obtained by taking an oil and in
this oil some wax is dissolved and then you heat this cold fluid into a high temperature
maybe like say 80 degree centigrade right. Now, if we start cooling it down at some point
you start seeing the formation of these needle like structures and these needle-like
structures are the wax crystals and that form whenever you take this fluid and then cool it
down to a temperature, something called as a wax appearance temperature.
So, typically if there is a way of transporting fluid this crude oil under conditions where
the temperature is always above the wax appearance temperature, it will continue to flow
like a nice fluid there is no issues. But the moment the temperatures surrounding this
pipeline goes down below this wax appearance temperature, you suddenly start seeing the
formation of these crystals and eventually these crystals what they do is they start
depositing on the walls inner wall of the pipes and they know this the skin of the wax that
is deposited it basically goes in time and ultimately it may even lead to a blockage ok.
This is an example where you start with you know pumping of fluid or transportation of
fluid and then your particles are formed during its transport because of the fact that you
know the temperature conditions are kind of you know altered during their flow.
281
Now, what do you think would be the implications of you know the formation of
deposition of particles in terms of fluid mechanics any thoughts? So, of course, there are
several implications right. Now we know that you know whenever you have the there is
generation of particles the viscosity of the fluid goes up right we kind of saw an example
in the previous in the in one of the lecture that I gave, wherein you have a base fluid you
saw adding particles we said that the viscosity was going up right.
Therefore, the generation of particles can lead to significant increase in the viscosity of the
fluid therefore, there is going to be associated transportation costs. So, one has to be you
know worrying about that. As it is evident from the figures the deposition of the particle
on the inner wall it reduces the cross section area right.
Effectively the cross section area that is available for the fluid flow is going to reduce
therefore, is also going to be there a going to be reduction in the flow rate as well ok. So,
it could have implications on the fluid processing that you are trying to do. Of course, the
too much of you know wax deposition can lead to blockage it can completely stop the fluid
from you know transportation right and so, we in the previous lecture we talked about the
skin friction and form friction right and we said that skin friction basically comes from the
roughness that is on the inner wall of the transport you know fluid you know the pipe
through which the fluid is being transported right.
282
Now, when you have; when you have you know wax deposited; obviously, is going to
affect the surface properties right and therefore, it is also going to affect the skin friction.
So, therefore, so, the formation of particle has is of relevance both in terms of trying to
understand when do the particles form plus as well as to now to look at you know fluid
mechanical aspects in terms of you know somebody has a question.
The first point first point whyone of the class that, I we I talked about viscosity
measurements ok. So, let me just draw something here. So, we had basically discussed
about a particular plot, you know where the viscosity was measured as a function of say
shear stress where in what we said that you know the viscosity would remain constant as
a function of shear stress when you have case where you only have a fluid ok.
Now I put in a small amount of particles what happens you know the viscosity goes up ok.
Now the increase in the viscosity purely comes from the because of the fact that you know
there is some dispersed species in the fluid ok. Therefore, the viscosity can increase they
of course, you know when you want to pump a fluid that has a higher viscosity of course,
you know you would have to worry about you know the pumping cost associated with the
fluid particle system ok.
So, in that context you know the formation of particles depending upon what is the
concentration of particles that are being generated, it can significantly increase the
viscosity and that could in turn have you know effect on you know the transportation in
terms of you know the pump requirements and things like that.
They there are two aspects in the example that I gave you it so, happened that you know
initial stages of wax crystal formation, the example that I gave initial you know when these
things start appearing initially they are nicely dispersed in the fluid ok. In that context I
have a fluid particle system right a homogeneous fluid particle system, what you seeing
here in terms of deposition happens at a little later stage first there is going to be some
formation of first layer of wax crystals ok. So, its not you know it just happens
instantaneously it happens during a period of time, but initially you could think about crude
oil at least when the temperature are around this wax appearance temperature is a
homogenous dispersion of you know such particles in a fluid ok.
283
So, now ok so, at least with this I just want to set a base for you know just to say something
about you know particles and that there is some you know particle fluid systems and they
can be some implications in terms of you know understanding their fluid mechanics ok.
Now I am going to kind of now I want to move on to talk about the different fluid particle
process that we are going to see in the course, I just want to start by showing you a video
let me just show that ok.
So, what you are seeing is a movie, we can do this in our lab if you want if you are
interested we can do that. So, what is done is they have a container and the container is
filled with water to begin with and all that is being done is take a one drop of milk and just
drop it in the fluid ok. Nothing else is happening, just one drop of milk into a container
and it is being kept below a microscope objective and you are watching it what do you
see? I should go back right you saw the title right what do you see? Brownian motion right
this happens when the particles are tiny when you have a very tiny particle if I do the same
thing like say let us say you know I have a container I put in millimeter sized objects ok.
Larger stuff I put in a liquid this is what happen right that is because the particles are really
large, then the reason why this motion happens because there are solvent molecules that
are hitting the particles right whatever thing that you are jiggling around that is because
there are solvent molecules which are continuously hitting the particle and continuous
bombardment of the particle and the solid molecule is what is leading to this motion ok.
284
And this motion happens when the particles are small enough and if I progressively
increase the size of the particle, you will see that you know very tiny stuff can move a lot
make it little bigger it moves a little less if I make it a little bit more bigger it will not move
ok. This is because if the energy with which the molecules are hitting the particles may
become much larger than you know much smaller than what is what it is required to move
the particle when the particle size becomes larger ok.
So, let us put it that way this point. So, we going to look at in the context of the course we
are going to look at some examples where we are going to have a fluid and you know we
are going to set the particle into motion. In this case motion happens purely because of the
bombardment of solvent molecules with the particle, but I can there are other ways of
setting a motion of the particle right can you think of other ways of I have a fluid, but that
contains particles I would like to have them move any examples? Any?
You can temperature increase will I do not know if that is the best way to answer this
because if you have a like say a tube for example and say that you know I have this tube
contains fluid and particle if I heat one end if I cool the other end then there is going to be
a temperature gradient right ok. Now because of this temperature gradient the things can
move that is what is called as a thermophoresis right.
What we saw just now is an example of any other examples of how to make them move?
285
What is that somebody says electrophoresis ok. As the name itself says in this case you
can set the particle into motion by applying an electric field right about something called
diffusiophoresis where in the basically the motion is set by creating a concentration
gradient right.
So, we are going to look at some aspects of this in the course. I am going to have one or
two classes on looking at particle characterization because before you start looking at a
fluid particle system, you should have an idea about you know what is a particle where do
they come from, you know what is the typical characterization that one has to do before
you work with fluid particle systems.
So, I am going to have some aspects of that in the course. We are going to talk a little bit
about something called a Stokes Einstein relationship which is which basically comes from
the Brownian motion aspects ok. We will talk a little bit more about it in the course some
aspects of electrophoresis just to give you an idea about motion of particles in a fluid that
is going to be the few aspects you know.
Now, some aspect that we are going to look at in the course is to look at if I have a fluid
particle system, how do we separate particles. So, any thoughts?
286
Like say imagine let us think about a case I have a container say that this container has say
water and take say some muddy water or a rain water if you want to call it there is lot of
particles in it right. So, you have a you can think about this as maybe a clay particles right
because when you get rain water lot of you know clay particles will be present in water.
So, if I want to like say separate these things what do you think with the best way to do
what I would like to do is to take this and then somehow ensure that you know everything
is at the bottom and there is a clear fluid.
So, how do you do that any example? Any thoughts? Right. Something called as
sedimentation all you have to do is you know take this container which has these you know
particles and you let it stand then what is going to happen is, your gravity is going to kick
in and the particles are going to slowly settle to the bottom. So, we are going to look at
some aspects we are going to we are going to look at sedimentation of single particles we
also going to look at sedimentation of collection of particles ok. So, in this example I
mentioned that they are going to settle down because of gravity right.
So, are there other ways of inducing sedimentation? Are there other ways by which I can
separate on the particles right in this case what you can do is you can use a? What is that?
You can do coagulation as well somebody says coagulation. So, what it basically mean
what he is trying to say is that, I have a system like this what I am going to do is whatever
particles I have, I am going to make them agglomerate or coagulate what happens when
287
they agglomerate or coagulate? The particles are going to become larger and when things
become larger they can settle right.
So, you can do that and the point that I wanted to say was that you know instead of gravity
I can use an external field right I can use what is called as a centrifugal forces or your
electric or magnetic forces right I can you know. So, there are ways by which I can either
exploit gravity for you know separation of particles or I can use some external fields. So,
we are going to look at some aspects of that in the course ok.
We are going to start with talking about settling of single particles there is something called
is a free settling that will happen when the particle concentration is lower or you know
when there is when you have maybe very few particles in the dispersion that is your free
settling and hindered settling when you have really large number of particles we are going
to look at some aspects of that.
Then we will move on to settling of slurries where you have a dispersion of particle, but
concentration is really large ok. And people do something called as a batch sedimentation
test these are kind of tests where you take dispersions in a column long column and you
basically watch them as the particle settle. In indeed you have an experiment in the next
semester, where you are going to do something called as a batch sedimentation test to look
at the sedimentation of behavior of slurries we are going to do some of that and we will
talk about some aspects of using external field ok.
So, now in processing you often come across cases, where I would like to have a fluid
particle contact ok. What I mean by that is say there you know I have a particle now I
would like to pass maybe a stream of say liquid or a gas over it there are certain devices
for doing this ok. Can you think of examples where I want something like this to happen?
I have let us think about like say solid catalysed reactions right.
So, where you have something like you know A plus B going to C if this reaction happens
on a catalyst surface therefore, you would like to have a catalyst I would like to hold it in
this place maybe by some means, but I would like to have the stream of A come in stream
of B come in and I would like to have the contact between the two occurring on the catalyst
surface and you know and that would lead to some reactions right.
288
So, there are a lot of processes in chemical engineering, where you have to ensure a proper
liquid solid contact or a liquid gas solid contact you could have a two phase system where
I am talking about liquid ok. I am talking about a liquid solid contact or a gas solid contact
or a liquid plus gas solid contact and there are devices for doing that and one of the very
commonly used device is something called as a packed bed its a fairly simple set up what
you have is basically a long column.
And in the column right the region that is marked you have you see these spherical things
these are particles these are kind of put in the column that is what is called as a packed bed
right its a solid particles you know packed into a bed and then in this particular example
what you have is basically there is a liquid that is coming out. So, basically there is a
scrubbing liquid and this scrubbing liquid is let me put this way.
At the bottom there is an entry for something called as a contaminated gas this gas contain
some liquid droplets ok. The objective is to have a clean gas go out at the top and whatever
gas that is entrained that is there in the whatever liquid that is entrained in the gas I want
to take it out ok. What I do is I introduce a liquid and what this liquid does when it is
coming down the column is that, it will take out all the liquid that is coming along with
the gas right and when it comes down it takes out all the liquid. So, that you know the
clean gas can go out. So, there are such devices of course, in this case it is used for some
kind of a mass transfer operation right.
289
So, you basically want to transfer the liquid that is there in the gas stream that is entering
I want to transfer that into a liquid stream that is basically coming down right. So, the now
if you want to do this without the packed bed is very hard right because you know of
course, the gas will go up and your liquid that is coming from the top would come down
right, but a proper contact between the liquid and the gas is not going to happen right.
Therefore, when it when these fluids flow through the packed bed this when I say packed
bed this is a bed of particles of course, there is going to be gap between the particles right
and when the when things flow between these gaps right. So, there is going to be a proper
liquid gas contact in this case right.
So, therefore, there are cases where you want to ensure either a liquid contact liquid gas
contact you know or a liquid particle or gas particle all kinds of you know fluid particles
operations where you want to ensure a proper contact between different you know species
people use this concept called packed beds we are going to look at a little bit more of that
in the lecture.
Now the other examples where you know one could look at ensuring a good contact
between the particle and the fluids is by something called as a fluidized bed ok. So, what
you are looking at is basically a case where on the left again you have a column and you
know all these particles are kind of packed into a bed now what you do is if you increase
the velocity ok.
290
So, I just wanted to say that in the packed bed what happens you know what keeps the
packed bed in the packed bed configuration is that there are basically support plates ok.
There are two plates that are basically holding these particles and you know no matter what
kind of fluid velocities you use, the particles are going to be in a fixed state ok. Now if you
have a fluidized bed if you remove the top plate the top plate is not there anymore if I
increase go on increasing the velocity of the fluid that I am sending from the bottom at
some point they are going to be in a nice dispersed state right that is what is called as a
fluidized bed. So, we are going to look at some aspects of that.
So we are going to look at flow through packed beds and fluidized beds, we are going to
analyze flow through packed bed in terms of you know we have learned flow through pipes
we going to use that concept to look at flow through packed beds and then look at some
pressure drop and things like that.
291
So, I have again a nice video if you in pharmaceutical industry there is a need for looking
at coating drug particles ok.
And coating of drug particles can be done for several reasons one of the reasons could be
that you know I would like to control the release right.
292
So, if I have people talk about controlled release right, one of the way of controlling the
release is that you know I have a particle and I do a coating on the surface ok. Or this is a
case where there is some kind of agglomeration that people somebody was talking about
a process called granulation ok; that means, you start with smaller particles and then you
kind of induce some sort of aggregation between the particles ok.
Fluidized beds are used for granulation purposes what you do is, you have a packed bed
to start with and then you introduce some gas and then introduce some way if you know
293
joining the particles. There is another example where fluidized bed using being used for
coating applications. So, you fluidize a bed to begin with and then whatever coating
material that you want; that you want to put on the surface, you again put it as a another
fluid and then you know you can ensure a nice homogeneous uniform coating of you know
the stuff on the particle surface.
So, the idea would be to make this core shell kind of particle where the core is your maybe
a drug molecule for example, drug particle and then you know your and the coating could
be something that you are trying to use for controlling the release ok. I just want to end by
the last one piece. So, we talked about separation by sedimentation and you know and
using external forces right any other ways of separation you can think of? Have you heard
of an operation called filtration? Right ok.
294
If you have you know I am sure all of you may have seen a water filter right if you look at
water purifier; there is a pre filter that takes water from you know tap and you know it
removes some particles to begin with it goes into a purifier there is going to be some carbon
filters and things like that right. So, we are also going to look at some aspects of filtration
during the course look at types of filtration, there are some special specific types of
filtration something called as a constant rate filtration and constant pressure filtration
devices, we are going to look at analyzing filtration in the context of flow through packed
beds ok. So, these are the some of the things that we are going to do in the class.
295
Fluid Mechanics
Lecture - 29
Characterization of particles 2
So, in the previous class we kind of had a introduction to what we are going to say in the
rest of the course right in terms of fluid particle mechanics. So, one of the thing that came
out of the discussions last week right, was that if you consider a particulate matter we said
you know a particulate matter could be either in the form of solids or solid particles or
liquid drops or gas bubbles right that is what we said right depending upon. So, you have
we said that you know there is a dispersed phase and a continuous phase and you know
your continuous phase could be a gas or a liquid ok
And your dispersed phase could be either particles liquids or gas right in that sense your
particulate matter when we are going to discuss in the course, it could be solid particles
liquid drops and gas bubbles right. Now, what do you think is that if you want to
distinguish the three can you think of one parameter which is kind of differentiates solid
particles from the other two.
So, when we say that you know we are dealing with fluid particle systems or particle filled
systems and I am working with solids you know filled in a fluid or liquid or gas bubbles
296
in a fluid. What do you think would be one important parameter that you would have to
you know worry about, when you are dealing with liquids and gases compared to solid
particles, deformability of the particles right. Because when you are working with you
know solid particles I mean if you work you know under certain conditions you can assume
that you know the solid particles will continue to remain as you know rigid, hard you know
they retain their size, shape you know during the processing operation, but whereas, if you
consider liquid droplets or gas bubbles it could so happen that you know they could deform
right.
What I mean by that? You know you could be working with like say a spherical liquid
droplets to begin with you know, but; however, during processing operation it can become
elongated right. So, such deformability you know of the particulate matter you would have
to worry when you are dealing with you know liquid droplets or gas bubbles.
Now, for any of these particulate matter what is the constituent of each of these particulate
matter right, everything is basically made up of atoms or molecules right. So, when we
have atoms and molecules that constitute this particulate matter you should know a little
bit about something called as inter atomic sorry atomic or inter molecular forces right, all
of you know about this.
So, you know a little bit about inter molecular forces inter atomic forces, I guess in the
previous semester in thermo I think there was a section on inter molecular forces right all
297
of you know what it is ok. So, the only thing that is important is that you know. So,
whatever the atoms and the molecules that make up these particles we should know that
you know there is something called as a inter molecular or inter particle forces between
them. And in general, your inter molecular force you know is typically a summation of
some attractive forces plus some repulsive forces right ok. And there could be different
contributions to the total you know interaction potential between the atoms and molecules
ok.
You would have heard of what are the things that contribute to this intermolecular forces
that you heard of atomic bonding there is something called as ionic bonding, there is
something called as a covalent bonds and metallic bonds Van-der-Waals forces right. So,
all of these things contribute to the inter atomic or inter molecular forces.
And one of the very famous examples of inter molecular forces is something called as a L
J potential, I do not know how many of you have heard of L J have you heard of this
Lennard-Jones potential right. So, typically if you were to have you know total potential
which is powered as a function of distance r it turns out that there is going to be a repulsive
force and this repulsive force goes as 1 over r power 12 and there is an attractive force that
goes as 1 over r power 6 ok.
So, in this case your total potential U is 1 over you know r power 12 minus 1 over r power
6 and negative because it is a attractive force ok; negative is attractive force right. So,
298
maybe just let me just do that right here. So, that is your, you know attractive force and 1
over 12 it is a repulsive force right.
𝑎 𝑏
𝑈𝑡𝑜𝑡𝑎𝑙 = −
𝑟 12 𝑟 6
And if I were to sum of these two typically your plot would look like this something like
this it will be something like that. And that is a total potential that is U in this case you if
you want to write as U total or just U and what can you say from this plot I can say
something about what is the depth of the minima of the energy minima.
The depth of the energy minima will tell you something about how strongly are the atoms
or the molecules are bound together in a given material right that is what the depth of the
minima represents, and that minima position this is what is called as it tells us something
about the equilibrium position right. On an average what is the typical distance between
the atoms of the molecules that constitute a given particle that is your equilibrium
separation.
Only thing is that the functional form of this interaction potential would definitely depend
on the type of particles that you are dealing with right. Typically people think about
Lennard Jones potential when you are thinking about like say molecules like methane you
know argon and things like that, but if you move to solid material then you would have to
worry about you know other kinds of functional form of the interaction potential.
So, in the case of the Lennard Jones is 1 over you know r power 12 there is typically a
constant a and constant b, and that is constant a and b depends on the type of material that
you are dealing with. And so the attractive contribution to Lennard Jones potential comes
from the Van-der-Waals forces right and the repulsive contribution basically comes from
the fact that you know every atom or a molecule has a electron cloud right. And you know
when two particles come close enough the repulsion between the electron cloud is what
gives rise to the repulsive forces yeah, any questions?
Now, if I look at this plot, can I say something about the range of interactions, can I say
something about long range interactions and short range interactions, which one of them
is long range and which is short range in this case, what I mean by that is, what I mean by
long range and short range is say there now I have two atoms or molecules you know I
299
have held them together say they are you know I start bringing them close enough. Now,
if the distance between the atoms or molecules is very large in this case if I look at this
particular plot I can clearly say that you know any distance above you know what I have
marked there; there is no interaction right, because the interaction is essentially 0 right.
The moment I start bringing them together, you can see that the particles start feeling the
attraction first right you know this the curve that is below this horizontal it starts becoming
nonzero of first right ok. Therefore, I can say that in this case attractive forces are long
range compared to the repulsive forces right. So, whenever people talk about you know
long range or short range forces everything depends on what kind of interaction forces you
are working with ok.
In this case Van-der-Waals are long range compared to you know the electrostatic
interactions, but it would depend on you know the kind of particle system that you want
to work with.
Now, specifically if you come to solid particles right; solid particles people may be
working with particulate solids which be crystalline or amorphous right ok. And again
crystalline you know all of you have done I am sure some basic material science where
you will know that you know crystalline material can exist in something called as Bravais
lattice right have you heard of this ok; Bravais lattice ok.
300
So, there are 14 Bravais lattice right you know cubic, hexagonal, trigonal, monoclinic and
things like that. And of course, in the crystalline material there is a ordered arrangement
of you know atoms of the molecules that make up the solid, but of course, in the case of
amorphous materials you know there is no long range order right. So, yeah; so and of
course, you know you can think about when you think of a solid particles your solid
particles could be made up of a metal or a polymer or a ceramic particle or it could be a
some kind of a composite as well right.
Now, so this is a little bit of basics about you know what the particles are made up of right.
Now, is this really important when you are trying to think about fluid particle mechanics?
It may or may not, it depends on you know what kind of fluid particle operations you are
trying to do.
For example, if I am like say in powder technology, you people worry about somebody
mentioned about making particles by reducing the size right. If I have a large particle and
if I am reducing its the size of the particle you know by some means maybe I am
hammering it or you know some means then I would have to worry about you know what
would be the modulus of the material you know whether because you know a certain type
of materials are much easier to break compared to the other.
So, you know then you know all these, you know some information about the
intermolecular forces inter you know inter atomic forces you know what is the toughness
of the particle, you know all these information’s may become important. But; however,
when I am worrying about you know like say there is a solid spherical particle of radius r
you know and there is a fluid going around it really does not matter whether the particular
is you know crystalline are you know on amorphous right.
So, therefore, having a basic idea about the solid particles what they are made up of
whether the bonding between the atoms that make up the particle is it ionic you know
crystalline, is it ionic covalent or metallic. So, all these information could be useful to
some cases and again having an idea about the Bravais lattice could be useful in some
applications. Just to give you an example people make up particles by a particular say in
reaction right.
So, in reactions when you want to make particle, what happen initially you have a nuclei
that is formed right. Now, the nuclei grow and then it becomes a bigger particle in the end
301
right. Now, the way they the next atoms comes and attached to this nuclei, it would again
depend on the specific sort of you know the you know the orientation of different
crystalline molecules that are in the particle and stuff like that.
And you know there are a lot of people who talk about adsorption of some things onto a
particle surface and there are cases where people have kind of identified that you know
some molecules attached to one one one plane of the particle you know compared to the
other planes and things like that. So, therefore, the having details you know having a basic
idea about you know the material science of particles becomes important in some field ok,
but in some other field you know you may be happy with thinking about the
characterization of materials in terms of knowing their size, density, shape and things like
that ok.
So, let me just wind up at this point and then we will just go to you know some basics
about particle characterization ok. So, if I have like say a particle, now I want to
characterize these particles. So, what kind of properties do you think you know I would
like to characterize these particles for and I have just drawn you know some object, but let
it need not be this right it could be any you know any particle right I mean I have something
like that. So, I would like to characterize these particles. What do you think are the
properties that I should worry about, yeah what is that?
302
Yeah of course, I mean of course, density is one you can think of course, size of the particle
itself right ok. So, basic calculation that I can think about you know I should know what
is a, the size of the particle, I should know the density. You can also think about something
called as a specific surface area and of course, you can think about you know
characterizing the shape as well right, because you know you may be working with
particles are different yeah.
So, when I say size we will come back to that ok. So, let us say that I have at the particles
in general right you take any particle in general, it can be a regular particle or a irregular
particle right. What do I mean by regular particle you could have like say a sphere right or
you can have like say a cube like particle or like say a cylinder or like say a disc like
particle right you know. So, maybe yeah that is the side view the top view is going to be
something like this right that is your disc like particles right you.
So, when people work with particles, the particles can be something called as a regular
particles, I mean when I mean by regular particles you know we know something about
the shape of the particles right. So, again in the regular particles I can say that spherical
particle is an example of something called as a isotropic particle ok. All the other things
that I have done here the example of what is called as a anisotropic particles right because
if you take a sphere right no matter you know from which side I look it will look exactly
similar right.
Whereas if you look at other shapes like you know cylinders, discs like particles, say
ellipsoids and stuff like that you know they do not look the same in from different
directions. So, you can think about regular particles being a spherical and non spherical.
303
And again depending upon the kind of particle that you are going to deal with you may
need either one dimension or two dimension or three dimension for a characterizing the
particles right. If I take a like see a spherical particle all I need to know is what is the
diameter of the particle DP as long as I mentioned one dimension the size of the particle
is no right.
And the same is also true for a cube right if you take a perfect cube all the size is like say
if the side length is known all I need is one dimension for characterizing in the particle.
But if I can you think of examples where I need two dimensions to characterize the
particles.
Example would be a cylinder right. So, I should know what is the length of the particle
and I should know also what is the diameter of the particle right. So, similar could be even
if you take a disc right; if you take a disc again you should know what is this height and
what is the diameter or the radius right any examples where you should specify all three
dimensions for characterizing the particle.
Student: Cuboid.
Cuboid right, if we example could be cuboid wherein you know you have to have all the
three dimensions specified. Again a general class of ellipsoid also need a three dimensions
right, you could have ellipse which can have three different dimensions if you take
304
something that looks like that. So, that is your dimension a right, if this is b, there is also
another dimension right the c, when a is; if this is a and there is another dimension c here
right.
If you have a case where a and b is equal to c there is an example of something called as a
prolate ellipsoid right ok. If it is a prolate ellipsoid then of course, you need a only two
dimensions because the two dimensions are the same and the length is different that is an
example of a prolate ellipsoid.
But; however, if you take a general case of ellipsoid you would have to identify three
dimensions you have to specify it for it to be completely characterized ok. When people
talk about a particles that are not spheres that is non spherical particle one of the parameter
that people always talk about if you go up to literature something as an aspect ratio ok. So,
in the case of like say a cylinder here your aspect ratio is your L by D right, that is the
length of the rod divided by the diameter right.
And of course, if you take up a prolate ellipsoid again its going to be either a by b or you
know a by c right. So, people think about you know a non dimensional length non
dimensional way of expressing dimensions and in the case of elongated particles aspect
ratio is something that people use ok.
𝐿 𝑎
𝐴𝑠𝑝𝑒𝑐𝑡 𝑟𝑎𝑡𝑖𝑜: ,
𝐷 𝑏
So, now, when you work with regular particles; I mean, like say if somebody comes and
tells you that hey you know tell me what is the dimension of this particle right. I mean, for
me I can quickly take up a scale and I can measure it you know I can specifically say, what
is the size of the particle right.
305
Now; however, if you are working with like say some arbitrary shape particle say that you
know you have you know some shape like that, and if I want to calculate what is the
dimension of the particle. What do you think one could do? I have an arbitrary shape
particle, I would like to characterize this particle in terms of the dimension ok, what do
you think one could do, any thoughts?
So, let us say that you know this is an image of a particle that I obtained by using like say
a microscope all of you know what is a microscope right. So, what is typically done is you
take some say solid base, you spread particles in the surface I take this and I put on around
a microscope right. There is a IPs right there is an IP, so you basically go and focus on the
particle surface.
Whenever, I capture an image what I actually get is something called as a projected area
right. So, what the microscope does is, it measures the projected area of the particle right
of course, you know microscope is a, you know you can only do a 2D imaging right. If I
have a spherical particle of course, if I look it from the top it will appear, so a circle right.
Now, whenever you have a image that I acquired using a microscope and if I have some
way of doing something called as a image analysis ok; that means, I have some way of
you know characterizing you know my the particle in the image right. So, what people do
is when you have a irregular object like this, one of the dimension that people talk about
is what is called as a equivalent sorry equivalent circle diameter. What you do is I have an
306
image of an object I have some way of measuring the area and what I do is I find out what
is the area of the circle that has exactly the same area of the projected particle you know
area that I have taken ok.
If I have a way of doing that from that I can basically back calculate what is the area of the
circle whose area is equal to the area of the particle under consideration ok, that will give
me what is called as a equivalent circle diameter. This is one way of you know
characterizing the size of an irregular you know object, there are other methods. In one of
the method what is basically done is if this is your area what you do is you basically draw
a bisector ok. A draw bisector such that the area on one side of the bisector and the area
on the other side are the same. And I say that the length of this bisector is the dimension
of the object, this is a in literature is called as a Martin's diameter ok.
So, what you do is you take an object you start; you basically find out a bisector and as
long as the area on the two sides of the bisector of the same. You find out what is the length
of that bisector which basically divides the particle into two areas and the dimension
corresponding to that is the diameter of the particle.
There is also something called as a Feret’s diameter. In this what is basically done is what
you do is you basically construct two tangents ok, what you do is you construct two
tangents ok. Now, when you construct two parallel tangents if these two parallel tangents
exactly and come pass the particle that you have the distance between these two tangents
is what is called as a Feret’s diameter.
Again, these are the guidelines in the literature for thinking about you know how does one
go about finding the dimensions, but what is the dimension that I have to use for a
particular application, it depends on several factors. It would depend on you know what
kind of access do I have for a particular technique. For example, I am able to discuss about
the equivalent circle diameter, Martin’s diameter, Feret diameter because there is a
microscope. I am able to capture the image, this is a projected area projected image of the
particle and there is a access to an image analysis and I can easily calculate the area of the
you know the projected area and I can equate that to pi r square right and I have a way of
back calculating what is the radius right.
So, because I had access to microscopy and image analysis I could think about equivalent
diameter, Martin diameter and Feret’s diameter ok, if you want to do it you know with a
307
scale and a pencil is very hard right you cannot do it ok. So, therefore, the kind of
dimension that I am going to associated with a non spherical or a sorry irregular particle
depends on what is the access that I have for a particular technique for measuring the
dimension plus it will also depend on what am I going to do with this dimension ok, what
is it; what is the use of this and what to what use I am going to put it depends on that.
But these are some guidelines that people use in the literature and you should be you know
aware of this. Now, there is yeah somebody has a question. Ok. Yeah. Yeah.
Yeah, there will be a different line yes I completely agree with you. So, what are you
saying is. So, in this case I have drawn a horizontal line right, in this case there is a
horizontal line that basically bisects right. I could have had a vertical line, I could have it
in any angle as I said it depends on you know this is one. So, if I were to do it what I would
do it that maybe the best description would be you draw tangents at different orientations,
you get a dimension for every orientation, you get that you know that diameter that bisects,
the area into two equal you know area halves and take an average of that that I think that
would be the best representation of a Martin diameter ok. You understand what I am trying
to say.
So, instead of drawing one line you could think about doing multiple lines right as long as
you know the area on both the sides are equal I am going to get a dimension. So, you could
think about it enough the average martin diameter could be more in diameter 1 plus 2 plus
308
3 divided by 4. So, you have to figure out you know what is the best dimension that you
would like to extract out of such analysis and then you know and it as I said, it depends on
you know what application you are going to put this dimension into in your calculations
yeah ok.
Now, so the other way of kind of working around you know irregular shaped particles are
part of they are not spherical is by exploiting a quantity called sphericity. This is typically
defined by
𝑆
(𝑉𝑃 )
𝑃 𝑠𝑝ℎ𝑒𝑟𝑖𝑐𝑎𝑙 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒,𝐷𝑃
𝜙𝑠 =
𝑆
(𝑉𝑃 )
𝑃 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑢𝑛𝑑𝑒𝑟 𝑐𝑜𝑛𝑠𝑖𝑑𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑤ℎ𝑜𝑠𝑒 𝑛𝑜𝑚𝑖𝑎𝑙 𝑠𝑖𝑧𝑒 𝑖𝑠 𝐷𝑃
That is your I will explain a little bit I see some strange faces, but you know yeah I am
going explain this a little bit ok. So, now, all you have to do is say that I have an arbitrary
object like this ok, what you do is you say that the volume of the particles are under
consideration right. So, let me put it this way ok.
309
Let us do an example and I will come back, let us say that I have a cube say that the
dimensions of the cube is a, I would like to find the what is the a ϕ S for this cube ok. So,
𝑆
(𝑉𝑃 )
𝑃 𝑆𝑃𝐻𝐸𝑅𝐸
𝜙𝑠 =
𝑆
(𝑉𝑃 )
𝑃 𝐶𝑈𝐵𝐸
4 3
𝑎3 = 𝜋𝑟
3
𝑆
(𝑉𝑃 )
𝑃 𝑆𝑃𝐻𝐸𝑅𝐸 6/𝐷𝑃
𝜙𝑠 = = = 0.806
𝑆 (6𝑎2 /𝑎3 )
(𝑉𝑃 )
𝑃 𝐶𝑈𝐵𝐸
Now, let us go back to the definition ok. So, when you are extracting the size by equating
the volume right when you are extracting the size by equating the volume your V P here
and V P here is exactly same right, what are you doing is you are taking up a sphere and
you are saying volume of the sphere is equal to the volume of the particle in the
concentrations there.
There therefore, your ϕ S definition can also be the surface area of a sphere divided by
surface area of particle under consideration of course, whose normal size is D P right. So,
you can either express it as S P by V P of a spherical particle divided by S P by V P of a
310
particle of whatever arbitrary shape that you are considering as long as your diameter
nominal diameter is the same as a sphere or you can also express it in terms of the surface
areas itself ok. What you do, but no, but of course, when you want to get the equivalent
diameter you would have to equate the volumes, volume with the sphere is equal to the
volume of the particle under consideration you get the nominal size of the sphere and of
course, you have to use that for your calculation ok.
Now, so this number that we got right 0.806 does it make sense for you? Because we know
that what do you think will be the value of ϕ S it cannot, it will is going to be always less
than 1 right because we know that you know for a given material sphere occupies the least
surface area right ok. Therefore, this is always going to be less than one of course, if you
take a spherical particle your ϕ S is going to be one because the surface area exactly the
same, but any deviation from spherical shape is going to lead you to a value which is less
than 1, the more closer to 0 it is; that means, it is highly non spherical ok.
That means, if I take a very very long needle like particle for example, maybe this pen. If
I take this and if a calculate a sphericity of this and if I compare the sphericity to or
something like this, but a shorter length of course, the sphericity works the shorter one is
going to be larger compared to the longer needle right.
So, therefore, this sphericity kind of tells you something about how do I think about kind
of considering the shape effect just by you know. So, you will look at lot of expressions
which are kind of developed for spherical particles, but the deviation from the spherical
shape is kind of accounted for by incorporating is sphericity into the calculations ok. So,
in that context this is an important parameter that you will come across you know maybe
in some calculations ok.
311
Now, we also said; so, therefore, so the size is important and we also said shape is also
important we have discuss some aspects of shape, what about specific surface area? It is
typically you know defined by ASP; specific surface area, it is basically defined as the
surface area of a particle divided by the mass of the particle ok. It typically has units of
you know sorry it has units of meter square per kg right.
If you take a particle a spherical particle for example, your 4 pi R square is your surface
area right if you take mass of the particle I can express because density is equal to mass
by volume ok, I can express mass in terms of density the particle into volume of the particle
right, yeah. So, therefore, I can write this as 4 pi R square divided by ρ of the particle into
4 by 3 pi R cube right therefore, this becomes 3 divided by ρ times R right.
𝑆𝐴 𝑚2
𝐴𝑆𝑃 = ( )
𝑚𝑃 𝑘𝑔
4𝜋𝑅 2 4𝜋𝑅 2 3
𝐴𝑆𝑃 = = =
𝜌𝑉𝑃 4 𝜌𝑅
𝜌 3 𝜋𝑅 3
So, this specific surface area is an important quantity because when people are working
with like say any surface reactions or anything that involves surfaces, what would be your
freely interest is what is the total surface here that is available for any process ok. For
example, if you look at surface reactions or adsorption and you know things like that what
312
is really important is given a certain mass of material, what is the total surface area
available ok.
So, if you especially if you look at particulate samples which are smaller in dimensions
like nano particles and you know and colloidal particles and things like that one of the
parameter for which all the particles would be characterized is something called as specific
surface area, which is the surface area of the particle divided by the mass of the particle
ok. And of course, as you would see it would depend on the size of the particle, it would
depend on you know the shape and you know parameters like that.
The last thing that I wanted to quickly brush up on is talking about density of the particle.
So, how does one measure density of the particle? So, I will give, I have a like say either
a liquid that contains some particles or say I have some powder sample ok, it could be a
dispersion or it could be a powder sample and I would like to measure what is the density
of the particles you know that I have in the system.
So, what do you think is the best way of doing this, some thoughts? So, this there is a and
there is a technique called; there is a technique called solution densitometry ok, what is
done is this ok. Say that I have prepared dispersion say that I could have a powder sample
or I could have a dispersion given to me if I have a way of making up something called as
a stable dispersion. What I mean by stable dispersion is that you know I have a dispersion,
if I take samples from different you know locations you know within the container they
313
should have same number of particles and they should exists as individual particles and
things like that ok.
So, if I have something like that, if I am able to prepare a stable dispersion what I can do
is I can use something called as a ideal mixing rule. And I can say that volume of the
dispersion that I have prepared volume of the dispersion that I have prepared is equal to
volume of the particles plus volume of the a solvent right or your dispersing medium if
you want to call it as. Of course, this is if your particles are not reacting and things like
that right as long as you have inert particles in the fluid this is going to be true ok.
Now, so the; so what I can do is I can manipulate this equation ok; I can manipulate this
one of the things that I can do is I am going to divide this by the mass of the dispersion
everywhere right. Now, what I am going to do is I am going to multiply this V S by M D
term right sorry this is M S right sorry M D. I can the second term I can multiply it by
mass of the solvent divided by mass of the solvent I can do that right and this term I can
multiplied by mass of the particle divided by mass of the particle yeah.
𝑉𝐷 𝑉𝑃 𝑀𝑃 𝑉𝑆 𝑀𝑆
= +
𝑀𝐷 𝑀𝐷 𝑀𝑃 𝑀𝐷 𝑀𝑆
Now, if I look at volume of the particle divided by the mass of the particle, this is 1 over
ρ of particle right multiplied by, so you have mass of the particle divided by mass of the
dispersion. Similarly, this term is going to be 1 over ρ of solvent multiplied by mass of the
solvent divided by mass of the dispersion. Similarly, this is going to be 1 over ρ of
dispersion right I can do that.
1 1 𝑀𝑃 1 𝑀𝑆
= +
𝜌𝐷 𝜌𝑃 𝑀𝐷 𝜌𝑆 𝑀𝐷
𝑀𝑃 𝑀𝑆
= 𝑋; = 1−𝑋
𝑀𝐷 𝑀𝐷
314
Therefore,
1 1 1
= 𝑋 + (1 − 𝑋)
𝜌𝐷 𝜌𝑃 𝜌𝑆
1 1 1 1
= + ( − )𝑋
𝜌𝐷 𝜌𝑆 𝜌𝑃 𝜌𝑆
What this tells you is that if I have a way of measuring; if I have a way of measuring 1
over ρ D that is a dispersion density as a function of mass fraction, if I have a way of doing
that you would get a straight line where your intercept is going to be 1 over ρ of solvent
and your slope is going to be 1 over ρ P minus 1 over ρ of S. Therefore, just by measuring
the intercept and the slope I am I can basically calculate what is the density of the particle
ok.
This is a simple solution densitometry root what is being done is there you basically
preparer or series of dispersions 5 or 6 samples, I know what is the mass fraction of
particles in the dispersion I go and measure the density of each of these stable dispersions
and I plot 1 over ρ of dispersion as a function of mass fraction just by measuring the slope
on the intercept I can actually calculate what is the density ok.
So, basically in today’s class what we have done is we kind of talked about basics of
particulate matter we said that; you know the all the particles whether you are going to
315
deal with solid particles, liquid droplets or gas bubbles they all are made up of atoms and
molecules there is something called as a intermolecular forces which holds these particles
together in these particles ok.
We talked about one functional form of interaction potential, L and J potential. So, you
should know a little bit about what is the strength of attraction or the you know the depth
of the energy well and what is the equilibrium separation distance between the particles,
then we talked about regular particles, irregular particles some way of characterizing them.
And we also talked about sphericity how do we think about you know thinking in terms of
characterizing irregular shaped particles for sizes and things like that ok. So, with that
maybe I will stop today tomorrow we will continue with some other aspects of fluid
particle mechanics, so.
316
Fluid Mechanica
Lecture - 30
Motion of a particle in a Fluid
We will talk a little bit about motion of a particle in a fluid. So, when we look at this title
right, motion of a particle in a fluid, do you come across cases where you see motion of
particles in a fluid? I mean I am talking a particle, but you know in a general sense it could
be a object right, it could be any object it could be any object. Can you think of examples
where you come across motion of objects in the fluid?
Yes right, some examples; I am sure all of you know right some examples of, it is a
submarine right so, if you have a submarine in you know you know in like say sea for
example, right it is moving underwater right. A lot of aquatic life ok, it could be whale
right, it could be fish right or it could be any other thing ok. You can you also come across
motion of bacteria in a fluid right, plus if somebody is interest in swimming right you can
think about swimming of people in a fluid right. So, there are a lot of instances where
motion of an object or a particle comes into picture ok. I am going to do is I am going to
ask you to do a simple calculation ok.
317
Let us think about your; I am going to have a table ok; I am going to say this is going to
be your object ok. Let us take three cases, maybe four cases; a whale and let us think about
a human or a man and let us take a bacteria ok. They are all swimming in water and the
properties of water like say is ambient temperature so, your ρ is going to be 1000 kg per
meter cube is the density of water and your viscosity, what is the value of viscosity of
water 25 degree centigrade? It is 1 into 10 power minus 3 Pascal second right; that is your
viscosity ok.
So, I now let us think about the characteristic features of these objects ok. Let us say that,
what do you think is a dimension of whale, characteristic dimension? What is the
dimension of whale typical dimensions?
Student: 20.
20 meters right. It is 10’s of meters. Let us say we can approximate it to be something like
say 10 meters with the order of meters ok. So, let us say its 10 meters what about humans?
What is that?
Roughly again, you know let us say of the order of meter right, 1-meter right is of the order
of meter. What about bacteria?
What is that? What do you think would be the size of the bacteria?
318
What is that? Micro?
Student: Micron.
Micron right, typical length scale could be of the order say 1 micrometer ok. There are
there are lot of bacteria’s, there are lot of viruses’ and typical dimensions of these things
are of the order of a micrometer. That is your characteristic dimension. What do you think
about the characteristic velocity with which these things move? What do you think would
be the speed if like say let us say bacteria, what do you think would be the speed ok? So,
they typically swim with the velocity which is the order of say 10 power 10 micrometer
per second ok. That is the speed with which you know they move in a fluid ok. What about
humans?
What is that?
For me about depends on you know from person to person. So, if you if somebody is a
great swimmer, you know they will do you know little faster right. So, typically again it is
going to be something like you know 1 meter per second. That is the velocity ok, you can
kind of you know swim the length of your body in a you know a second or something like
that ok. What about a whale? Similar dimension so, it is going to be something like 10
meter per second ok. That is the typical ok. Now I want you to calculate, what is the
Reynold’s number that is associated. I want you to calculate what is the Reynold’s number
that is associated with the motion of these objects in water ok. That is the question.
So, how do we go about doing it? You can just do that, just plug in all these numbers. I
would like you guys to tell me the Reynold’s number that is associated with the motion of
these three different objects, you know which have different length scales, which I you
know have a different velocities with which they are moving. I would like you guys to
calculate these numbers and tell me what is the Reynold’s number that is associated with
the motion of these fluids these particles. you do that quickly. All of you know right the
definition of Reynold’s number all of you know ok.
Reynold’s number is it is defined as inertial forces divide by viscous forces right; that is
what you learned and typically it is defined as ρ V bar some length scale by viscosity right.
That is your definition of Reynold’s number.
319
𝜌𝑉̅ 𝐷
𝑅𝑒 =
𝜇
So, can you calculate what is the Reynold’s number. So, mu is the viscosity of the fluid
right so, that is given to you right. What about ρ? Ρ is the density of the fluid that is also
given to you. What do you say about D? D is your characteristic dimension right. When
you talk about flow through a pipe typically D is going to be the diameter of the pipe right.
Now, because we are interested in looking at the motion of objects in the fluid ok, that D
should actually be the diameter of the particle ok, because the relevant length scale in this
case is the size of the object or the dimension of the object, which is swimming in a fluid.
Therefore, your Reynold’s number definition your your characteristic length scale is going
to be D P which is the diameter of the particle ok. That is one modification that you would
have to do for the Reynold’s number formula that you learned.
What about V bar? Again, you know flow through pipe problem V bar is your average
velocity with which the fluid is flowing through the pipe; however, in the case of fluid
particle you know systems ok. If you have an object which is moving in a like say a
stagnant fluid for example, then your V bar is going to be the characteristic velocity with
which the particle is moving in the fluid ok. So, with this you should be able to get what
is the Reynold’s numbers right. Can you just can someone tell me what is the Reynold’s
number for case one?
Student: 10 power 8.
What is that 10 power?
Student: 8.
10 power 8. What about the next one?
Student: 10 power 6.
10 power 6.
10 power.
Minus?
320
Student: 5.
Minus 5, right? Now you also learnt a little bit about laminar flow and turbulent flow right
in the class so far. So, now, there are certain range of Reynold’s number for which you are
going to consider the flow to be laminar flow. There is going to be a range of Reynold’s
number for which there is going to be a transition regime and of course, there is going to
be a range of Reynold’s number beyond which your flow is going to be turbulent flow
right, and when you talk about flow through a pipe right, if you have like say fully
developed flow you know your parabolic profile and all that what is your Reynold’s
number for laminar flow conditions?
What is that?
Approximately about 2000 right. Anything that is less than or equal to 2000 is going to be
it depends on textbooks so, is 2100. So, it is about 2000 anything less than that its laminar
flow. What about turbulent flow? That is your laminar flow, something about 4000 right.
Anything about 4000 is about is turbulent flow. In the case of fluid particle systems so,
these numbers get a little shifted ok. So, typically when you are you refer Reynold’s
number in terms of a Re with P at the as a you know subscript; that means, it is a particle
Reynold’s number.
If the particle Reynold’s number is less than 1 that is your laminar flow condition. If your
particle Reynold’s number is greater than 1000’s that is your turbulent flow condition and
anything in between is going to be your transition regime ok.
𝑅𝑒𝑃 < 1 → 𝐿𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤
𝑅𝑒𝑃 > 1000 → 𝑇𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝑓𝑙𝑜𝑤
Now, if you look at you know the flow through pipes and problems where you had you
know fluids flowing through conduits, your transition from laminar to flow it is actually
governed by you know your velocity of the fluid, viscosity, you know the parameters right.
However, it turns out you know what you have is a case where you have everything, which
is basically moving in the same fluid right, you have your fluid in which the objects are
moving is fixed, that is your water ok, with the given density and viscosity; however, all
that you change was the object right. You had a whale swimming you know a human and
321
a bacteria and you can look at you know the Reynold’s numbers right they are markly
different ok.
Therefore, the flow around objects are going to be very different than the flow or you know
in pipes ok. And so, typically when your Reynold’s number is less than 1, you can think
about it as a laminar flow conditions and depending upon the fluid object combination that
you are going to look at ok, you are going to the object is going to experience very different
kind of flow field around it ok.
Now we will move a little bit into so, we are going to take a simple case of fine particles
ok, in a fluid ok. What I mean by fine particle is, so, I have a container, I have particles
ok, the dimension of these objects if I were to think about dimension of these objects is
going to be in the range from about 1 nanometer to 1 micrometer ok. That is the dimension
of the object. So, typically people call them as colloids or nanoparticles right.
If you remember the movie that I showed a couple of classes ago; so, there was a movie
in which you had objects you know which were kind of moving around right and so, now,
if you take that such tiny particles and if I put them in the fluid, I am going to ask a question
as to what are the different forces acting on the particle?
I have tiny particles which are in a fluid and I would like to have an idea about what are
the different forces that are acting on the particle. Can you name some forces? Of course,
322
the first one is going to be; obviously, gravity right your gravitational force ok. So, if you
call it as F g ok, that is going to go as delta ρ there is a density difference ok. Delta ρ is
your ρ p minus ρ of fluid ok, where ρ P is a density of the particle and ρ F is the density of
the fluid times volume of the particle right times the gravitational constant right your g
acceleration due to gravity right. That is your gravitational force.
𝐹𝑔 = Δ𝜌𝑉𝑝 𝑔
Any other force?
What is that?
Student: Drag force.
Somebody says drag force. Do you think that is going to be a drag force? So, we you have
learnt that you know the drag force comes into picture whenever there is a relative motion
between the particle in the fluid and especially when you take the tiny particles 1
nanometer or 1 micrometer particle say it turns out you know they were moving around
and that moving around was because of what is called as a Brownian motion right.
And whenever there is a relative motion between the particle in the fluid, you know your
drag force is going to come into picture. People call it as drag force there is also another
term for it is called as a hydrodynamic force. You can also call it as hydrodynamic drag
and that if you say it as F D, any thoughts what is a functional form for this? I am sure all
of you have learned this at some point. It is 6 pi ok, mu which is the viscosity of the fluid
some average velocity with which the particle is moving and your R or P or whatever right
your size of the particle ok. That is your drag force.
𝐹𝐷 = 6𝜋𝜇𝑣̅ 𝑅𝑝
Any other force? So, is gravity, is drag force, any other force that you think is important?
What is that?
Student: Friction.
Friction, drag they are the same know; when people also call this as frictional forces if you
want to write another name ok. Other one what is called as a Brownian force and this
323
Brownian force it goes something like as F B; it goes as K B T that is your thermal energy,
divided by R P ok. So, KB K B is something called as a Boltzmann constant right and T
is your absolute temperature the value of K B T is something like 1.38 into 10 power minus
23 joules per Kelvin ok.
𝑘𝐵 𝑇
𝐹𝐵 =
𝑅𝑃
𝑘𝐵 (𝐵𝑜𝑙𝑡𝑧𝑚𝑎𝑛𝑛 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡) = 1.38 𝑋 10−23 𝐽/𝐾
𝑇 → 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒
Let us think about again calculating these forces ok. I would like you guys to calculate
what is your what is the FD ok.
(Refer Slide Time: 18:21).
So, if you take a lets think about a simple case of 1 micrometer particle ok, like this is
going to be your R P and again let us go back to your viscosity of the fluid is 0.001 Pascal
second or 1 into 10 power minus 3 pascal second. Your density of the fluid is again 1000
kg per meter cube ok., let us take a particle the density of like say 1100 kg per meter cube
and if you take a 1 micrometer particle typically it is known that you know it kind of moves
around with the velocity which is the order of 1 micrometer per second ok.
324
That is a typical ok, if people have done these measurements, so what they have done is,
they have taken a microscope, they have observed the velocity with which the particles
move and that typically for a 1 micrometer radius particle it is of the order of 1 micrometer
per second. If you put in all these numbers can we get an in get values for say F g and F
B. You just quickly work it out.
What is that?
Volume is 4 by 3 pi r cube right. You know the you know the dimension of the particle
you can actually calculate what is your volume of the particle. Yeah what are the numbers?
Yeah somebody has numbers. What is F g? This should be of the roughly 4 into 10 power
minus 15 Newtons. This also should be of the order of or if you have these numbers let us
let us quickly check yeah. So, so the number that you will get would be your drag force is
of is of the order of 4 into 10 power minus 15 Newtons and your FB would also be of the
similar order ok.
𝐹𝑔 ~4 𝑋 10−15 𝑁
𝐹𝐵 ~4 𝑋 10−15 𝑁
Now, so what I can say from this calculation is that, if you have a particle of 1 micrometer
radius and if it is immersed in a in water and if the density difference is about 100 kg per
meter cube, it turns out that you know your F g and F B they are comparable right. Now
what happens if I reduce the size of particles? If I have like say instead of you know 1
micrometer, if I take say 10 nanometer particle ok. Obviously, what you will see you know
is you will see that you know your F B which is the a Brownian force is going to be much
greater than F gravity right ok.
Therefore, so whenever one is working with you know fine particle system it is a good
idea to kind of think a little bit about different forces are acting and also do a simple
calculation of the different force that will be acting on the particle that will give you some
information about whether your gravitational force are really important in terms of making
your dispersion or the fine particle dispersion unstable ok.
Whenever you know, when I drew this picture right so, now whether the particles continue
to remain you know in the solution or they would go back and sit at the bottom, that would
325
depend on the balance between the different forces that are acting on the particle right.
Now there is a gravitational force which wants to pull the particles down and you have
Brownian which wants to keep it in a suspended state right. Therefore, if you go for finer
and finer particle sizes it so, happens you know your Brownian force overweighs the
gravitational force that is one of the nice way of imparting nice stability for the particle
dispersion that you are working with ok.
Now, that we know a little bit about fine particle dispersion and different forces now, I am
going to ask a question as to so, I said that these particle fine particle they exhibit Brownian
motion right. Do you think there is any practical use of this chaotic motion that you know
the particle exhibit or the fine particles exhibit you know when you put them in a fluid?
Do you think they are any applications of it or any uses? Anyone has read a little bit about
Brownian motion? You read at some point for sure right in your plus two or you know
maybe before that right.
So, it turns out that you know the concept of Brownian motion was really useful and also
it is useful you can use this for actually measuring particle size ok, number one. Number
2, people have also used this Brownian motion to actually prove the existence of atoms
and molecules ok. Before there was a debate about you know the whether the molecules
are there atoms are there or not ok. So, people kind of use some nice simple experiment
that they could do by monitoring the Brownian motion of particles you know in a fluid and
326
they were able to kind of come up with a come up with the explanation for the fact you
know the molecules do exist, plus have you heard of Avogadro number right?
So, we know that you know 1 mole of any substance has 6.023 into 10 power 23 number
of atoms right or molecules right. Now, one of the proofs for the calculation of Avogadro
number, it actually came from measuring the Brownian motion ok. What someone called
John Perrin he is a French scientist; what he did is he used this Brownian motion concept
and from that he was actually back he back calculate what is called as a KB which is the
Boltzmann constant and there is a relationship between Boltzmann constant and R which
is a universal gas constant and your Avogadro number. So, he used that to back calculate
what is the Avogadro number. It turns out that you know the number that he got was very
close to you know what is known now ok.
So, we will talk a little bit about these concepts in the next class ok. So, I am going to start
with showing you again some more videos of Brownian motion ok. Yeah explain a little
bit about why is it important you know and what kind of measurement that people did and
then we are going to derive one very important equation; which is what is called as a
Stoke’s Einstein’s equation; Stoke’s Einstein’s equation which is one of the fundamental
equation that people use for measuring the particle size data ok. So, we will talk a little bit
about that in the next class. I will stop here today yeah.
327
Fluid Mechanics
Lecture - 31
Brownian motion and electrophoresis
One is to look at, one is to look at something called as Stokes Einstein relation this is
applicable to something called as Brownian particles ok. As, I was mentioning in the
previous class this is one of the widely used relationship for a or relation for finding of
particle size data ok.
If, you want if you have a particle dispersed in a fluid, if you want to measure it is
diffusivity, or if you want to measure it is particle size ok. This is one of the widely used
relation ok. We are going to talk something about Phoresis ok, which basically refers to if
you have a particle and in a fluid, and if you set it to motion ok. Motion of a particle in a
fluid what is called as phoresis.
And, you can set a particle into motion by using external field right. And, either you could
use something called as a concentration gradients ok, or you could use thermal field or
thermal gradient, or I can use some other force ok, it could be electrical forces you know
or magnetic forces right. So, if you set the particle into motion by using you know by
exploiting the concentration gradient, this is something called as a diffuso-phoresis ok. If,
328
you use thermal fields something called thermo-phoresis ok. And, if you use electrical
field it is what is called as a electrophoresis right ok.
So, we are going to look at some aspects of you know phoresis ok. This is what is going
to be the plan for today. So, I am just going to start by showing you again a movie which
I showed in the last class ok.
So, this is what has been done is that you know the white dots that you see, these are
particles ok. These are something called as a fluorescent particles ok, when people use
fluorescent particles the idea is that, if you use a regular optical microscope ok, which
works by using the white light, you can see objects up to about a micrometer or larger ok.
Of course, you can also look at things that are smaller than micrometer, but you cannot
resolve it ok, you cannot get a good image by using optical microscope. What people do
is they use fluorescent particles ok, by which I can actually push the limit of microscopy I
can also look at things that are much smaller than a micrometer size. You can even look
at things that are 200 nanometers, 300 nanometers ok.
So, this is sub micrometer particle and these are fluorescent particles and white dots as I
said these are spherical particles dispersed in the fluid, nothing is being done. All that has
done is you are just putting the particles in the fluid and you are just watching you know
329
with a microscope ok. And, what do you see is that particles are moving right, this is
without the influence of any external force they are just moving by themselves ok.
And, if you can see there is no particular direction in which the particles move right they
are just moving chaotically and that is because ok. So, you have molecules of the fluid
around each of these particles and there is a bombardment of the fluid molecules with the
particle ok. And, depending upon the bombardment it will just move in some random
direction ok, it could be in x y z any direction ok.
Now, such a chaotic motion is something called as Brownian motion right ok. Such a
chaotic motion ok, such a motion which is because of bombardment of particles with the
molecules of the fluid is what is called as a Brownian motion ok. So, the particles really
do not move in any particular direction ok, this is a random motion ok, which is purely
existing, because there are molecules and atoms that constitute the fluid in which these
particles are dispersed ok. Now, from yesterday’s lecture, we know that whenever we have
a relative motion between the particle and the fluid ok. There is going to be something
called as a drag force right ok.
We also said it is also called as frictional force ok, you can also call it as hydro dynamic
drag force right. Now, so, you can think about of course, we already know what this
hydrodynamic force is right yesterday we already wrote up an expression for it right, but
330
the other way of thinking about this you know drag force is going to be you can actually
bring in the concept of friction factor ok.
So, in this case C D is something called as what was that?
Oh it is. So, know oh my god is that ok. C D is something called as a drag coefficient and
it is defined something like this ok. Drag force divided by some area over density time
some velocity head. Do you recognized this formula, some similarity that you may have
seen in the previous lectures. So, if you look at this carefully, this is of the form your
friction factor, you know was something like wall shear stress divided by you know some
velocity square right, that is some something like that right that is what it was density ok,
multiplied by the velocity head and there was a wall shear stress ok.
𝐹
(𝐴𝐷 )
𝑃
𝐶𝐷 =
𝜌𝑢2
( 20 )
𝐴𝑃 (𝑃𝑟𝑜𝑗𝑒𝑐𝑡𝑒𝑑 𝑎𝑟𝑒𝑎) = 𝜋𝑅𝑃2
Now, your shear stress is actually force per unit area right. So, F in this is what is called
as a drag force and your A P is something called a projected area ok. And, this projected
area what is what is being done as if you have a spherical particle ok. If, it is that is fixed
331
in space and if there is a liquid moving over it ok. Now, what you do is you look in the
direction perpendicular to the flow ok. What is that area that you see ok.
If, you have a spherical particle, if I look perpendicular to the flow what I would see is a a
circle right. So, your A p is going to be area of the circle that you see, that is your projected
area ok.
And, of course, rho is the density of the fluid and u 0 is the velocity with which the your
particle is moving in the fluid ok. Now, lot of people have done, so now, again we did this
calculation yesterday if you take the Reynolds number for the particle, we said you know
it is going to be a diameter of the particle times some average velocity, times rho the
density of the fluid divided by mu right. And, it turns out that because of the particle
dimension that we are working with this is going to be much much less than one right,
yesterday we calculated this is going to be of the order of 10 power minus 6 ok.
𝐷𝑃 𝑢0 𝜌
𝑁𝑅𝑒,𝑃 = ≪1
𝜇
Now, whenever you are working under conditions where the Reynolds numbers are much
much less than 1, we said that comes into some kind of a low Reynolds number of flows
or the laminar flows right. And, people have done a lot of experiments and they have found
that this C D for the Reynolds number much much less than one goes as 24 divided by the
Reynolds number ok.
24
𝐶𝐷 =
𝑁𝑅𝑒,𝑃
Now, what I can do is, I can take this relationship and substitute for C D in this equation,
I know what is the average velocity ok. I know what is rho and as I said your A p is the
projected area which is you know pi R p square. If, you substitute that can you just let me
know what your FD is can you just do this simple math? So, you substituted for C D as 24
by the Reynolds number and an Reynolds numbers is given by this ok.
332
And so, your A p is pi R p square or if you want to you know is pi D p square by 4 right
ok, if you do that can you guys let me know what FD is right.
𝐹𝐷 = 6𝜋𝜇𝑢0 𝑅𝑃
So, you basically recover back the Stokes law right that is what we are done right, now if
this is your velocity with which the particle is moving ok. If, I know what is the average
displacement over time is what you your average velocity right. Now, I have a movie in
which the particles are moving ok. I can track you know the particles in every frame of the
video; I can get their location ok.
And, therefore, I have a way of calculating, what is the average displacement of the particle
ok, per unit time is going to give me what is the velocity with which the particles are
moving right. Now, if you if I want to calculate what is so, now, that I know that there is
a particle that is moving there is a drag force. So, I want to calculate, what is the energy
that is required to move the particle? Ok.
So, how do I calculate that, I want to calculate what is the energy that is required to move
the particle right. So, the work that has to be done for moving the particle or the energy
that is required to move the particle is essentially your FD times x bar right. Your force
multiplied by distance and we know that you know it is moving it has more than average
distance x bar.
333
Now, so, this energy that is required to move the particle, where does it come from it comes
from the thermal energy right your K B T. So, we said that you know the reason why the
particle are moving is because the solvent molecules are hitting the particle ok. And, why
do solvent molecules are hitting the particle, because your fluid is at a is at a finite
temperature ok, whenever you have a finite temperature.
So, you are going to have every molecule is going to be associated with the energy which
is given by Boltzmann constant times your temperature ok. Therefore, your the energy that
is required to move the particle, which is FD times x bar actually comes from the thermal
energy ok. Now, what I can do is I can substitute for FD is 6 pi mu u naught is x bar divided
by t times R p times x bar is equal to K B T right.
𝐹𝐷 𝑥̅ = 𝑘𝐵 𝑇
𝑥̅
6𝜋𝜇 ( ) 𝑅𝑃 𝑥̅ = 𝑘𝐵 𝑇
𝑡
I can rearrange this and I can write it as x bar square by time is equal to K B T divided by
6 pi mu R p right. Yeah. And, this is D which is the diffusivity of the particles in the fluid
ok. And, it should have dimensions of meter square per time right meter square per second
ok. Therefore, your diffusivity of the particle is equal to K B T divided by 6 pi mu times
R p, this is what is called as a the Stokes Einstein relation ok.
334
𝑥̅ 2 𝑘𝐵 𝑇
𝐷= =
𝑡 6𝜋𝜇𝑅𝑃
This is a simple arguments of you know getting this equation, but there are other ways of
doing this ok.
So, all that we did was we said that you know there is a particle is in motion and because
of it there is going to be a drag force ok. And, we said that this is basically Stokes drag
because of the fact that it knows numbers comes out to be much much less than 1 ok. And,
the energy associated with moving the particle is your drag force times the average
displacement ok. And, that has to come from the thermal energy just by equating that, you
will get in the relationship between the diffusivity of the particles in the fluid, and the
thermal energy. And, this is some kind of a friction factor ok. People also write this as K
B T divided by f friction factor which is basically 6 by mu R p yeah correct.
Yes, why is it not the case? So, all that we are seeing is you know I mean ok. So, this is
going to be I would say this is kind of more of an ideal idealized case right ok. And, it
turns out that you know of course, if you take particles which could be rough for example,
if the particle is kind of set into rotation, because of you know this bombardment.
In such cases you would have to worry about you know accounting those you know cases,
but if you take like say rigid spheres, smooth spheres of course, there are some kind of
assumptions you know where you can use this, but this is typically applied for rigid
particles ok. And, the particles has to be have to be in colloidal size range of course,
because they have to exhibit Brownian motion. And, of course, this is for a smoother
surface and things like that ok.
Of course, you know in real systems you know, but it turns out you know this kind of this
equation. Typically is valid for you know a lot of colloid that people use and you know
every days you know experiments or life so, yeah ok. So, now that we know that there is
a relationship that exists between your diffusivity and you know K B T and in the friction
factor, what someone called John Perin ok.
335
Did was he basically did the very first experiments ok. On quantifying Brownian motion
ok, this happened around 1920s ok. And all he did was he basically had this video of the
particles moving in a fluid and he tracked the particles. So, what I what he did is this is a
the you know the frame of the movie that I have got, I have one particle here. And, what I
do is add some if this is time t ok. I take at some time other time interval say t plus delta t,
I look at where the particle is ok. Now, I look at another frame say t plus 2 times delta t, I
look at where the particle is ok.
So, so, in the end what he did was he basically came up with a picture of the location of
particle as a function of time right. So, you know it was basically exhibiting some chaotic
motion. So, all that he had was he had some information about the time frame ok, that is
you know the distance you know he had information about what is the delta t between the
2 frame that he was considering. And, of course, he had tracked the particles; that means,
he had access to the particle position as a function of time ok. So, once he did that. So, he
calculated what is called as a mean square displacement ok.
336
And, that mean square displacement is given by something like this ok. All, you have to
do is if you have basically tracked you know the particle for a time t ok, if you have
observed it for enough time. So, what he what he did is your x at time t i plus t minus x at
time t i whole square plus y at t i plus t minus y at t i whole square right, divided by n is
the mean square displacement ok. All that you do is you basically know the positions right
ok. And, if I have like say I have taken 10 frames right, I know the distance it travels in
every frame ok. So, that is your mean square displacement.
2 (𝑡)
(∑𝑛𝑖=1[𝑥(𝑡𝑖 + 𝑡) − 𝑥(𝑡𝑖 )]2 + [𝑦(𝑡𝑖 + 𝑡) − 𝑦(𝑡𝑖 )]2 )
< Δ𝑟 >=
𝑛
Now, that the mean square displacement could be measured by experiment what he did is
he made use of the theory of of Brownian motion that Einstein had developed ok, which
basically said that, if you observe a particle for a for long enough time that is for t going
to infinity ok. Your mean square displacement ok, go says 6 times D t, where t is your
observation time and D is the diffusivity of the particles ok.
𝐴𝑠 𝑡 → ∞; < Δ𝑟 2 (𝑡) ≥ 6𝐷𝑡
Now, because he was able to measure this ok, Perin was able to measure it he can actually
calculate what is diffusivity ok. Now, your Stokes Einstein relationship is K B T divided
by 6 pi mu R p ok. Now, he was careful enough to pick particles of exactly same size ok,
337
because he had a access to an electron microscope, he basically took particles which are
about 0.53 micrometer in radius, he was able to work with well characterized particles.
So, he knew R p ok. And, because he could do Brownian motion experiments, he was able
to calculate D, that is known now ok. And, of course, he used water as a fluid in which the
particles were put ok. He knew the viscosity from this what he did is he calculated K B ok,
the Boltzmann constant ok. And, we know that Boltzmann constant is basically your
universal gas constant divided by the Avogadro number ok. That is why that is how he
was basically able to calculate what is the Avogadro number and the number that he got
from his experiments, as I said yesterday was very close to 6.023 into 10 power 23 ok. So,
this is yeah go ahead.
Which one?
Student: The mean square.
The mean square displacement.
Student: Yeah.
So, if you look at the video right. So, you had particles all the particles were kind of
exhibiting some motion right. Now, what I can do is I can actually go and pick one particle.
I can look at that particle as a function of time ok. Now, in each frame I locate the same
particle ok. Now, so, therefore, what I can do is for every frame, I can associate a location
right, which is basically your x comma y ok, because you are looking at in 2 dimension
ok.
Now, once I have access to the position of the particle ok. And, if I know what is the time
frame that I am basically considering from my the video right. I can choose that hey I have
a like say video for say 5 minutes, I am going to look at every 5 seconds you know what
is the location. So, if I have information I can basically back calculate what is it the mean
square displacement yeah ok.
So, you know so, as I said right this is a useful equations which you should remember, you
know and it is kind of used in a lot of commercial instrument that people buy for measuring
particle sizes. The basic equation that basically goes into the calculation of the sizes is this
particular equation ok. They will have some way of measuring diffusivity ok. It could be
338
done either you know if you use a microscope. There will be some algorithm by which I
can actually calculate the mean square displacement from which I can calculate D ok. Plug
in that D here ok, I can calculate you know R p, which is the size of a particle right or there
are other ways of getting D ok.
And, there are some instruments where what people do is they measure, if I have a particle
in a fluid you shine a laser light ok. Because, the particle is moving ok, you detect what is
called as a scattered intensity ok. And, because of the chaotic motion of the particle the
laser intensity also fluctuates and from that fluctuation there is another way of back
calculating your D ok.
So, no matter what the instrument is ok, whether you get d from the fluctuations of the
scattered intensity or by using some kind of particle tracking. In the end the size of the
particle is basically determined by using stroke size in Einstein relationship ok. So, we will
move on to the next topic, which is for phoresis ok.
In specific we are going to talk about something called as electrophoresis ok. Now, for
electrophoresis so, what you do is. So, all of you know if you have like say, I have a cell
say, if I have a cathode for example, and an anode right. Now, if you have ions in solution
ok. If, you have like say positively charged ions or negatively charged ions in solution like
this is say this is going to be your a fluid right.
339
Now, we know that you know the moment you apply some kind of a voltage difference
right a potential difference, we know that you know ions are going to move right. Now of
course, you know the positive charged ions are going to move towards the cathode right.
It is going to be moving this way and negatively charge is going to be if this is your
negative and positive right. So, negative ions is going to be moving in this direction right.
Now, so, we know that the charged ions do move when you apply an electric field right.
Now, similarly you can also consider a charged particle ok; say that you know I have a
particle that has charged ok. I can consider this is a macro-ion ok. So, now, if I take Na
plus Na plus right so, the charge on that is your valency times e right, where e is the
fundamental elementary charge right. So, in this case for Na plus Ze is equal to plus one
times e right that is your the charge on Na plus ok.
Now, if I have Ca 2 plus is going to be 2 times e right. Similarly, if I have a Macro-ion say
that you know there are some 10 charges on the particle surface ok. Therefore, you know
I can say that you know there is there are 10 charges multiplied by e is going to be the
charge on the particle this that say q. Now, in electrophoresis what is actually exploited is
the fact that you know if you have similar 2 ions, the particles have to be charged for it to
move in a potential you know that you apply right.
340
Now, let us think about, how does the particular acquire charge? So, I have a particle ok.
And, that is charged any thoughts as to how the particles can be charged yeah have
something to say.
If, you have surfaces you know the particles would get charged right, I mean that you
would have you know learnt in your you know 10 plus 2 or something like that, but that
would happen you know that is because of friction right. Now, if you have particles in
solution ok. Now, we are talking about particles that are dispersed in a fluid ok. Now,
would have I would like to have them to be charged, there are different mechanisms by
the by which the particles can be charged ok. One is that if I have a particle ok.
Now, the chemistry is so, advanced that I can actually graft the particle surface ok. I can
actually put some charges on the particle surface ok. I can maybe make the surface,
functionalized with some charges. Now, whenever I have particle like that, the moment I
put it in water, what will happen is this particle with you know this surface groups, it gives
0 COO minus ok. This mechanism of charging a particle is by, what is called as a surface
charge dissociation ok.
You can buy particles like this ok. If, you want particle with any functional groups, you
can get them in the market ok. So, if I say hey I want a particle with the COOH groups on
the surface you can buy it ok.
So, therefore, one of the ways by which the particles can acquire charge is that, you know,
you have a particle and there are some surface functionalization. This surface
functionalization is done by putting some chemical molecules right, molecules on the
surface. The moment you put such particles in water there is going to be some dissociation
and that is going to leave a particle charge ok, this is one mechanism. There are other
mechanisms, where if you have particles in a solution, if there are some ions in solution
ok.
It could be H plus or OH minus or you know or Na plus or so, whatever ok, then what
could happen is your ions can go and get adsorbed on the particle surface ok, there could
be a physical adsorption of ions on the particle surface ok. So, this is this mechanism is
something called as adsorption of ions right. It could be absorption of ions, it could be
adsorption of any charged molecules, it could be adsorption of charge surfactants, it could
be adsorption of charged polymers ok.
341
So, now that we have a charged particle ok. And, say that you know I am going to apply a
potential difference right. He is going to move right just similar to the ions that we talked
about. Now, if you apply an electric field E ok, if you are applying electric field E the force
the electrical force or you know if the force with which the particles move in the applied
field is given by the charge ‘q’ times E ok. So, you have done this at some point right. If,
you have you know coulombs law right, your force is q 1 q 2 by 4 pi R square right.
Now, if you want that is the force of attraction or repulsion between you know of course,
there is 4 pi epsilon 0 epsilon R right sorry yeah ok, that is your right. Now, electric field
is basically yours force per unit charge right is q if you say q 1, q 2 instead of that if you
have q squared all right. So, let us say I am going to do that ok, you have 2 charges right.
So, q divided by 4 pi epsilon 0 epsilon R times R square, you know that is your electric
field right. That is the definition of so, that is how the force and the field are related right
all of you know this right yes, no.
So, basically the way you know the electric field is defined is it is basically force per unit
charge ok. So therefore, if I am saying that I know I am applying an electric field E ok.
The force with which the particle start moving, either towards a positively charged
electrode or a negatively charged electrode is given by F eq, which is q times E and as I
said q is the charge that the particle carries ok. And, we have been saying that you know I
am going to consider that this is a Macro-ion.
342
So, if I say that you know I have a particle surface with let us say 10 charges, 10 dissociable
groups ok, your q is going to be 10 times e ok. But for in general if you can call it as q,
which is the charge that the particle has right. Now so, of course, you know if you did not
apply the electric field the particle was moving, because of the Brownian motion right.
Now, that now that you applied a field is going to first it is going to accelerate right; it is
going to start moving faster towards one of the electrode ok. And, how fast it moves it
depends on the strength of the applied field right.
Of course, you know it cannot accelerate infinitely right, because it is also going to feel
the drag, because of the presence of liquid is going to be slow down right, as it moves ok.
And, ultimately it is going to start moving with some stationary or a constant velocity and
that constant velocity is basically determined by the balance between F e l and your drag
force right, because you know whenever you have a particle moving in the fluid there is a
drag force.
Now, the drag force basically opposes the motion ok, ultimately there is going to be an
equilibrium and after that initial acceleration the particle slows down and it is going to
start moving with a constant velocity ok. Now, this F e l is q times e ok, and F D is your 6
pi R p times you know mu times some velocity v right. That is the velocity with which the
particles would move right, after that initial acceleration.
Therefore, I can actually get what is v by E, which is the velocity which with the particle
is moving divided by the applied electric field that basically goes as q divided by 6 pi R p
into mu right ok.
𝐹𝑒𝑙 = 𝐹𝐷
𝑞𝐸 = 6𝜋𝑅𝑃 𝜇𝑣
𝑣 𝑞
=
𝐸 6𝜋𝑅𝑃 𝜇
This velocity, velocity per unit field is something called as a is denoted by a quantity ‘u’,
which is called as electrophoretic mobility ok. So, the velocity which with the particle is
moving per dived by the field that you apply ok, that velocity is what is called as
electrophoretic mobility.
343
And, if you have a way of calculating what is the electrophoretic mobility ok, you can if
you know the parameters of the particle and the fluid that you are using, I can actually
calculate what is q that is the charge on the particle ok. Therefore, what people do is they
p people exploit this concept called electrophoresis; that means, the fact that I can put up
I can set a charge particle into motion by applying electric field, I can exploit that to
basically measure the charge of the particle ok.
And, again the experiments are going to be very similar ok. What they do is they will
construct a rectangular cell ok, that has 2 electrodes an anode and a cathode you put up a
dilute dispersion of particles. The reason why dilute dispersion of particle is important is
because it turns out, that this expression that we wrote up right. If, F e l is equal to q times
E it is actually valid only under infinite dilution; that means, the number of ions that you
should have in solution or the number of charged particle that you have in solution should
be very very few.
Because, if that is not the case what is going to happen is if there are a lot of particles you
know next to each other ok, the motion of one is going to be influenced by the other as
well right. So, you would have to avoid that ok. So, this kind of formalism is valid only if
you are working with a very you dilute cases ok, where the concentration of ions in your
solution is very very less, or if you want to talk about colloidal particles that your the
particles number of particles in your dispersion is going to be very very few ok.
So, and the and the experiment that is done is as I said we take a rectangular cell you put
in 2 electrodes ok, you put in this dilute dispersion. I can put this below a microscope I
apply a known electric field ok; that means, I know what I am applying and basically I
know what is the either I am applying ok. Now, I am recording a video of how the particle
is moving, because I can track particles I can actually get what is the velocity with which
the particle is moving right. All, I have to do is take different frames I locate their position
and from that I can get that what is the velocity with which the particle is moving.
So, therefore, I have some way of calculating at the electrophoretic mobility ok. Once, I
do that ok. If, I know what is the if I have some other independent way of measuring the
particle size either by you know some using an electron microscope or you know some
other means I know R p. If, I know what is the fluid, that I am using ok, I can basically
344
calculate what is the electrophoretic mobility. Again, this is again one of the very common
technique that everybody uses for measuring the charge on the particle.
345
Fluid Mechanics
Lecture - 32
Sedimentation and separation
So, we are going to continue with the same theme that is.
So we are going to continue the same theme that is the motion of particles through a fluid.
So; however, we are going to look at the case of non-Brownian particles that is the particle
that are much larger in size. So, it turns out that you know the motion of larger particles
there is non-Brownian particles is kind of you come across this in a lot of processing
operations any example. So, can you think of some examples where you come across
motion of particles in a fluid?
So, when you concern operations such as say removal of say dust particles from air or you
could also have a case where there is a flue gas which is coming from you know one of
the industries right and if this industry could be carrying a particle ok. So, in general you
are the particle that we are going to consider like other cases it could be a solid particle, it
could be a liquid drop or a gas bubble and again your fluid can be either you know air or
a liquid right or a gas or a liquid ok.
346
So, there are several operations in which you have particles moving in a fluid one example
could be as I said the removal of dust particles you could also have cases where you have
like say solid particles in a waste fluid and if you want to like say discharge the waste
liquid, you may have cases where you know you would like to recover particles from liquid
waste plus you also have cases where in a lot of industries which produce acid ok. You
have acid mist is basically is fine droplets this is basically acid and before you exit a mist
you know into atmosphere you would have to recover you know this droplets from the you
know the waste ok.
So, there are a lot of cases where you do come across motion of either solid particles, liquid
you know droplets or gas bubbles in the fluid and therefore, whatever we are going to do
today is going to be of interest in that context of some of these operations now whenever.
So, we said that you know when you think about I have some issue ok. So, now, so, when
you had fine particles right we kind of saw that you know there the particles were moving
by themselves of course, there was a thermal energy because of the presence of fluid ok,
but in a sense they were moving by themselves.
But whenever you look at larger particles you know what you need for the particles to
move is it? External force right ok. And this external force can either come from density
difference that is the density difference between the particle and the fluid ok. So, can you
think of an external force where the density difference can be exploited to you know move
347
the particle under consideration particle right. Can you think of an example? Can you think
of an external force where I can exploit the density difference to effect the motion of
particle in the fluid? Gravity right ok. So, your gravitational force is an example of a case
where I can actually exploit the density difference to set a particle you know in the fluid
to move ok.
Any other example where I can exploit the density difference for setting a particle to move?
What is that osmosis why is osmosis related to density difference osmosis is basically a
typically you have some kind of a permeable membrane right where you know you can
look at separation in terms of you know there is a. So, there could be a membrane which
is permeable only to some species right. So, in that context you can use that, but can you
think of centrifugal forces like we know centrifugal forces again one force that I can
actually exploit the density difference any other external force that I we can think of again
something we discussed the other day it could be electrical forces, it could be magnetic or
it could be thermal forces right.
So, essentially so, you can set a particle into motion by using any of the external p. So, we
going to see some examples where we are going to look at the motion of a particle under
the influence of gravitational force and centrifugal force in this class ok. Now so, we will
think of a simple example ok. So, let us think about simple example.
348
So, I have a like see a column and this column is say is filled with fluid and let us say that
the densities of fluid is ρ, the viscosity is mu and I have a particle to begin with you know
may be sitting right on at the top of the column I am holding it and let us say that the mass
of the particle is m and ρ P is the density of the particle ok.
And I just let the particle move down under the influence of some external force. So, let
us say then the external force is acting on the particle is F e ok. So, whenever a particle is
set into motion say that you know there is an external force that is acting on the particle
say that is a F e ok. Now we have kind of discussed in the class that whenever there is an
external force which sets the particle into motion, there is always going to be a drag force
right whenever there is a relative motion between the particle in the fluid, a drag force
comes into picture therefore, there is going to be also drag force acting on the particle ok.
Now, when the particle enters the fluid, it is going to displaced its going to displace some
fluid as well right therefore, there is also going to be buoyancy as well right there is going
to be F b, this is your external force this is your drag force, this is your right buoyancy or
buoyant right now what you can do is you can think about writing a simple force balance
ok. So, we can use Newton’s second law that is your F is equal to m a ok.
So, can we think about the direction of each of these forces? So, if say that you know the
external forces say acting down. So, if you have now that you know you consider particle
in the fluid. So, you have maybe let us make it simpler. So, I have a particle in a fluid this
is your F e which is say acting downward what could be the direction of F D which is the
drag force is going to be acting up right that is your F D because we know that the drag
force opposes the motion of the particle and typically it acts in the direction opposite to
the external force and of course, its direction is you know its going to be parallel to the
external force as well and buoyancy. So, because if this the fluid is being displaced up
right therefore your buoyancy is going to be acting in the upward.
Somebody has a question? So, its going to be acting up right. So, therefore, my force
balance basically becomes F e minus F D minus F buoyancy. Now that is mass times
acceleration if the particle is moving with a velocity v if you have a particle in the fluid
and if it is moving with a velocity v your acceleration is d v by d t ok. Now in general your
F e which is the external force, I can write this as some mass time some acceleration a e
that a e would depend on the type of external force that you are considering if you are
349
going to consider gravity its going to be m times g right and if you are going to consider
the centrifugal force it could be something else right.
𝑑𝑣
𝐹𝑒 − 𝐹𝐷 − 𝐹𝑏 = 𝑚
𝑑𝑡
𝐹 = 𝑚𝑎𝑒
Therefore, I could write m times a e that is your external force minus F D how do we get
an expression for F D? We will go back to the definition of the drag coefficient ok. So, we
had said that your CD is F D divided A P divided by ρ u square by 2 now ρ v square by 2
now because you know v is a velocity which was the particle is moving therefore, I can
write F D as CD times ρ v square by 2 times a p right that is your right that is your the drag
force. Therefore what I can do is I can substitute for these in the expression and I can write
sorry with that ok.
𝜌𝑣 2
𝐹𝐷 = 𝐶𝐷 𝐴
2 𝑃
M times a e minus CD times ρ v square by 2 times A P minus. So, you have a buoyancy
right F b. So, this you actually obtain from the Archimedes principle right if m is the mass
of the particle that you are you know that you are considering, m divided by ρ P is going
to give me the volume of the water that the particle is going to displace what is that ρ P is
350
a density of the particle right times ρ which is the density of the fluid that is going to give
me what is the mass of the fluid that is displaced times the external force external you
know a e which is the acceleration because the external force is what is going to be your f
b. Therefore, I can write this as m divided by ρ P into ρ times a e should be equal to m into
d v by d t right.
So, I can do a little bit of rearrangement:
𝜌𝑣 2 𝑚 𝑑𝑣
𝑚𝑎𝑒 − 𝐶𝐷 𝐴𝑃 − ( ) 𝜌𝑎𝑒 = 𝑚
2 𝜌𝑃 𝑑𝑡
𝜌𝑃 − 𝜌 𝐶𝐷 𝜌𝑣 2 𝑑𝑣
𝑎𝑒 [ ]−( ) 𝐴𝑃 =
𝜌𝑃 2𝑚 𝑑𝑡
Now this particular equation it is the governing equation for something called as a 1
dimensional flow of particle through a fluid and the reason why is 1 dimensional is because
you are basically considering a particle which is moving in a direction where the direction
of motion of the particle is parallel to the direction of the external force and the buoyancy
ok.
But; however, you may have cases where that need not be the case you could have a case
where let us say a simple case of instead of you know have having a particle settled in a
column of fluid like this if I tilt the column itself ok. And if the particle is still settling you
know under say the action of gravity for example, you know you are the direction of
motion will not be parallel to the direction of the external force and the buoyancy ok. In
such cases you know your particle will not move in the a 1 dimensional fashion it could
be moving in a 2 dimensional fashion or 3 dimensional fashion that depends on the kind
of you know the problem that you are trying to work with.
At least in this case we are considering a case where the particle is like say in the case of
gravity is dropping down you are then the external force is basically acting external force
as well as the buoyancy is basically acting parallel the rest direction may be different, but
they are acting parallel to the motion of the particle. Now so, now, it turns out that if you
look at this expression right.
So, I can simplify this further right if I consider the external force F e to be a gravity I can
write this F e to be you know your m times g right are or if you consider external force to
351
be some kind of a centrifugal force its going to be m times r times omega square right
therefore, I can substitute for these I can simplify this further ok. One interesting thing to
note from this equation is that if you look at like say let us write this equation for the case
of gravity ok. So, that is ok.
So
𝑔 𝐶𝐷 𝜌𝑣 2 𝑑𝑣
(𝜌𝑃 − 𝜌) − ( ) 𝐴𝑃 =
So, if you look at let us say gravitation you know the case of gravitational you know
settling of particles are you know the flow motion of particle that and the gravity this is
your constant right you know your once you fix a fluid particle system and once you fix
the external force to be gravity in this case your there is a constant term ok.
That means if I go back to this example right. So, I had a particle. So, the moment I drop
the particle ok. So, your this term is fixed right; however, if you look at this term right
because of the fact that the particle is going to accelerate right the particle is starting from
rest the moment I drop the particle is going to accelerate because of which this is this term
is going to go on increasing right because you know if this term goes as v square if the
particle starts to go to higher velocity therefore, this term is going to go up you know go
on increasing.
352
However this cannot increase forever right at some point you’re going to reach a case
where this term would be equivalent to this term that. So, under these conditions the
particle is said to start moving in a fluid with a constant velocity and because its going to
move with the constant velocity your d v by d t is going to be 0 and this constant velocity
is the maximum velocity that the particle can achieve under a given fluid particle
combination ok. So, therefore, I can equate this to 0 and get an expression for that constant
velocity which is what is called as a which is typically divided by u t which is what is
called as a terminal velocity ok.
That means under the action of all these forces the particle that initially accelerates through
the fluid it is going to reach a constant velocity at some you know point during its flow
and that constant velocity is what is called as a terminal velocity we kind of looked at the
application of this terminal velocity concept when you looked at the measurement of
viscosity of the fluid right. We said that there is something called as a we if we looked at
you know we looked at different cases of measurement of viscosity right there is some
case where this is called as a falling ball viscometer right.
Do you remember that there is something called is a falling ball viscometer in which what
is what you do is you make a particle you know flow through a fluid make it fall through
a fluid and then you basically exploit such you know 1 dimensional equation of motion to
really get an expression for that terminal velocity and that terminal velocity as we going
to see it is going to be related to the viscosity of the fluid in which the particle is moving
right. So, people kind of use this concept to get the viscosity of the fluid in commercially
available instruments for measuring viscosity ok.
So, let us do that. So, let us going to let us equate this to 0 and then get an expression for
sorry ok.
353
So,
2𝑔(𝜌𝑃 − 𝜌)𝑚
𝑢𝑡 = √
𝐴𝑃 𝜌𝑃 𝐶𝐷 𝜌
Now, the equation that we wrote right it is kind of a general equation that is valid for
objects of any shape ok. We really did not make any assumption that the particles are
spherical or cylindrical or whatever right. So, the 1 dimensional equation of motion that
we wrote right let us go back this expression or this equation is valid for particles of any
size you know or any shape right. So, now, what you can do is, you can now if your
working with the particles of specific geometry I can simplify this expression further and
I can actually get right.
So, now if I want to work this out like say if I am considering a spherical particle ok. We
said that this A P which is the projected area, we said there you know what you should do
is if you have a particle is moving in a fluid you should look at a plane perpendicular to
the fluid and you should look at what area do you see ok. If its a spherical particle your A
P is going to be 1 over 4 pi D P square right where D P square is the diameter of the particle
I can substitute it for A P and what about CD ok? So, what people have done is, people
have done a lot of experiment ok.
354
And kind of tabulated you know the results of you know the measurement of CD which is
a drag coefficient as a function of Reynolds number this is the CD that we have been
talking about ok.
So, that is your CD on the y axis your x axis is the Reynolds number ok. It turns out that
if you look at this plot there are 2 limiting cases there is a case where if you look at the
low Reynolds number case there is basically a line with a constant slope ok. This is what
we kind of exploited in the some of the previous examples where we wrote the CD to be
24 by Reynolds number right that basically corresponds to that Stokes law regime its what
is called as a Stokes law regime which is valid when the Reynolds number is less than 1
ok.
So, therefore, if I am working with conditions where the motion of the particle in the fluid
occurs under the condition such that the Reynolds number is less than 1 for CD I can
replace this as 24 by Reynolds number. However, if you look at the case where you know
the Reynolds numbers are larger you know for example,10 power 3 in that region there is
a dotted line, that dotted line basically corresponds to a case where your CD is more or
less constant and typical value of CD in that case is basically.
So, therefore, you basically have 2 limiting cases this is CD is equal to 24 by Reynolds
number that is for the case where you know your Reynolds number is less than 1 and you
have another case where CD is 0.44 that is for the case where you know your Reynolds
355
number is typically greater than 1000. Therefore, I can exploit you know the experimental
data that is available on a on particles of different sizes that are being done and tabulated
you know or all kind of which is available in the form of CD versus Reynolds number plot.
24
𝐶𝐷 = ;𝑁 < 1
𝑅𝑒 𝑅𝑒
𝐶𝐷 = 0.44 ; 𝑁𝑅𝑒 > 1000
I can substitute for these limiting cases and I can consider the of course, these are valid
only for spherical particles the whatever plot that I showed its valid only for the motion of
spherical particle in the fluid, if somebody is working with other types of particles or you
know other geometry you know of particles you would have to think about using an
appropriate CD versus Reynolds number plot ok.
Now that we have done that so, can we just replace let us think about the case one where
Reynolds number your CD is 24 by Reynolds number which is what is called as a Stokes
law regime can you substitute for A P and CD and get me an expression for the settling
velocity or the terminal velocity ok. Let us do that for both for the Stokes law regime and
also for the Newton’s regime which is for cases where you know your Reynolds number
is much greater than 1000, can you do that and let me know what the results are.
356
So, I can actually write m in terms of the density of the particle and the volume of the
particle right I can do that therefore, this ρ P ρ P gets cancelled of course, I can express v
P in terms of the particle dimension right if you know pi D P cube by 6 is your volume of
the particle right and of course, I can substitute for A P to be 1 over 4 pi D P square right.
So, if you substitute for all of that and of course, your CD is equal to 24 by Reynolds
number right I say do all of that the expression I am looking for is something like this u t
should go as g into D P square into ρ P minus ρ divided by 18 mu is what you would get
for something called as a Stokes regime ok. And your u t should basically go as should go
something like 1.75 into square root of g DP in to ρ P minus ρ divided by ρ that is for
Newton’s regime, which is basically you are simple substitution right ok. So, I should be
able to get this right ok. Just work it out and let me know if you have issues with these
things.
2𝑔(𝜌𝑃 − 𝜌)𝑚 2𝑔(𝜌𝑃 − 𝜌)𝜌𝑃 𝑉𝑃 𝑔𝐷𝑃2 (𝜌𝑃 − 𝜌) 𝑔𝐷𝑃 (𝜌𝑃 − 𝜌)

𝑢𝑡 = √ =√ = = 1.75√
𝐴𝑃 𝜌𝑃 𝐶𝐷 𝜌 𝐴𝑃 𝜌𝑃 𝐶𝐷 𝜌 18𝜇 𝜌
Now it turns out that you know the concept, that we have kind of discussed today it’s kind
of useful in a lot of cases where people are interested to separate particles from a fluid ok.
There is actually a process called elutriation say that you have a powder sample say that
you know I have you know some sample are given to me which basically consists of
357
powders powder and say that the powder has particles of different sizes ok. Now if the
objective is to separate the particles that are present in the powder into different you know
fractions larger size smaller size and stuff like that what is typically done is you take this
into some kind of a container you fill the powder here now what you do is, you let a
basically a gas or any you know liquid ok.
Now, what you can do is, what I can do is if I have a way of regulating if I have a way of
regulating the flow rate of the fluid that I am basically you know using then depending
upon; depending upon the velocity with which the gas or the liquid is moving up the
column depending upon the velocity with which the fluid is moving up the column if the
velocity of the fluid is larger than if the velocity of the fluid is larger than the terminal
velocity of a particular particle that is there in the you know in the container. So, either the
particle is going to be carried along with the fluid or will continue to remain in the column
ok.
So, therefore, having an idea about the terminal velocity, its also called as a terminal
settling velocity because when you are talking about motion of a particle under the
influence of gravity is basically settling right. So, therefore, depending upon the if you
have an idea about whether the particles are kind of smaller or larger, whether the Reynolds
number is you know less than 1 or more than 1000 depending upon these conditions. So,
you should be able to come up with some estimate of what is the settling velocity of the
particle.
Now, once I have an idea about that basically by regulating the flow rate that is the velocity
with which the fluid enters the column I can actually regulate that to separate particles in
a different fraction. So, therefore, so, if I want a particle you know to remove particles of
larger size of course, I would have to maintain a higher flow rate therefore, so, this is one
of the cases where the settling principles are kind of used for separation of particles by this
technique called elutriation ok. We will look at some applications of this in the class when
you do some tutorial problems ok. So, any questions so far?
Correct.
So, typically when we said that you know when you start the particles you know say then
you start with particle being addressed ok.
358
I see a lot of people talking. So, if you want to talk you can leave. So, say that you know
you start with particle being on at rest let you know and you leave it and let it move under
the influence of gravity. Now it turns out that you know that the typical time that it takes
you know for it to attain that terminal velocity, it’s usually very very small. In fact, you
have an experiment in the next semester what are you going to do is, you basically have a
column something like this what you do is, you have some markers on the column these
markers are typically used to basically find out what is the distance that the particles travel
in a given time ok. Now what I can do is, I can actually plot the velocity as a function of
time ok.
F you do that you will see that you know. So, basically what you do is an experiment you
know when the particle basically enters this you start a stopwatch when it leaves you
basically stop right. I know the dimensions the distance that it travels and I know the time
I can basically plot it right. It turns out that you know if you look at of course, when you
do this terminal velocity measurement of course, you do not look at that initial region you
know where the particle could still we are you know accelerating; however, it very quickly
reaches the terminal velocity ok. So, what you one of the best way to do this would be I
look at velocity at successive intervals, I look at you know at what instant you know the
velocity between 2 successive regimes is constant that is how people find out the terminal
velocity.
So, I agree that I know initially you know there could be some case you know where they
would also need not be terminal velocity, but typically you kind of work with a really large
column its not that you know. So, they basically the fluid that your using for separation it,
typically expenses a reasonable amount of time in the residence time is typically larger.
So, in that timeframe you know the particle does attain a terminal velocity and your
separation is typically based on whether the velocity of the fluid is either larger or smaller
than the terminal velocity of the given time ok. Any more questions no?
I just said that you know your you know your buoyancy right. So, if you this is basically
based on the Archimedes principle right all you have to do is I have a fluid I have a particle
that enters the fluid that is going to displace some volume of the fluid ok. I should calculate
what is the mass of the fluid that it is displace and time acceleration right. So, your external.
Correct.
359
So, I only said you know in this case F b I wrote it as m by ρ P times ρ into ae right typically
we you write it as m times g right. So, in this case g was the acceleration due to gravity I
said there you know I am going to use a general term ae which is the acceleration for a
given external force that is all. So, therefore, if I use gravity your ae is going to be g if I
use you know centrifugal force is going to be r times omega square right that that was the
only.
No it depends on the external force which you are considering for example, if I. So, when
you use external like say when people use centrifugal forces right typically you know
centrifugal courses forces are measured in terms of whether the centrifugal force is 1000
times g 500 times g and stuff like that right.
Now, when you do that you are you really. So, if the gravitational if the acceleration
because gravity is very very little then you know you are only talking only about the
dominant centrifugal force in that case in such a case your g has to come from the external
force that you act that you are using though this weight concept is true only when you are
talking about are the buoyant force coming from the weight right its only true only for the
gravitational force.
But however, if you use other external force you would have to worry about the
acceleration that comes from the external force correct right. So, now, no I mean it’s the
same thing right your this is your mass that is displaced right m by ρ P times ρ that is the
mass of the fluid that is displaced by the particle times your acceleration because m by ρ
P is a volume times g is your mass times the acceleration right ok.
Any other questions? So, I would like you to guys to go back and work out these equations
ok. Just write down this 1 dimensional equation put in appropriate terms I would like you
guys to derive those two equations the Stokes law and the Newton’s law because that we
are going to use them for some of the tutorial that you are going to do later.
But you know you see if I have like say let us think about let us think about something like
that I have a like say a play that is rotating now they say I have put a fluid with some
particle ok. Now if you look at you know the relative you know the I mean. So, what you
are trying to say is true when you know when you are the centrifugal force that you are
using there, it becomes kind of comparable to gravity ok; however, if you take cases where
360
you know you are really working at a very large rotation speeds ok. So, I think the motion
because of the gravity you can actually neglect.
But again now you know you would have to think about resolving it you have to find out
you know whether the gravitational force you know the gravitation activation and the
acceleration that is because the different you know are in the same direction are you not
worried about resolving and then take appropriate components for your analysis ok. So,
we so, you should also think a little bit about working out the equation that we developed
also for cases of other particles ok. So, we looked at a simple case of spherical particles,
but; however, you know you can also develop similar expressions if you are you know
thinking about settling of let see cylinder for example ok
If I have a cylinder which is basically falling with this configuration you know under the
influence of gravity, then you know your ap which is a projected area that is going to be
the a of a rectangle right your length times the diameter. So, you should have to you know
take a appropriate ap right everyone agrees right. So, if I have a cylinder which is rotating
ok so, which is falling.
So, the area that is perpendicular to its flow it is going to be the rectangle right that is you
know your length times diameter; however, if it is rotate you know if its basically
following with this axis in the direction of the external force, your again consideration is
going to be same as the sphere right pi r square it basically sees a circle right. So, should
be able to develop these equations for particles of other arbitrary shapes and they should
be able to work out the settling velocity expression.
Yes it will right because the plot that we showed that I showed for CD was in a Reynolds
number, it was written there is only for a spherical particle case ok. However, people have
done experiments on you know particle there are rough irregular you know again discs
like particle cylinders and stuff like that you know.
Also you would have to worry about all of that as well. So, even if you say that you know
you have to think about you know the area that is in contact with the fluid right. If you say
that that is going to be if its 4 pi r square right its going to be 4 times larger if you want to
do that way. I do not have an exact reason as to why it is done like that maybe I will just
think about it maybe you know answer to the next class.
361
So, the definition of ap is the area that the fluid a plane that is perpendicular to the motion
of the fluid right. So, if the liquid is flowing in a particular direction or a particle move
right. So, you have to take an you know plane which is perpendicular in motion and the
area that that plane sees is what it what that is how it is defined.
But I will just look up and let you know if the why it is in a particularly way ok. So, I will
stop here today and then tomorrow we will try and look at tomorrow we are going to have
a tutorial based on this whatever you discussed so far. So, is going to be in the in MSB
356.
362
Fluid Mechanics
Lecture - 33
Settling velocity - Stoke's regime and Newton's regime
So in the last class we were talking about 1 Dimensional Motion of particle through a fluid,
right ok. So, what we did was we kind of came up with a governing equation that basically
describes you know when a particle is moving in a fluid, but that is in 1 dimension. Such
a case I mean as we were discussing it could be a simple you know, if you take a case
where if you have a column of liquid say that the column the liquid in the column is
stagnant and if you have a particle that is moving under the influence of an external force.
So, we basically set up a working equation, right. And that working equation would look
something like this right a times e into ρ p minus ρ divided by ρ p right is it ρ p or ρ p right
minus C D into A P into ρ v square by 2 times m was equal to dv by dt right that is the
working equation right everybody knows how to get this right.
𝜌𝑃 − 𝜌 𝐶𝐷 𝜌𝑣 2 𝑑𝑣
𝑎𝑒 [ ]−( ) 𝐴𝑃 =
363
Now, as I was saying this is, this equation is a general equation which is valid for particle
of any shape, right and for any fluid particle combinations, right. Now if you want to
simplify this further before we do that I said that you know, as soon as a the particle has
dropped into the column what will happen is, you will start with the particle being
addressed now once you drop it, it is going to accelerate right and then this acceleration is
going to die down soon.
Because of the fact that there is a drag that is acting on the particle and ultimately the
particle will start moving with a constant velocity, right ok. So, at some point in time your
acceleration is only 0, ok, that is because the particle is going to reach a constant velocity.
And you can get that constant velocity as I mentioned the previous class it is denoted by u
t or the terminal velocity people also call it as terminal settling velocity and that we
obtained it as some a times e into ρ p minus ρ divided by ρ p into 2 times m right divided
by C D A P into ρ under square root ok, right that is what we had developed.
2𝑎𝑒 (𝜌𝑃 − 𝜌)𝑚

𝑢𝑡 = √
𝐴𝑃 𝜌𝑃 𝐶𝐷 𝜌
𝑎𝑒 = 𝑔 (𝐺𝑟𝑎𝑣𝑖𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑓𝑜𝑟𝑐𝑒)
𝑎𝑒 = 𝑟𝜔2 (𝐶𝑒𝑛𝑡𝑟𝑖𝑓𝑢𝑔𝑎𝑙 𝑓𝑜𝑟𝑐𝑒)
Now, if you want to simplify this further, I was mentioning that you know you would have
to. So, of course, we talked about two different cases right; one I can say that a e is the
acceleration because of the external force it could be working with any external force that
you wish, ok. But if you take a simple case of external force being gravity we said you a e
is going to be g which is the acceleration due to gravity. And of course, if you take your
acceleration if your external forces like say centrifugal force, ok. We are going to replace
that a e with r omega square right.
So, now if you want to simplify this further we said that you know we should know what
is the C D, right if you look at the equation here right ρ p is a you know the property the
particle ρ is a fluid A P is. So, there are certain parameters which depend on the fluid and
there are certain parameter that depend on the fluid that you are considering, right. Now if
you want to simplify this further we said that you would have to go back and look up things
like this, ok.
364
This is a an example where you know, what is plotted is a the drag coefficient C D on the
y axis versus Reynolds number on the x axis, right. If you look at this plot there are kind
of two limiting cases, ok; one is in the low Reynolds number regime, ok. When the
Reynolds number is less than 1, your C D versus Reynolds number can be represented by
an empirical equation which is something like C D is equal to 24 divided by the Reynolds
number, ok. I can simplify the expression that we just wrote for the Stokes law settling
regime.
And if you look at the high Reynolds number regime when your Reynolds number is more
than 1000 up to about 200000, your drag coefficient more or less remains constant, right
typically about 0.44. Therefore, what I can do is, I can actually substitute for these things
in the working equation. So, what I am going to do is simplifying this, ok.
365
So, let us; so therefore, your u t was if I take a e to be g right. So, it was 2 times m times g
into ρ p minus ρ divided by C D A P into ρ under square root, right that was the did I miss
something, I have ρ p as well, right. So, now, what I can do is you have m here right and
you have ρ p here I can replace that with V P which is the volume of the particle right and
I also have C D which is sorry A P which is as I said it is a projected area which is pi D p
square by 4 and C D as we have been saying this is C D is 24 divided by particle Reynolds
number or Rep or N Rep.
Therefore that is 24 divided by D p U t ρ divided by mu right where mu is the it is cos to

the fluid. So, if you put in this, we can do that quickly. So, therefore, your u t is going to
be 2 times instead of m divided by ρ p, I am going to put it as phi D p cube by 6 right that
is your volume, right multiply by half g ρ p minus ρ divided by I have C d here. So, I am
going to write it as, ok.
Let us do for A P first, A p is going to be I am going to have 4 in the numerator it is pi D

p squared that is for A P, right. And I also have ρ here. Now for Reynolds number it is
going to be I am going to multiply by mu divided by 24 times D p u t into ρ, right that is
your mistakes. So, your C D is 24 divided by Reynolds number right. So, that is going to
be D p by ρ mu that is that is, right. So, therefore, I can cancel these two here and I can
cancel, I think I have done a mistake, right. This is going to be D p here in to u t here in to
ρ right. So, then you have D p square here and there is D p power 4. So, this going to be
366
D p square will come in the numerator. So, you have 4 here and there is going to be so, 6,
right.
Therefore there is 1 here 3 here ok. So, therefore, basically if you work it out and of course,
your ρ and ρ is going to get canceled. Therefore, if you work it out it turns out the your u
t is going to be g into D p square into ρ p minus ρ divided by 18 mu because 3 times 6 is
18 and you have mu in the you know denominator and your numerator is going to be g D
p square into ρ p minus ρ you can work it work this out, ok. This is the working equation
if you have like say a particle which is settling in the Stokes regime ok.
So, Stokes regime is basically if you have your particle Reynolds number is less than 1.
Your u t is going to be g into D p square into ρ p minus ρ divided by 18 mu, ok.
𝑔𝐷𝑃2 (𝜌𝑃 − 𝜌)
𝑢𝑡 = ; 𝑁𝑅𝑒,𝑃 < 1
18𝜇
Now, similarly if you work out for Reynolds number, that is greater than 1000, but less
than 200,000, ok. In that case your C D is going to be constant and that is 0.44, ok. If you
substitute for C D in the expression for u t and then if you simplify it further what you
would do is you would get u t as 1.75 into square root of g into ρ p minus ρ into D p divided
by ρ, ok. You can work this out at home ok.
367
𝑔(𝜌𝑃 − 𝜌)𝐷𝑃
𝑢𝑡 = 1.75√ ; 1000 < 𝑁𝑅𝑒,𝑃 < 200,000
𝜌
Therefore, these are the two limiting cases, for the settling of particles in a fluid, ok. That
you know expressions for the terminal velocities which are applicable for the limiting
cases one for when the Reynolds number is less than 1 is what is called as a Stokes settling
regime. Other for cases where the Reynolds number is in between 1000 and 200,000 is
something called as a Newton’s settling regime, ok. Now so, of course, these are applicable
for a spherical particle, right, because that is what we considered in this case pi D p square
by 4 being substituted for A p which is the projection area which is only true for spherical
particle right. So, therefore, if you have other objects of other sizes, so, it turns out that
just.
So, if you want to generalize for you know particles of any shape, what you should do is
you should have some correlations like this ok. What you are seeing in the plot is again C
D versus Reynolds number, but if you look it up there are different plots, ok. There is one
plot for cylinders sorry for a spherical particle is a continuous line, right. The continuous
line is for the spherical particles, which we just saw right the two limiting cases of C D
being 0.44 or C D being 24 by Reynolds number and the other two are for particles of
different shapes, ok. One is for disks other one is for cylinders and these have been
368
generated by taking particles of particular dimension and particular shape and they have
been held in some particular condition.
For example in this case axis of the cylinder and face of the disc are perpendicular to the
flow direction ok. If you have a liquid that is flowing, the axis is in a direction
perpendicular to the flow that is like this, ok. Therefore, when you want to work out the
equation for settling velocities for different shaped particles you would have to worry
about the orientation of the particle as well, ok. Whether the particle is oriented in the
direction or in the direction perpendicular to the flow.
When the particle is oriented perpendicular is something was a bluff body is that is a
terminology people use, in the literature. So, therefore, I would have to take care of you
know such issues when you want to simplify these equations further for calculating settling
velocities for different shape objects. Now, if you are given a problem that, I say that a
problem is given to you and you want and they want you to calculate what is the settling
velocity of the particles okay.
Now, how would you go about doing this say that, you know there is an problem that is
given I can say that there is a particle of 100 micrometer diameter. It is settling in a liquid
column, I say that the column is filled with water. I will give the density of the liquid you
know water and then also viscosity of fluid. And if you would ask to calculate what is the
settling velocity of the particle how would you go about doing the calculations.
369
So, one crude way of doing this would be you calculate your u t for the Stokes regime, get
what is your u t and then from that you basically calculate what is the Reynolds number
after that I have done the settling velocity calculations. You assume that you know u t is
say g D p square into ρ p minus ρ divided by 18 mu, you calculate the settling velocity
then you get u t then you back calculate what is your Reynolds number, ok. If it is D u t
by mu, right if the Reynolds number falls less than 1, then I say that you know whatever I
have assumed that the particle is settling in the Stokes regime is correct ok.
Therefore, my assumption is right, ok. But that need not be the case all the time, right.
Now you can end up with a Reynolds number which could be more than 1, it could be
more than 1000. So, therefore, you kind of come up with a kind of a dilemma, as to you
know whether I should be using the Stokes regime or the settling regime or the Newton’s
regime ok. So, therefore, what you can do is you can come up with a simple non
dimensional number, ok. Which would, which can help us in terms of identifying the
settling regime without going into this guesswork?
I said that the guess work you assume that you know the particle is settling in the Stokes
regime. You calculate u t back calculate your Reynolds number see whether it is less than
1 or you know look at the typical range or typical number that you get out of the
calculation. And then if that is not the case then you go back and then assume the Newton’s
regime. Then again calculate u t, calculate again Reynolds number right. So, instead of
doing that what you can do is you can rearrange the expression that you have for your
settling velocities, right.
370
I know that u t is g D p square into ρ p minus ρ divided by 18 mu for the Stokes regime. I
know that you know my Reynolds number, for the upper limit of the Reynolds number for
Stokes regime is 1, right anything less than 1 is a. So, now, what I can do is I can substitute
D p into ρ into u t divided by mu I can substitute for.
U t from the Stokes regime, so what do you get sorry, that is going to be your g D p square
into ρ p minus ρ divided by 18 mu, right. So, what I can write this as. So, this is going to
be I am going to write it as cube and I am going to write this as 1 over 18, right.
𝐷𝑃 𝜌𝑢𝑡 1 𝜌𝑔𝐷𝑃3 (𝜌𝑃 − 𝜌)

=
𝜇 18 𝜇
Now let us look at the exact similar formulation for So, we know that your Reynolds
number should be more than 1000 for the Newton settling regime. So, similarly I can write
the Reynolds number as D p into ρ by mu into 1.75 right, square root of g into ρ p minus
ρ into D p divided by ρ, right that is your expression further. So, now, what I can do is I
can take the D p inside, right. So, I can write it as D p cube, right. I am going to take is
that and then I have ρ here, again I am going to take ρ square inside I am going to write as
ρ square, now of course, there is going to be mu square as well right, ok.
371
So, now I am going to cancel a few things, right. So, that is going to be there is one mu
goes, right and what I has to have, but I forgotten that should be ok. So, did I do a mistake
here, no right 18 there is going to be mu square here, right.
Right ok. So, look at the term that is in the there is g here right, there is D p cube here,
there is ρ p minus ρ did I cancel one of the, what is the expression for the stoke settling, I
made a mistake, right there is no. So, what we did is right, D p into ρ was in the numerator
that is going be so, can you help me out. So, I have 1.75, let me just write it. So, your, it is
1.75 into square root of g into ρ p minus ρ into D p divided by ρ, right that is my Reynolds
number.
So, here that is going to be ρ square here now that is going to be ρ here. So, therefore, it is
seems. So, it is right. So, I think that is fine. So, I have a row here, right. I have a row here,
g here write D p cube ρ p minus ρ mu square is it, ok. Therefore, I can actually define a
factor K, ok. I can define a factor K as g D p cube into ρ p minus ρ into ρ here divided by
mu square to the power of one third, right.
So, therefore, this is going to be 1 over 18 into k cube. And here it is going to be 1.75,
again to k cube to the power of one and half is it.
So, therefore, from the just by rearranging the two expressions that; we had one for the
Stoke settling regime another one for the Newton’s settling regime. What we have been
able to do is, we have been able to get 1 over 18 into k cube is equal to 1 is the case for
Stokes settling regime ok. And 1.75 into k to the power of 1.5 is equal to 1000 for the
Newton’s settling regime, ok.
1 3
𝐾 = 1 → 𝑆𝑡𝑜𝑘𝑒𝑠 𝑠𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑟𝑒𝑔𝑖𝑚𝑒
18
1.75 𝐾 1.5 = 1000 → 𝑁𝑒𝑤𝑡𝑜𝑛′ 𝑠 𝑠𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑟𝑒𝑔𝑖𝑚𝑒
1/3
𝑔𝐷𝑃3 (𝜌𝑃 − 𝜌)𝜌
𝑊ℎ𝑒𝑟𝑒, 𝐾 = [ ]
𝜇2
If you work it out in this case your k comes to be something like 2.6 ok. If I you know k
cube is 18 therefore, k is going to be cube root of 18. So, that is going to be 2.6, ok. And
similarly the k for the Newton’s settling regime comes out to be something like 68.9, ok.
372
If I substitute the lower limit of the Reynolds number for the Newton’s regime and if I
substitute to higher limit of the Reynolds number which is 200000 your k comes out to be
something like 2,360. Therefore, the idea is this ok.
So, what you do is you calculate the k which is basically defined as g D p cube into ρ p
minus ρ into ρ divided by mu to the power of one third you calculate that. If the value of
the k that you obtain if it is less than 2.6 then you use the settling velocity for the Stokes
regime and calculate your terminal velocities.
However, if the k factor that you are getting if it is between 68.9 and 2,360, that is the
range of Reynolds number over which the settling occurs in the turbulent condition, that
is the Newton settling regime, ok. In such case you would have to use the appropriate
expression for the terminal velocity for the, you know particle in the Newton’s regime ok.
So, therefore, this formalism, in terms of calculating k, as a criteria for deciding whether I
should go for a Stokes or the Newton’s settling regime it is useful instead of doing a
guesswork of assuming your particular settling regime and a back calculating your
Reynolds number.
Any questions, do you have any questions with this? It is just a manipulation, right. It is
just a rearranging of you know the expression for each of these you know u t’s and then
you know equating that to the limiting value of the Reynolds number for respective
conditions, that is all we done yeah ok. So, now, we will talk a little bit about, before we
talk about applications of these things.
373
So, now what do you see is a plot, of a variation of settling velocity with size, on the y axis
you have terminal velocity on the x axis you have equivalent spherical diameter. And there
are different curves right 1 2 you know they are numbered from 1 to 9, ok. And they have
taken different fluids you know for example, 3 is an aniline, 7 is nitrobenzene, ok. So,
basically what we are looking at is, a case of fall of liquid drops in water, ok.
So, far when we looked at settling, we only talked about solid particles, right we talked
about spherical particles rigid you know hard particles. Now, if you want to plot, terminal
velocities is a function of size of the particle, ok. It would always be a monotonically
increasing function right because in the case of Stokes settling your u t goes as D p square,
right. And in the case of Newton’s settling regime it goes as D p to the power of half right.
So, therefore, in both the cases as I increase the size my settling velocity should always go
on increasing right; that means, it is a monotonically increasing function, ok. Now, if you
look at this plot can you see some difference; the first difference which is very apparent is
that it looks like the settling velocity increases, in analogy with what is known for smaller
dimension for example, maybe in the range from 0.0 to 0.6 in that centimeter size range,
the thermal velocity increases with size.
However, it attains a maxima and then it basically drops back, right ok. So, any thoughts
is to why that could be the case. So, everybody understand the plot right, it is a plot of
terminal velocities is a function of equivalent spherical diameter. So, why does equivalent
374
spherical why is it not the diameter of the drop, why is it a equivalent spherical diameter?
The obvious answer is going to be because the drop right, the drops can also change the
shape.
So, therefore, when one is dealing with drops in a fluid I would have to worry about shape
changes ok; that means, depending upon. So, the reason why it would go to a maximum
and then drop off because one naive answer could be; that in the initial for the smaller size
regime the particle, the droplets will remain spherical ok. However, when you make the
droplets bigger and bigger at some point they are going to flatten, right and then your drag
is going to be more, more drag means it is going to settle slower, right. So, anyway we can
think of we will argue a little bit of along these lines in a minute.
This again, some another some data again for, so I do not know if you notice this, right.
This is a paper from IIT Kharagpur which is published in 1950s, ok. So, people who are
worrying about settling of you know droplets and you know other stuff you know then it
is a right, because it is and again the settling of drops in a fluid is also relevant to one of
the common thing that you guys see. Any example raindrops, right ok. When the rain falls
right it is going to be these are going to be water droplets as they approach earth they are
going to again flatten in the direction you know of the gravity, right. So, in depending upon
the size of the droplets that are going to be generated, ok
375
If that size the droplets are very small, then you can assume the drops to be spherical in
shape they would you know come like a hard sphere; however, if you have a, if there is a
larger raindrop the moment it exceeds a particular size, if I were to go by this plot for like
say 4 which is some carbon disulfide. There is peak right I can draw a vertical line and I
can actually get what is the limiting size for which the droplet behaves like a sphere. But
beyond that line you know where you see the drop in the terminal velocity that is when the
shape changes would have occurs, ok.
Again there are more data this again from some other people, again the characteristic peak
in the terminal velocity followed by a you know a decrease is evident here as well. Any
thoughts as to why the other nice things to notice is you know, the peak where it the
appearance of the peak right is seems to be shifting towards smaller sizes right. If I look
at the bottom most plot which is for aniline, right or the top most plot which is for
tetrabromoethane it turns out there you know the peak is kind of shifted right to the left
side ok. Any thoughts why it could be? That is because there are different density, right.
If you look at these fluids for example, aniline has a density of 1.02, which is very much
close to that of water right.
However, if you look at tetrabromoethane the density is 2.97 much denser, ok. Again that
is also evident from the slope, right. If the slope for at least in the initial regime where
there is a linear increase, right the slope is larger for 2.97. Because the density difference
376
is more we know that, they should be settling much faster right, compared to the case
where the density difference is very little for the case of aniline, right.
This is a plot of again drag coefficient as a function of Reynolds number. It appears that
for the low Reynolds number cases for example, Reynolds number from a 10 to about 400,
roughly ok. All the data for the droplet seems to be in line with what is expected for solid
spheres, but; however, if you go for larger Reynolds number; larger Reynolds number is
because larger dimensions larger size of the droplets.
Therefore, you can see a significant change that you see because that is evident from the
upturn of this C D versus Reynolds number plot, right. Again this is another indication of
the fact that the droplets have changed their shape and one has to worry about such things
you know when you look at.
377
These are some images from again different literature the extreme the plot on the left.
Again extreme left is that is where the droplet continues to remain in the spherical shape
during it is settling ok; however, if we consider a larger droplet it turns out that you know
that the drop start becoming elongated right.
And this elliptical the deformation of the droplet leads to droplets being in different shapes
and of course, because of that your drag forces are going to be very different than what
you would see for the spherical particle case. And that is therefore, you would you would
expect that you know the particles would slow down because of the increase in the drag
force ok.
378
So, in general the main reasons for the difference between the motion of the liquid drops
and that of the rigid spheres is because of the deformation of drops ok, which becomes
evident when you go for larger sized droplets ok, number one. And this deformation can
lead to either shape changes or shape oscillations ok. If you look at the movies ok, you
know how these droplets settle.
Now you will see that you know the drops start going from one shape to the other ok, there
is going to be drop shape fluctuations which you would have to worry about ok. And
whenever you have a smooth or a you know a liquid like surface; the flow on the surface
as well as flow within the droplets are going to be different ok. In the case of solid sphere
spheres I do not have to worry about flow within the particle, right.
However, in the case of you know liquid droplets they are going to be some internal
circulations which can also lead to a change in the way that droplet would behave in a fluid
ok. So therefore, so when you are working with spherical particles; spherical particles and
rigid particles of well-defined shape and dimension, things are kind of fairly well laid out
ok; that means, you know there as C D versus Reynolds number plot you should know
little bit about settling regime you know the correlation between C D and Reynolds number
all of that is this kind of well laid out you know.
379
You can use those things to understand the settling behave behavior of solid particles;
however, if you go to liquid drops as a bubbles, you will have to worry about some of these
considerations.
380
Fluid Mechanics
Lecture - 34
Applications of settling - I
So, we are going to look at some applications of what we have learned so far. So, though
we termed you know the last few classes as we are looking at motion of particles in a fluid
ok, and that naturally occurs when you are talking about settling of particles in a fluid
right.
So, we looked at something called as a falling ball viscometer right ok, in which you
basically had a container right, a capillary, and you take a particle of known properties ok,
that means, known diameter and known density ok.
And if you have a way of measuring what is the density the fluid that you are working with
right, and then you know if you have a way of calculating or measuring what is the terminal
velocity that is the maximum possible velocity that the particle attains in a given external
fluid, and if you are working in the in you know under the influence of gravity right if
right. So, therefore, what I can do I know there you know this u t, I can really you know it
goes as g D p square into rho p minus rho divided by 18 mu right that is for your stokes
regime ok.
381
And if you look at the formula for settling velocities only in the Stokes regime there is a
dependence of the you know settling velocity on the viscosity right. If you look at the
Newton’s regime, your settling velocity does not depend on the viscosity of the fluid ok.
Therefore, what they do you know, they use particles of appropriate dimension ok, so that
the settling falls in the, you know in the stokes settling regime that is your Reynolds
number is less than 1 ok. As long as you ensure that I can basically exploit the Stokes
settling principle. And if I am working with particles of well in a defined sizes and densities
and fluids you know known you know density, I can actually calculate what is the viscosity
right that is one of the application that we have looked at already so far.
Separation of so settling principles are kind of exploited a lot in particle separation ok. If
you have like say particles of one size or multiple size ok, one type or multiple types ok,
you could have a case where I have particles of same type of a given density, but different
sizes. Or I could have a case where I have a mixture of particles in a particle 1, 2, 3 of
different materials; it could be like say glass beads plus polymer beads for example, ok.
So, if you have particles of different sizes and different properties ok, so you can actually
exploit these things. We will talk a little bit about some of these things in the remainder of
the class.
So, before I go further looking at separation of particles, I just want to give you a little bit
of heads up with some places where you know sedimentation is kind of really useful. I
382
know how many of you have heard of the ESR which is called as a erythrocyte
sedimentation rate. Have you heard of this term? No? Yes? Somebody says yes. Where
you have heard?
It is a test that is done when somebody goes for a diagnosis right. What they do is, they
take a small amount of blood sample ok, and one of the tests that they do is this
sedimentation measurement ok. The experiment basically involves taking blood in a small
capillary, you just let it sediment ok. And looking at the sedimentation rate, you can
actually tell something about whether the blood belongs to a male or a female, somebody
who is young or old ok. Plus it also is one of the nice ways of measuring you know or
finding out if there is something of a inflammation whenever somebody has some disease
there is going to be inflammation right. So, basically this sedimentation rate measurements
are simpler measurements for kind of as a first tool for diagnosis ok. And so in that context
you know people do.
So, I again picked up some you know image from Wikipedia wherein these are basically
capillaries which are filled with blood and made to sediment you know for some period of
time. And you can clearly see that you know there is a interface between the clear capillary
and the you know the brown or reddish mark you know from finding out you know what
is that interface I can say something about sedimentation right.
If I had a you know homogeneous so maybe just to tell you, so initially the blood was say
in this case the blood was filled up to this level ok. After you know some time, you can
clearly see in all the whatever particles that you had in blood RBC is another stuff that has
you know now the mark is here. Therefore, this is a clear plasma there is no particles there
ok, and all the particles are now kind of concentrated here right.
So, basically by measuring this interface between the clear fluid and you know the samples
where there are blood particles, I can actually measure the sedimentation rate ok, this is
what is typically done. And depending upon as I said disease conditions and another type
of you know a person you are dealing with will have different values ok. So, this is one
applications you know where he actually can actually use exactly the same equations that
you used ok. You if you know the dimensions of a blood, if you know the viscosity of the
blood, you can actually also calculate what is the settling rate.
383
I am going to talk about an application that kind of people have. So, in the class so far I
only talked about mostly sedimentation and the influence of gravity right. Of course, we
did mention that you know you can use external force in the form of centrifugal forces to
do sedimentation as well right.
So, there is a nice concept called paper-fuge ok. Any thoughts what it could be? Ok that is
actually it comes from paper that appeared in 2017 in a journal called Nature Biomedical
Engineering. It is published by a group in Stanford which is headed by Manu Prakash ok.
And there is one person who is Saadbhamla who is in involved with this book, is he is in
IITM undergrad. He finished his B.Tech few years ago, finished his Ph.D Stanford ok.
So, the title of the paper is Hand-powered ultralow-cost paper centrifuge. Ok the reason
why they call it as paper-fuge, is because it is a centrifuge and we can actually make it
make it out of paper right. So, I just have a video from ok, let me just try and show you a
video of this.
384
I am sure all of you have played with this at some point.
385
Ok. So, there is a lot of you know study before this kind of let me just I have some you
know slides from the paper. So, an interesting aspect of this is if you noticed, so when you
are pulling it right, there is a there is a linear motion that you are applying that kind of kind
of translations or rotation right.
And what they did is they looked at so, basically what is happening when you do this, there
is a winding and unwinding right. And you know at some point if you clearly see that you
know the spinning disc actually stops right. You know when there is a complete winding
386
right. So, what they have done is they have taken different designs ok, they did experiments
with you know a lot of different designs of these discs ok. And they used high speed
camera to measure what is the rotation speed. And as it was mentioned in the video, they
can actually reach speeds which are as high as a 1000 you know a 125000 rpm ok, and that
can actually translate into a gravitational force of the order 25000 to 30000 g that which is
really really huge ok.
And you have a plot here that is measuring phi max which is the maximum rotation speed
as a function of disk diameter right. It turns out that you know if you make it smaller, you
rotate it faster right if you really want to go for; go to a very high gs right, therefore, you
would be using a smaller diameter disk. And if you go, if the; this diameter becomes larger,
now your speed is going to go down ok. So, depending upon you know what kind of speeds
you want to generate you can actually you know exploit the fact that you know the
maximum phi max varies with the disks speed.
And this is again the device that was mentioned what you have here is that is a kind of a
hollow tube in which you can put a capillary with any sample that you have right, and all
you have to do is close it and then spin it ok.
387
This is a demonstration of what basically happens that is the fraction of RBCs in the blood
that the y-axis go from 0 to 1 right, and you know your x-axis is the time ok. And of course,
as the clear fluid you know kind of starts you know appearing, I can also measure what is
the plasma volume right. When, the fraction of the RBCs in the blood is like 1 here, your
plasma volume is 0 right because it is a homogeneous fluid here ok.
At some point you know somewhere here you start seeing some small volume of you know
clear fluid that is when a in a small volume of the plasma appears here. So, basically it is
a simple demonstration of the you know the how the separation happens in terms of
monitoring what is the fraction of RBCs in the blood and the fraction in and the plasma
volume. And these are the microscopy images taken at different location, and the stop
region where it is only plasma, there is absolutely no cells ok, that means, all the cells there
were they have kind of settle, and that is the interface your between the clear fluid and you
know the settle you know RBCs and that is your image where you know you have lot of
RBCs ok.
So, it is a very simple tool which is kind of, you know kind of developed which is again
exploits the fact that you know you can basically use very high external force right.
External force of very high magnitude to really separate stuff ok, any questions? Yeah.
So, do you understand the plots? Ok, the plots is basically you have a when you have the
blood particles you know RBC is kind of homogeneously dispersed homogeneously
388
dispersed in the entire fluid you know then RBC fraction is 1 ok. The movement you know
a small amount of you can also think about this in terms of the volume ratio as well right.
I have the entire volume of the fluid ok, and of course, I do not have any fluid you know
without the particle. So, in that case my you know the reference is 1 ok. Therefore, in the
next stage what I see is I have a very small fraction of the clear fluids are form that volume
divided by the total volume of the blood that I had taken that this gives you some fraction
as is it basically a measure of you know how much of sedimentation has occurred right.
Now, on the on the right what you have is a basically microscopy images ok. The fact that
you know I clearly see a visually there is a difference for sure between when I started, and
you know and at some instant during your, you know a separation. And other way of
confirming this would be just to take out some small sample from different locations and
you do microscopy ok.
And if I do microscopy from the after small sample or I take from the top, I do not see
anything right ok. However, if I take some sample from the bottom ok and if I do
microscopy it is not a very high magnification image here. But what you see is can you
see some small circle like things here right, these are basically RBCs ok, red blood cells
which have a typical dimension of in about 8 micrometer ok. The scale ball here is about
10 micrometer ok. So, therefore, you will see typically I mean the object that you see are
sub 10 micrometer dimension ok.
This is just to say that you know the visual observation that you see here is you know it is
kind of confirmed by doing some microscopy measurement that is all, yeah.
Oh, it is a basically I mean no this is a see what they have done also you are talking about
see this you see this large error bars right ok. These are actually done over several
measurements ok. There could be some fluctuations because of you know conditions as
well you know the person you know who is being whose blood has being drawn, there
could be some other fluctuation. So, basically you know but the point is you know that
you basically these some kind of a plateau right, if you wait for sufficiently long time you
see that you know absolutely that that means, at that point all these RBCs are kind of
packed there is no more space for them to even you know sediment further you know if
you even if you continue to do prolonged you know centrifugation so on yeah ok.
389
So, I have kind of talked about elutriation in the last class, maybe you know I was trying
to draw picture myself, so it was probably it was not clear. So, elutriation is again a
technique that is typically used for separation of particles ok. So, if you have a mixture of
particles like say just to give an example let us say I have a mixture which is like say 1
micrometer, 10 micrometer and 100 micrometer particle ok.
Now, elutriation is actually it is a reverse of sedimentation ok. Now, if you look at the
schematic here, what you have is there is a there is a feed stream ok, that feed stream
basically carries the mixture of particles that you know are that want to be separated ok.
So, I have a mixture of different sizes, you could think about maybe feed stream consisting
of say 1 micron, 10 micron you know 100 micrometer particle that enters the feed. And at
the bottom you have a fluid stream ok, and with this fluid stream, what I can do is I can
basically maintain a particular flow rate right. I can; I have a control over which what is
the flow rate with which I can send the fluid stream in right. Because I have control over
the fluid stream you know volumetric flow rate, I also have control over the velocity right.
Now, what happens you know these particles are going to come down because of gravity
ok. Now, what I can do that I can actually manipulate the fluid stream velocity such that it
to maintain a particular value such that. For example, let us say that 1 micrometer, I have
a particle, and 100, 10 micrometer particle. I can calculate their settling velocities right,
you know under gravity what is the settling velocity that they going to have.
390
Now, if I have information about that, if I want to separate out ok, if I want to separate out
100 micrometer particles from this mixture. What I do is, I maintain a fluid stream velocity
such that the velocity of the fluid stream is larger than the settling velocity of 1 micrometer
particle and the 10 micrometer particle such that when the fluid stream goes up ok, the
smaller particles basically come out ok, 1 and 10 micrometer particles, they come out of
the you know the container and the larger ones basically sediment out ok.
Therefore, what you do is I can actually manipulate the fluid stream velocity, and I can
actually by, you know by design I can actually find out or I can I can say hey I only want
1 micrometer to come out ok, I can do that. I want a mixture of 1 and 10 to come out, so
basically that is how the typically separation is done in the case of elutriation ok. So, what
you do is the particles with the terminal velocities less than the upward velocity of the fluid
are basically carried out you know carried along with the fluid ok, that is what happens in
the elutriation ok. And this typically is used if you have you know particles of different
sizes, but of same density present in a given mixture ok.
391
Fluid Mechanics
Lecture - 35
Applications of settling - II
Now, there are techniques where you kind of exploit a concept called differential settling
ok. And this concept of differential settling is typically used when you have mixture of
particles of different sizes and different densities ok. So, now, in a method called sink and
float method, what you do is somebody gives you a powder mixture and it has say particles
of you know type A and type B. And say the density of the particle is rho A and density
of the particle is rho B ok.
Now, what I can do is, I can put them in a column with a fluid. And if the density of the
fluid that you are using ok, if it is in between the density of the two particle that you have
in the mixture, what will happen is one of the particles would sink and the other would
float right. So, that is why you know hence this name the sink and float method ok. So, in
the sink and float method what you have is a mixture of particles of different densities and
a use of fluid column whose density is in between the density of the two particle that you
are considering, and then you basically separate it out ok, that is what is done in here ok.
392
And there is a concept called equal settling that is what we are going to do in the ok. Now,
so in the differential settling what you do is again you have a mixture of two particles ok,
maybe again you know like say let me just ok. Again you have particles A and B ok, and
the density of the particle you know as it is written here is rho of particle B, sorry particle
A here ok and this is the rho of particle B right. Now, what is done in you know the equal
settling concept is there you know you basically use the fact that there is a difference in
the terminal settling velocities between the particles of different densities ok.
I am just going to. So, what you do is we know that you know this is a the settling velocity
for Stoke’s regime right and this is the settling velocity for again Stoke’s regime, but again
for the particles of type B here right, this is for type A, this is for type B. Now, if I say that
the settling velocities of you know type A and type B are the same ok, what I do is I can
get a ratio of the particle diameters right ok.
What I do is rho tA is equal to rho tB right, the density of the fluid is the same that gets
cancelled right, sorry your viscosity the same, your g gets cancelled. So, basically because
of the fact that this and this is the same, I can actually get an expression for the ratio of the
diameters as the ratio of the density differences right that is for the Stoke’s regime. And
this is for the Newton’s regime right. Any question? It is a simple fact you know what I
have done is, I basically equated the settling velocities right.
393
Now.
How this works you can kind of understand you know from a you know from a diagram
ok, from a schematic. What you have here is this is the terminal velocity ok and this is the
particle size ok. Now, on the y-axis sorry on the x-axis, you know there are these vertical
lines right, these vertical lines here right. So, what you have is, you have a mixture of
particles whose size varies from D p 1 to D p 4 ok. I have two particles of type A and type
B and the size range of the particle in the mixture is the same that goes from D p 1 to D p
4 ok.
Now, there is an horizontal line that is drawn here ok. And this horizontal line and this
horizontal line, it corresponds to the equal settling right, because you know I am just
drawing a. So, this one the plot and that A is for the is how the terminal velocity, after type
A particle varies as a function of size. This is a plot for how the terminal velocity varies as
a function of particle size for B type of particle right.
Now, what can I say about the size range from D p 1 to D p 2 of B type of particle. Just by
looking at this plot, what can I say about the settling velocities of the particle size range
from D p 1 to D p 2 for type B particles? Can I say that the type B type of particle in the
size range from D p 1 to D p 2 settle, you know the slowest right, because any particle of
type B from D p 2 onwards settles faster right and all the particles of type A from D p 1
all the way up to D p 4 settle faster, therefore, the size range of B from D p 1 to D p 2 that
394
it settles slower than any particle of A type plus all the particles of B going from D p 2 to
D p 4 right. Everyone agrees with that?
So, therefore, I can actually what I can do is, I can I have a column ok, which is filled with
a fluid. I introduce; I introduce all the particles you know at the surface of the you know
the fluid column ok. And because I know that D p 1 to D p 2 of type B settles the look
slowest ok, I can wait for the particles to settle for sufficiently long time. So, there you
know all the particles of type B from D p 2 to D p 4 settle and all the particles of type A
from D p 1 to D p 4 settle, therefore, what you have in the fluid would, be what do you
have in the fluid that will be suspended would be only the particles of type B from D p 1
to D p 2 that is because they settle the slowest right, right.
So, what can I say about this size? So, now, this is the if I go vertically that is the maximum
settling velocity of or that is the settling velocity of the largest size B type right, settling
velocity of the largest size B type is this. If I move vertically sorry horizontally here, D p
3 D p 3 of type A, D p 3 of type way type A will have the same settling velocity as D p 4
of type B. Therefore, the size range from D p 3 to D p 4, the size range from D p 3 to D p
4 will settle the fastest right.
So, therefore, if I do experiments for a relatively short period of time ok, I again I have a
column, I introduce the particles of the surface of the fluid ok, if I withdraw the fluid ok,
so if I withdraw the particle from the sediment after an appropriate time, all the particles
of type A from D p 3 to D p 4 would have settled to the bottom. And what you would have
in the fluid would be all the particles of B from D p 1 to D p 4 and all the particles of A
from D p 1 to D p 3 right, they will continue to remain in the fluid ok. So, therefore, what
people do is people exploit this thing called differential settling, basically to separate out
mixture of particles into different sizes ok.
And one thing I did not mention is that in this case the density of the medium should be
less than the that of either of the substances right. In the case of sink and float method, the
density of the fluid is in between the density of the two particles right, that is where that is
why one would sink and the other one would float. In this case because you want both of
the both the particle to settle and that will have happen only if the density of the fluid is
less than the density of the two particle mixtures right.
What is that?
395
Any answer for this question. So, why are we not using sink and float method for this, why
are we think about you know equal settling concept? Why should we look for the density
which is smaller than the density of the two particles that I have in this system? Any
thought?
That is one possible explanation. So, what are you saying is there you know in the sink
and float method, I would have to use particles of a fluid whose density is in between the
two particles right. However, if the density difference between the two particles itself is
not too huge ok, then you know you may have difficulties or challenges in terms of you
know finding out a fluid whose density is between the two ok.
But the more I think it is because of the practical difficulties ok, maybe there is an example
here. So, say that you know I have a two particles, say A there is a something like the
galena which is one of the ores of you know one of the material ok, whose specific gravity
is 7.5 ok, that means, you know the density is 7500 kg per meter cube right. Let us look at
quartz, it has a specific gravity of 2.65 ok, 2650 kg per meter cube is a density.
Now, you look at any fluid let us say water density is 1000 ok. Now, go back and take a
look at you know the list of fluids and the density it will be very difficult to really have a
density which is you know more than 2.65 ok. Of course, there are ways by which I can
actually manipulate the density of the fluid. What people do is I have like say water I
introduce salt ok. Depending upon the density of the salt that I am dissolving if I say you
know like say I take NaCl right sodium chloride ok and I dissolve in water. What will
happen to its density, it will go up right, you know if the density of the salt that I am putting
in if it is larger than if it forms a homogeneous solution, the density will definitely go up
ok. There are ways by which I can manipulate the density of the fluid.
But however for cases where the density of the particles are much larger than you know
like say 1000 kg per meter cube right, so it will be difficult to really find a fluid whose
density is in between the density of the two fluids ok. Therefore, people do exploit this
differential settling where the density of the fluid that you are using is less than the density
of the two particles of you are considering the mixture ok.
396
So, now, there is something called as a sharpness of separation ok. What I mean by that is
how given a mixture of particles of different sizes and different you know densities, how
cleanly can I separate out the particles into pure fractions ok. What I mean by that is in the
previous example, say you know I have a particle size going from D p 1 to D p 4 ok.
Now, I take this mixture this has you know type A, type B, all the sizes from D p 1 to D p
4, if I am able to separate them into pure A and pure B ok, then my sharpness of separation
is kind of one right. You know I am able to separate them into two pure fractions of
different you know population of particle right.
Now, it turns out that you know the sharpness of separation can be improved if the density
of the medium that I am using is increased ok. So, what I mean by that is in the previous
example see you had a specific gravity 2 and 2.65 ok. If I use a fluid of you know say I am
just going to mostly make it as that is your density of the particle A and particle B. If the
density of the fluid that I am using is say 1000, then you will get a certain fraction like say
maybe D p 1 to say D p 2 as a pure fraction.
Now, if I increase the density of the fluid ok, just to give numbers, let us say you are
separating particles from 1 micrometer to say 10 micrometer of B into pure fraction ok, if
I am using a density of that fluid which is 1000. If I make it 1500, it turns out that I can
actually increase the size range ok, I can maybe go from 1 micrometer to say 15
micrometer ok, because I am able to get a larger size range into a pure fraction ok. It is a
397
equivalent of saying you know I basically increase the there is a sharp, the sharpness of
separation is actually increased ok.
So, one of the ways of you know increasing the sharpness of separation is by using a
medium whose density is by increasing the density of the medium in which you know
which you are using. The other way of doing this would be that you know if I have a way
of choosing a size range which is closer ok, what I mean by that is let us take this example
ok. I have explained that you know I have by considering the fact that you know D p 1 to
D p 4 is a size range that exists between you know that is for the particle A and B, I talked
about the fact that you know this settles the you know slowest and this settles the fastest
right.
For simplicity, let us say that you know I only have size range from D p 3 to D p 4 ok.
What I have done this, now I have done a closer spacing right, earlier I had sizing going
from D p 1 to D p 4. Now, I have a mixture whose size range goes only from D p 3 to D p
4. If that is the case, I know that the size range D p 3 to D p 4 of A settles the fastest, that
means, I can actually separate out all the particles of A into pure fraction. And the fact that
you know thus again the same sizing D p 3 to D p 4 of B settles the slowest I can also
separate that into you know pure fraction right.
So, therefore, there are two ways of increasing the sharpness of separation one is you use
a fluid of a higher density or by increasing the density of the fluid. Or other one is if you
consider closer sizing of the feed instead of taking a size range going from D p 1 to D p 4,
you reduce the size range ok, either going from D p 2 to D p 4, or D p 3 to D p 4 if you do
reduce that you know if you have a closer spacing in the size you know that again give
you a another way of you know complete separation of the particle.
Ok.
Correct.
So, why would you say that or it will have some of A.
That you would have to work it out ok. Let me make another point which I did not make
it. So, in this differential settling method ok, you can get a pure fraction of A and pure
fraction of B ok and a mixture of A and B right that is what is a typical case. If you when
398
I talked about you know the sizing going from D p 1 to D p 4, we saw that you know D p
1 to D p 2 of B was coming as pure fraction and D p 3 to D p 4 of A was you know coming
as pure fraction and in between it was a mixed fraction ok.
However, if you took this example of closer spacing D p 3 to D p 4 both A and B was
coming as pure fraction right. Now, when you say that you know so the point that you
want to make you know you are saying that, so it turns out that the when you use a density,
so say there you know my size range is fixed ok, say that you know I have a size from 1
micrometer to 1000 micrometer that is the sizing that I am considering ok. So, both the
pure fraction that you get of A or B are kind of slightly expanded if you use ok.
We are going to have a you know tutorial question, you know in the next class, wherein
you are going to look at this example, where we will see that you know if you use a density
like say 1000 kg per meter cube, you will look at you know what are the sizing that comes
up your fraction ok. Now, if you work with the fluid of higher density, you will see that
you know there is an expansion or the you get an expanded size range both in the A type
as well as B type.
And of course, mix fraction anyway is going to be there for both the fractions, you know
unless you know you are actually at a particular you know kind of a fraction where you
know either you are working with a narrow size range or a part you know particle of a
particular fluid of a particular density ok. You will see that maybe in the next class.
So, the mixed fraction, that means, the mixture of sizes are going to be there in both the
cases whether you use a fluid of a particular density or a higher density, only thing that is
going to be changing is the size range of the pure A and pure B that you are going to get
you know by enhancing the density ok. Any more questions, anything that you want me
to repeat a few things ok?
So, today’s class we have kind of looked at applications of what we have learned right
both in terms of you know exploiting this in terms of doing some fundamental
measurements in terms of measuring viscosities by exploiting you know the settling
principles. We also looked at some applications where the settling principles are used in
diagnosis plus we also looked at a couple of examples where people use this motion of the
fluid you know particle in a fluid concept or the settling principles to separate out particles
you know if you have a mixture of particles of different sizes and different densities ok.
399
We talked about sink and float method and the differential settling method for separation
of particles ok.
Maybe with that we will stop today. So, far we have when we looked at motion of particles
in the fluid, we have always taken single particle right when you wrote this force balance,
we took one particle and then we kind of identify the different forces, you know we need
a force balance we got the working equation you know and then we got the expression for
settling velocity right. But that is going to be the case if you are looking at like say falling
ball viscometer as a method for measuring viscosity right.
However, even if you go to this case right you know where you have a blood sample and
you are really not dealing with one particle right. You know in this case you have a very
thin capillary ok, you have a blood sample and you have large number of blood cells right.
So, therefore, you want to talk about you know the collective behaviour now in the next
class where in we are going to look at cases where you have more than one particles in a
fluid you know and look at you know their settling behaviour in the next class ok. We will
stop today and then we will meet up tomorrow here.
Thanks.
400
Fluid Mechanics
Lecture – 36
Colloidal aggregates - Introduction
Today’s class we are going to kind of talk a little bit about again aspects of settling. But I
just want to post this problem by just saying that say that I have a container, and it has
particles ok. Now, if I want to and say that these dimensions of these particles are very
small ok, ok. Say that I have the dimensions of these particles, say they are very tiny ok,
like say sub micron size, say 7 nanometre particle or 15 nanometre particle, when the
particles are very small ok. It so happens that you know even if you want to use external
force like say we talked about centrifugal force right ok.
If of course, we know that for smaller sized particles, gravity is not good enough to settle
the particles down ok. However, you can say that look I am going to go for centrifugal
forces. We know that you know I can generate you know if you look at commercially
available centrifuges, you know you can use like say centrifugal force which is of which
is of the order of you know maybe even to 20,000 times g ok. It turns out that you know
even such high centrifugal force, it will be very difficult to sediment particles of you know
smaller dimension ok.
401
Now, in such cases what do we do? Ok, because, because when people work with you
know particularly dispersions, now that there is a lot of interest in going to smaller and
smaller sized particles right the nano domain ok. Therefore, there comes a case you know
where you are basically interested in looking at you know separating out such small
particles from solution, and it turns out that you know external forces you know kind of
become you know not very applicable in such cases ok.
So, what do we do in such cases? So, I have smaller sized particles. I would like to separate
them out or sediment them out from the solution. And you know I cannot use the external
forces, because the particles are too small ok. One way of doing this would be, what is
that? Make these particles bigger right, how do we do the, how do you make the particle
bigger? Not the particle bigger.
But what I can do is I can actually make the particles large enough in terms of you know
what I can do is, I can start with this case if you know homogeneous dispersed particles, I
can make them what is called as aggregate right ok. I can make the particles come together,
so that they form what are called as multi particle clusters ok, also called as aggregates.
Now, these aggregates you know if they are sufficiently big they could contain you know
many number of particles depending upon the size of the aggregate. And now when the
particles become larger in size we know that you know I can make use of the either gravity
or you know or the other external forces right.
402
Now, how do we go from you know in a state a to state b, what do we do? Can you think
of some ways by which I have a homogeneous dispersion of particles I want to go from
state a to state b, where I have aggregates of particles people also called as cluster of
particles people you know use term called coagulation, flocculation ok, these are different
terms that you come across in the literature, but I want to go from A to B. So, can you
think of some ways of doing that?
Any thoughts? Should I pick some names? How about Saurabh, is Saurabh here? Yes,
Saurabh. All of you have kind of I am sure you have kind of looked at this.
So, basically use additives right ok. Do you know of some process where people use
additives to make the particles aggregate or cluster together? You would have heard in the
context of water purification right people put in what is called as alum which is a aluminum
sulfate which is a salt that is typically added to water to ensure that whatever dirt particles
that you have that is there in water, they become larger in size ok. And the moment such
things are stored in a tank ok, they slowly settle down ok. Therefore, you basically you
know can obtain you know water purification right or achieve water purification ok.
So, one way of doing this would be additives right. But it turns out that you know the kind
of additives that you are going to use, it depends on lot of factors ok. It depends on the on
the type of particle that you have in the dispersion ok, whether the particles are say neutral
or whether the particles are charged. Basically the characteristics of the particles become
403
important. And again what kind of additives you use that again depends on the particle
fluid combination that you are working with ok.
Typically what is done is a two types of additives are typically used; one is basically adding
a salt ok. I gave this example of alum right which is an aluminum sulfate which is a salt
that is added in the water purification. And other kind of additive that people typically use
is a polymer ok. And the way the, the salt helps in the aggregation, and the way the polymer
helps in the aggregation, the mechanism is very different ok.
But the take home is that you know you have a dispersion of particles, very tiny particles
in size, very tiny particles, and I would like to you know make them larger; that means, I
want to cluster them to you know make a multi particle aggregate, and you can use
additives you can either do that by adding salt or by polymers. Of course, there is a lot of
other additives as well which people use, but for today’s class we will just stick to these
two. Now, I have added an additive, how does it work ok? So, the way all these additives
work is that it basically alters ok, it basically changes what is called as inter particle
interactions ok.
In terms of inter we talked about this intermolecular interactions right ok, where we talked
about you know Lennard-Jones potential which has a repulsive part and an attractive part
ok, and that is a plot of you know how the potential varies as a function of separation. This
is what I was mentioning you know in the tutorial that you have to work with dimensions
404
which are sub nanometre; otherwise any distance larger than a nanometre your you know
at least for the Lennard-Jones potential, your potential of interaction is 0; that means,
essentially the molecules are non-interacting right.
So, there is a repulsive part, there is an attractive part, and that is at overall potential, that
is the total potential here. And there is a minima and the you know basically I can get that
by just differentiating this equation, and you know you know equating to 0 right. And it
turns out you know that the minimum occurs at you know that is a minimum position and
that is a energy corresponding to the minimum right. So, similar to intermolecular forces,
you can also play with you know interparticle forces as well ok, when you when you talk
about particles in a solution.
So, I am just going to flash a slide where we talking about in this case we are talking about
colloidal interaction, that means, interaction between the particles in the case of charged
particles in a solution ok. Analogous to the Lennard-Jones potential, so the total potential
phi DLVO is what is called as a total potential. And DLVO actually stands for the four
scientists you know who basically came up with this formalism that the total potential in
the case of a charged stabilized dispersion, that means, a dispersion which contains
charged particles, it basically has the Van der Waals contribution and a electrostatic
contribution ok.
405
And this DLVO stands for the name of the four scientist Derjaguin, Landau, Verwey and
Overbeek ok, these are because these are the two groups of two scientists one from Russia,
one from Netherlands you know which came up with this argument ok. And like the
Lennard-Jones potential, there is a attractive part which is the Van der Waals that is the
one that is in the you know the below that you know zero axis right. And there is a
electrostatic repulsion term ok, and that is the total potential right.
Now, I do not want you this to know a lot of details about this, but the point that I want to
make is there is a Van der Waals term ok; the Van der Waals force of attraction, it depends
on a parameter A which is what is called the Hamaker constant ok, what is called as a H a
m a k e r - Hamaker constant ok. This basically contains all the material properties of the
particle that you are dealing with ok. This takes a certain value for say polymeric particle,
certain value for metal oxide particles, certain value for you know metal particles and stuff
like that ok. And this d is the separation distance ok. So, it depends on the the kind of
particle that you are working with and it depends on the separation distance between the
particle that is the Van der Waals term.
And the electrostatic term it depends on a parameter called kappa ok. And that kappa
depends on the salt concentration that people are working with I have a dispersion and I
have added certain amount of salt. I know the concentration of the salt that I am you know
working with ok. It only depends on the salt concentration; it depends on the the valency
of the ion that I am using I am. Whether I am using a sodium chloride that is Na plus and
Cl minus or whether I am using a CaCl 2 right Ca 2 plus and you know 2 times Cl minus
right. So, it depends on the valency of the ions that I am working with.
And of course, it depends on thermal energy K B T, it depends on the permittivity right of

the medium in which the particles are dispersed. N A is the Avogadro number; e is the
charge that the electron has right the elementary charge ok, plus there is a term called
capital gamma square which basically depends on the surface charge of the particle ok,
whether the particle is highly charged or less charged ok, it depends on something like that
ok.
So, what people do is there is a lot of work that has gone on in you know in looking at
charge stabilized dispersions and their interaction. And people have come up with a way
of manipulating the interaction between the particles ok. Either I can make the you know
406
the you know the attractive forces more or I can make the repulsive forces more ok,
depending upon whether I want to have a stable dispersion. What I mean by that is whether
I want to have a state like this or I want to have a state like this, I can manipulate this by
changing the interactions that is the point ok.
And just to give you a little bit about whenever I have a, so what you are looking at is a
schematic ok. What you have in the inner that is a charged particle ok. In this case, the
particles are positively charged ok. And the moment you have a the moment I put in a
particle in water ok, and if it acquires charges ok, and one of the ways of acquiring charges
could be say that you know I have like say a particle which has say COOH groups on the
surface ok. The moment I put it put this in water, what happens is that it is going to
dissociate and it is going to give a particle that is negatively charged.
Similarly, I am going to have 3 H plus ions right that is the what are called as the counter
ions ok. Now, that is a positively charged particle and these are the negatively charged
counter ions ok. And because of the fact that you know I have a positively charged particle
and there if there are negatively charged ions in the solution, the electric field that is
developed because of the that the positive charges on the particle surface, it is going to
attract the negative charges right. Because of which there is going to be a concentration of
you know counter ions or the negatively charged particle in this case close to the particle
surface ok.
407
So, in a way I have a picture of what is called as a charge that belongs to the particle
surface and the charge that is in the solution ok. This is what is called the electrical double
layer ok, that is the charges on the particle surface the constitute one type of you know
charged surface ok, and there is a charge surface that is surrounding the particle, and this
kind of combinations what is called the electrical double layer ok. I just want to introduce
the terminology ok, this phi electrostatic ok, it is also called as phi electrical double layer
because of the fact that you know the moment you have a charged particle in solution the
configuration around the particle looks something like this ok.
And this is the K that I was talking about this K, if you work out the units of K, it turns out
that K has a units of meter inverse ok. And therefore, this K inverse will have a unit of
meter ok. This K inverse tell you something about the length scale up to which I have to
worry about the electrical double layer interactions ok. I talked about the range of
interactions right, in the case of you know Lennard-Jones I said that you know any distance
greater than 1 nanometre, I do not have to worry about the interactions because beyond
that the interactions are not you know present in essentially right.
In the case of charged stabilized dispersion, if the particles are such that the distance
between the particles is more than K inverse ok, then I can practically neglect all the
electrical double layer interactions or the electrostatic interactions because now the
particles are far enough that the repulsive interactions do not come into picture at all ok.
However, the Van der Waals force of attraction typically occurs in the size range of in the
in the length scale of the order of you know a few nanometres up to about 10 nanometres
that is a distance ok.
Now, what I can do is I know a little bit about Van der Waals, it depends on this type of
the particle that you are working with. I know a little bit about the electrical double layer
interaction, it depends on the kind of salt that I am using it depends on the concentration
of the salt that I am using also the charge on the particle. What I can do is I can actually
play with the double layer interactions ok.
408
This is a case where what is being done is this is your phi net that is a total potential plotted
as a function of distance ok. What is being changed is the Hamaker constant? Ok what I
mean by that is your phi DLVO has phi Van der Waals plus phi electrical double layer.
What is being done is your phi electrical double layer is basically kept constant ok, the
only thing that you are changing is the you know Van der Waals force of attraction ok.
And that I can actually change by changing A right the Hamaker constant.
If I increase the Hamaker constant ok, if I increase the Hamaker constant, you can think
about you know this barrier this barrier as a energy barrier for aggregation, that means, if
the total interaction potential is negative ok. That means, if I somehow ensure that the
particles do not come to a distance which is smaller than the distance that is corresponding
this vertical line right. Any distance that is less than this if I ensure that the particles do
not come to such sufficiently small distances, the particle would continue to remain in a
well dispersed state right ok.
And for them to come to the such close separation where the attraction becomes dominant,
that means, what I have to do is I should ensure that the separation between the particles
never occurs you know what I mean by that is that I should ensure that the separation is
sufficiently large that this energy barrier is kind of prominent that the aggregation does not
occur ok.
The point is that look at the energy barrier the energy barrier is this in the case of 2.0 to 10
power minus 20 J Hamaker constant. I have reduced the Hamaker constant now; now it
409
becomes a little smaller. I have reduced the Hamaker constant further or sorry I have
increased the Hamaker constant further ok. Basically I am playing with the this barrier
right, I am basically reducing the barrier by increasing the Hamaker constant. And what
that leads to is a is at or some point your interaction become completely attractive ok.
Therefore if you keep the Van der Waals force of, so if you keep the electrical double layer
interactions same, if I play with the Van der Waals force of attraction ok, I can actually go
from a nicely dispersed state which is what is you know you would have if your Hamaker
constant is lower 2 into 10 power minus 20 ok. However, if I go on you know increasing
the Hamaker constant that is if I go to A into 10 power A is equal to 2 into 10 power minus
19, I have just reduced the Hamaker constant by one order of magnitude, your interaction
becomes completely attractive, that means, I am going from a well dispersed state to an
aggregated state just by changing the Van der Waals force of attraction ok. Any questions
you have?
The point on you know note is that you know there is an energy barrier for aggregation ok.
And does energy barrier occurs at sufficiently close distances right. If you look at this
right, if I look at this plot, so I would say that you know the energy barrier maximum is
about 1 nanometre right, it occurs at 1 nanometre. That means, if I somehow ensure that
the particles approach a distance that is less than 1 nanometre, that means, I am already
crossing the energy barrier then the particles are going to cluster together, and they are
going to make aggregate right.
And one way of changing this energy barrier is basically I use particles of different types.
Therefore, I am basically increasing the Hamaker constant going from 2 into 10 power
minus 20 to 2 into 10 power minus 19 that leads to the totally attractive interactions, and
you know that your particles would become unstable and aggregation kicks in ok.
410
Second way of doing this would be if I work with particles of different charges ok. I do
not want to, I do not want to get into the details, but the charge is basically measured in
something called as a surface, but surface charge is basically measured in terms of what is
called as a surface potential. The only thing that is important at this point is you know a
higher value of psi means higher charge that is the only thing that is important at this point
ok. And what I what I am doing is I am basically decreasing the surface charge or the I am
decreasing the surface potential going from 231 all the way up to 103 ok.
And you can clearly see that the if this is a if this is a barrier to begin with ok. I am basically
reducing the barrier by changing the surface potential right. Again this is a another way of
making sure that the aggregation occurs, that means, if I am want to go from a an well
dispersed state to an aggregated state, one of the way of second way of doing this would
be playing with the surface potential or the surface charge.
So, in this case, what is done is you have basically kept the again phi DLVO is Van der
Waals plus you know phi electrical double layer, I have kept phi Van der Waals same ok.
Your Hamaker constant is kept constant, your Hamaker constant is kept constant at you
know 2 into 10 power minus 19 joules ok. And you have kept the salt concentration same.
As I said the salt can also influence the potential, I kept that also same. The only thing that
has changed is a the surface potential and that is how you basically manipulating your
electrical double layer interaction ok.
411
And the third way of doing this would be by the addition of salt ok, which is what I said
is most commonly you know followed in the in practice. That again this is the energy
barrier with certain salt; certain salt correspond to certain value of kappa ok. And by
increasing the salt concentration, I am basically reducing the barrier, at some point you
know your electrostatic potential becomes completely the total potential completely you
know becomes attractive that is because again here my phi Van der Waals is kept constant.
I am basically reducing the electrical double layer interactions ok, I am reducing it by
changing the salt concentration ok.
So, therefore, whenever you are working with a particle that have charge, I can play with
the material properties or I can play with additives such as salt right and manipulate the
total interaction. So, that your overall interaction goes from a purely repulsive a case like
this to an attractive. So, there you know I basically induce the clustering of particles right.
412
Now, other ways of doing this would be by putting other additives a polymer for example,
again the use of polymers as additives to you know tune the interaction between the
particles is again a is a big field. There is a lot of literature if you are interested you know
I can go through it. But I just want to give you a feel for a particular term called bridging
flocculation.
So, what could happen is you know if I have particles in a dispersion, I add a little bit of
polymer. And if this polymer has a tendency to go and adsorb onto one of the particle
surface, if this is say particle 1 and if this is say particle 2 ok. If it goes and adsorb onto
the particle surface and if another particle comes in it its vicinity ok, it will also can adsorb
onto other particle as well right.
So, therefore, this polymer kind of bridges the polymer bridges the two; two particles. And
now if many such particles come into picture, for example, I could be one more here one
more here, ok, therefore, I can again make a particle cluster. In this case, the clustering of
particles is basically done by adding a polymer that goes and adsorbs onto the particle
surface. And when many particles come together, so you basically get a multi particle
cluster which is basically held together because of the additives that I have added ok.
Now, ok so, these are a couple of different ways of doing it. I mean again there are different
mechanisms of you know how the polymer addition helps in aggregation, so that we will
not talk.
413
Fluid Mechanics
Lecture - 37
Settling of Colloidal Aggregates
So, now that I have this case right. So now, I would like to look at the objectives of you
know the reason why I have kind of given you a brief background of you know inducing
aggregation is because I would like to look at settling of aggregates. So, far we were
looking at you know a simple case of you know I have a spherical particle settling in a
fluid a stagnant fluid, now I would like to look at settling of aggregates. So now, if you
ask me a question is to how these aggregates form whatever background that I have given
so far would help right ok.
So, you start with a dispersion stable dispersion add additives and you basically end up
with this contradiction right. I would like to think about settling of aggregates. So, what
kind of formulation do you think I should use do you think whatever we developed for the
case of single particle settling, do you think it is going to work? Do you think that whatever
settling principle that we kind of you know formulated for the case of you know single
particles, do you think that is going to work with aggregates?
414
So, it turns out that you know it is not going work ok, but, but the reason why the settling
of aggregates is much more complex; number 1 is that you know you may have aggregates
of different sizes in the right, I have this nice dispersion and I go from you know this state
to this state right you know I form aggregates, but aggregates could be of different sizes
ok. And because these are particle clusters; these aggregates can also deform right you
know they may change their size; they may change their shape. It depends on you know
whether these aggregates are kind of you know settling in a quiescent liquid that is that is
a stagnant liquid or is it in a flow you know flowing conditions everything becomes
important ok.
So, we are going to look at a simple and so, when it comes to the type of aggregates that
are typically formed, people use a concept called there is a concept called fractal
aggregates. So, what you are looking at is three pictures of typical aggregates that people
kind of you know have looked at in the literature and there are some ways of characterizing
these aggregate and this is basically done by using a concept called fractal dimension.
So, the reason why this fractal concept kind of is applicable here is because in a lot of cases
the aggregates that that form they are kind of self similar ok. What I mean by that is if I
look at you know a case b here, I look at aggregates in this small box or I go on increasing
the box size right. It turns out you know if I kind of do analysis of the structure of the
415
aggregates in terms of you know the number of particles in that box, in terms of the way
the particles are arranged it will more or less look very similar ok.
So, in so, therefore, the fractal kind of concept are typically used in characterizing
aggregates, what you are looking at is a case where there are three aggregates which are
of very different morphology right. In one case, it looks more like a linear right aggregate;
in the second case, it looks like a more like an open network of particles and in this case
is a more compact network right. What people do is, we know that if you takes say mass
of the particle goes as volume of the particle into a density right.
Now what I can do is I can say mass of the particle which is a function of radius basically
goes as 4 by 3 pi r cube into rho p right. Think about a solid particle right think about a
solid particle and if I were to start from the center if I go radially outward I can say that
the mass of the particle basically goes as radius to the power of 3 right.
4
𝑚𝑝 = 𝑉𝑝 𝜌𝑝 = 𝑚𝑝 (𝑟) = 𝜋𝑟 3 𝜌𝑝
3
Now, that is the case if I have a particle which is completely or an aggregate which is
completely filled with particle right or that is actually valid for a solid particle right; there
is no spaces right.
416
Now I go to a case where I have maybe I have a picture here right that you know your
mass of the particle going as r power 3 is only valid for a completely solid aggregate right.
Now the fact that you know I have a lot of vacant space here; there is a fluid that is also in
the aggregate right. Therefore, your mp will go for go as r power some dimension Df where
this Df is less than 3 right and this Df is what is called the fractal dimension and typical
values of D, Df is that if you go for an aggregate in a three dimension, the value can go
anywhere from 1 to 3; 3 is a maximum. And if you are working with two dimension, it can
go from one to about two because the maximum that you can have is you know power is
2 right
If you look at the picture here, in A what you see is an aggregate of factor dimension 1.21.
In B, you have a fractal aggregate which is a fractal dimension 1.75. In the case C, you
have a an aggregate which has a fractal dimension of 2.05. And what kind of aggregates
form and whether the fractal dimension is 1.75 or 2.05 it depends on again the kind of
aggregation that happens in the system ok.
A is a case where what was done is you have a container; you have say a positive charged
particle. Now what you do is you add a negatively charged particle to it ok. When that
happens, what will happen is the moment there is a contact between a positive negative
particle there is an there is a cluster formation maybe a two particle cluster. Now when the
next one comes; it can either I mean you know of course, you know it can go out other
location, but it can either go here or here right. In that sense there is a formation of a linear
aggregate and that is what leads to a very low fractal dimension of 1.25 and what you see
in the case B is what is called as a diffusion limited aggregation ok.
What does this mean is there a no again you have a dispersion of particles. The moment a
particle comes into contact with another particle, they just aggregate ok. So, in case of
diffusion limited aggregation people talk about what is called as a sticking probability and
sticking probability is very very high in the case of diffusion limited aggregation. Because
the moment there is an encounter between a particle and other particle or a particle in the
cluster, the particle comes and attaches that is it ok.
However in the case of case C which is what is called as a reaction limited aggregation
what will happen is, a particle joins in aggregate. It can also it will continue to rotate until
it basically finds a more stable configuration and that leads to the formation of more kind
417
of a compact structure ok. The reason why I am trying to say all of this is because
understanding, aggregation sorry settling of aggregates is very complicated because you
know you should know about the kind of clusters that form; whether the clusters are linear
like this, whether they are more open structures like here or whether its a more compact
structure ok. So, therefore, you would had to worry about the morphology of the
aggregates, when you are trying to deal with settling of aggregates ok.
So, in the remaining class, I am just going to and I am going to use lot of assumptions ok.
Let us say that if you want we have a an aggregate something like that ok. The particles
are all connected. How would you like to look at the settling of this ok? I am going to
assume that these aggregates are impermeable. What I mean by that is there is no exchange
of the solvent between the aggregate and the surrounding ok. Because you know we know
that these aggregates are its a composite you can think about aggregate as a composite
particle right. There are particles of course, there is a fluid right. But now if I assume that
there is no fluid exchange between the aggregate and the surrounding, they are what are
called as impermeable aggregates.
And let us say that you know I have a simple case of a dispersion which contains say 50
nanometer particles ok. And say that I have made the particles to aggregate and I have
formed a an aggregate which is like say some dimension, some example of say 500
nanometer particle. There is an aggregate of 500 nanometer you know dimension which I
418
am able to form starting with a 15 dispersion that contains 50 nano meter particles and I
have done that by adding additives; it could be salt or polymer or whatever ok.
Now, it turns out that you know even if I assume that this is a really densely packed
aggregate say that what could be the maximum packing density. If I assume that you know,
they are I have formed an aggregate and this is an aggregate a particle and say that, this is
the most densed most dense spherical aggregate that you can think off ok. What do you
think would be the packing fraction of particle in the aggregate?
Because I said that you know these aggregates are you know you have particles of course,
there is a space where there is fluid as well right and I said that you know it is impermeable.
Therefore, there is a particle plus there is also a fluid right. What could be the maximum
fluid contained? It depends on the kind of packing that I am going to assume right.
If I say that you know the particles in the aggregate are randomly packed your packing
fraction is going to be 0.64 right; that is a maximum packing density. If I say that you
know somehow I have a way of making an aggregate where all the particles are kind of
orderly arranged, some kind of a HCP kind of arrangement hexagonal close packing here
it is going to be 0.74 right. Even if you take the most dense aggregate, it turns out that you
know the typical number of particle that it may have is of the order of 27000 particle ok.
So, what I have done is I have taken a dispersion of 50 nanometer particles and then, I
have made a aggregate by adding salt or you know polymer and if I assume that the density
of the sorry the size of the aggregate that I have formed is 500 nanometer.
If I assume some packing density, I can roughly calculate what is the number of particles
in the in that aggregate right by a simple geometric argument and it turns out that you
know if I use a packing fraction of 0.74. This typically aggregate this aggregate will
consists something about 27000 particle right. Now we know that any dimension which is
sub micrometer size range, we know that if I were to calculate the Reynolds right your N
Re,P is going to be definitely less than 1. Therefore, I can still assume the aggregate to be
settling in the Stokes regime right.
So, right I can assume that right because you know I have made a fine; nanometer particle
say you know I have a way of somehow getting what is the rho of the aggregate I have a
an aggregate of like say size d aggregate and that is settling under the action of gravity and
419
say there you know I have some way of getting what is a rho aggregate. I can actually if
the fact that you know the your Reynolds numbers are less than 1. Whatever formulation
that we are done for the settling of aggregates which is u t is g D p square into rho p minus
rho divided by 18 mu. I can actually use exactly the same formulation I can write this for
settling velocity of aggregate which is g into D.
Now, aggregate dimension square into rho of aggregate minus rho divided by 18 mu I am
of course, talking about isolated aggregate right settling you know in a fluid right is that
ok? Any questions right?
𝑔𝐷𝑝2 (𝜌𝑝 − 𝜌)
𝑢𝑡 =
18𝜇
2
𝑔𝐷𝑎𝑔𝑔 (𝜌𝑎𝑔𝑔 − 𝜌)
𝑢𝑡,𝑎𝑔𝑔 =
18𝜇
All that is being done is I am just saying that you know the aggregates that are formed of
course, they contain a large number of particles because of the fact that the dimensions are
you know still sub you know micrometer. I can still you know use the stokes formulation
because you are now Reynolds number of the particle or the aggregate if you want to call
it as would still be less than 1. I can actually use the same formulism that I use for settling
velocity of the particle.
420
No you can calculate that right because I said that you know the if the particle dimensions
again you know you would have to look at the particle dimension, it turns out to know if
it is a 500 nanometer particle right and if I have the settling in water, if I were to take the
density of water and the you know right. It turns out; if you calculate Reynolds number, it
is the it will be definitely be less than one yeah ok. So, now, when you are working with
aggregates, it turns out that you know the particle the properties of the aggregates are in
between that of the. So, properties are in between the properties of particle and the fluid
um.
What I mean by that is your rho aggregate right which is a the density of the aggregate, I
can write this as rho of particle times 1 minus epsilon I will I am going to define what is
epsilon in a while plus rho of the fluid times epsilon where epsilon is what is called as a
fraction of liquid in the aggregate ok. That is a fraction of liquid in the aggregate that is a
that gives you the liquid fraction and 1 minus epsilon will give you the solid fraction ok.
𝜌𝑎𝑔𝑔 = 𝜌𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 (1 − 𝜖) + 𝜌𝑓𝑙𝑢𝑖𝑑 𝜖
Therefore, now you can say that you know the density of the aggregate is rho of particle
times the particle fraction plus rho of fluid times you know fluid fraction right ok.
Therefore if you think about density, density the aggregate would always be intermediate
to the density of the particle that formed the aggregate and the fluid in which you know it
is dispersed right. So, therefore, I can actually substitute for this I can substitute for this in
421
the expression for you know the u t aggregate, I can simplify this further can you do that?
u t aggregate is g D aggregate square into instead of rho of aggregates I can just write it as
rho of particle into 1 minus epsilon plus rho fluid into epsilon minus rho of fluid right
divided by into right. If you just you know simplify this further that is what you would get
right yeah.
2 2
𝑔𝐷𝑎𝑔𝑔 (𝜌𝑃 (1 − 𝜖) + 𝜌𝑓 𝜖 − 𝜌𝑓 ) 𝑔𝐷𝑎𝑔𝑔 (𝜌𝑃 − 𝜌)(1 − 𝜖)
𝑢𝑡,𝑎𝑔𝑔 = =
18𝜇 18𝜇
Now so what I can also do is I can define the. So, I can I can say that the volume of the
aggregate the volume of the aggregate is equal to the number of particles I have in the
aggregate or basically I do let us put it this way now I want to. So, let us say that you know
I have of course, I do not have a way of calculating epsilon right which is a you know, but
of course, if I can do some assumptions right. I know I can say that you know the maximum
value of this is going to be 0.74. If I assume I ordered packing if I assume ran random
packing is going to 0.64 and if I have a way of calculating fractal dimension, there will be
some way of kind of you know writing this the structure of the aggregate in terms of the
packing fraction right.
And rho p is the density of the particle if you know the parent particle that you are going
to use right. I know the rho p, I know the properties of the fluid in which the particle
settling right the only thing that I do not know is what is D aggregate that is the dimension
422
of the aggregate if we have a way of measuring the aggregate dimension there are ways
by which you can measure that ok. If I have aggregates in a solution whatever method that
people use for particle size measurement people have people can use a similar methods for
measuring what is the D aggregate therefore, if I have a way of calculating this or
measuring this I can actually substitute this back into this equation and I can actually
calculate what is a you know ut aggregation you know the there is a terminal velocity of
the aggregate that was settling in a fluid right.
Ah What you can also do is, I can actually do further simplification and then try and
express this ut aggregate in terms of the settling velocity of the particle itself and the way
to do that would be yeah. Yeah.
No I am. So, see what I am trying to say is that look if I assume that it is an impermeable
aggregate ok. I you know it has some characteristic size, it has some characteristic density
I am assuming it as a spherical you know aggregate and that is settling in a fluid in such
case; I mean this is going to work right. That is a projected area is still I am you know
going to assume it has a 4 pi r square again as I said that this is simplification ok. The
reason why it is a simplification is because if I take you know aggregates like this, you
know it is more complicated right you know for example, the case 1 I have to worry about
whether this linear aggregate that you have at what orientation is settling that becomes
important right ok.
So, we have kind of simplified this problem and all that I am saying is that you know if
you assume the aggregates to be a spherical ok. If I assume them to be impermeable the
effective you know your ap which is the projected area is still going to be you know pi r
square which is the pi times you know the radius of the aggregate square that is all right
ok.
So, maybe I will stop here. So, so we kind of set up the problem now um. So, we have a
way of thinking about if I assume that you know the particles are settling or the aggregates
are settling in stokes regime, I can whatever formalism that we did for settling of solid
particles, I can kind of rework that and then use a similar kind of formulation also for
aggregates. We will try and continue with the rest of the class tomorrow yeah.
Thanks.
423
Lecture – 38
Tutorial - 05
So, hello everyone, now we will look at the problem of calculating sphericity of the
particles. So, in the question in the question it is given that the figure shows the scanning
electron microscopy images of hematite ellipsoids synthesized by method called forced
hydrolysis. And the average dimensions and average aspect ratio of the particles are given.
And we are supposed to calculate the calculate and plot the sphericity of the particles as a
function of aspect ratio.
So, first we will see what is sphericity? So, sphericity is defined as surface area of a sphere
to volume of sphere, divided by surface area of particle by volume of particle; where
volume of sphere equal to volume of particle. So, its actually surface area of sphere by
surface area of the particle in which the volume of the sphere and volume of the particle
are same.
So, here we were given the hematite ellipsoids. And its given that let us look at the first
case where length equal to 57 nanometers and diameter equal to 42 nanometers. So, that
424
alpha is L by d which is 1.3 ok. So, volume of ellipsoid is to calculates sphericity we need
surface area of spear.
As well as surface area of the sphere, which has the same volume as volume of the particle
and surface area of the particle. So, first we will look at the expressions which will give us
the surface area of the ellipsoids. So, even before that we will look at volume of ellipsoid.
So, the volume of ellipsoid is given by 4 pi by 3 times a b c where a b and c are the
dimensions that we get from the ok.
Let us say that this is an ellipsoid and this is the major axis. So, half of it is a; half of it is
taken as a and half of it will be taken as b and in this direction we have c. So, these are the
a b and c and if you see when a b and c are same we will end up with 4 by 3 pi a cubed or
b cubed or c cubed which means 4 by 3 pi r cube which is a volume of an sphere.
Similarly surface area surface area of ellipsoid equal to 4 pi times a b to the power of 1.6;
plus b c to the power of 1.6 plus ca to the power of 1.6 divided by 3 whole to the power of
1 divided by 1.6. And again if here if we see when a equal to b equal to c, we will end up
with the equation very similar to the surface area often sphere ok. So, now we have the
expressions for volume of the ellipsoid surface area of the ellipsoid and we know the
dimensions.
425
And now first we will calculate those. So, even before that let us look at this case, so we
have we know that the volume of the ellipsoid is 4 by 3 times pi times abc. And to calculate
the sphericity, the volume of the particle should be equal to the volume of the volume of
the sphere should be equal to the volume of the particle.
So, pi 4 pi by 3 times abc will give us the volume of the particle. It will be equal to n
sphere, which has an volume of 4 by 3 pi r cube from this we will get that r equal to abc
to the power of 1 by 3. So, now we will calculate the surface area of an sphere with an
radius of r that is abc to the power of 1 by 3. So, surface area of the sphere we will be 4 pi
r square which is 4 times pi times r we know it as abc to the power of 1 by 3.
So, r square will be to the 4 pi times abc to the power of 2 by 3. So, this is the surface area
of this sphere. So, now, if we write the expression for sphericity, sphericity is usually
expressed with an phi s. So, it will be surface area of this sphere 4 pi times abc to the power
of 2 by 3 divided by volume of the sphere which is 4 by 3 actually its 4 by 3 pi r cube.
But we know that its a same as the ellipsoid and we will use the volume of the ellipsoid
which is 4 by 3 times pi times abc. So, this is the now the numerator this whole term will
be surface to the volume ratio of ellipses sphere. And now if you write for the particle we
know that its 4 pi times a b to the power of 1.6 plus bc to the power of 1.6 plus ca to the
power of 1.6 divided by 3 whole to the power of 1 by 1.6.
This is the surface area and we need the volume is 4 by 3 pi times abc. So, this and this
will cancel out 4 pi will go and we will end up with 3 to the power of 1 by 1.6 times. abc
to the power of 2 by 3 divided by a b to the power of 1.6 plus bc to the power of 1.6 plus
ca to the power of 1.6 whole to the power of sorry whole to the power of 1 divided by 1.6.
So, this is the expression we have for calculating the sphericity of an ellipsoid.
426
And now we will just substitute those values. So, for the first case we have we were given
that the length is 57. And diameter is given as 42 sorry its a nanometers. And we have seen
that in the ellipsoid a will be L by 2 and we considered that its symmetric in both the
direction. So, b and c are same. So, b will b equal to c equal to D by 2. With that we will
calculate the sphericity.
So, phi s sphericity will be 3 to the power of 1 by 1.6 times 28.5 which is half of 57
nanometers to 10 to the power of minus 9 times 42 by 2 is 21. So, 21 into 10 to the power
of minus 9 into 21 into 10 to the power of minus 9 whole to the power of 2 by 3 divided
by 28.5 into 21 into 10 to the power of minus 18 to the power of 1.6. Plus 21times 21 times
10 to the power of minus 18 whole to the power of 1.6, plus 28.5 times 21 times 10 to the
power of minus 18 whole to the power of 1.6 and whole to the power of 1 by 1.6.
So, this is the phi s and if you if we do these calculations we will get 0.9845. So, when
sphericity is a comes out to be equal to 1; that means, we have an exact spherical particle.
And here we are getting a value of 0.9845 which is very nearer to 1. And if we see the
images in if you see the images of a first one where alpha equal to 1.3 they look very
similar to spherical particles.
They deviate from they deviate a bit form sphericity. So, that is why we see this deviation
from 1 and we did not get the exact value of 1. So, phi s is 0.9845 for the first case and
similarly if we calculate for the second case phi s will be; phi s will be 0.9397. And for the
427
third case phi s will be 9179 for the fourth case phi s comes out to be 0.8714. Whereas, for
the fifth case we come to it comes out to be 0.7672.
So, these are the different sphericity values for a different aspect ratio of the particles. And
we see that when aspect ratio is being increased we see and decrease in the sphericity.
So, that is a last part of the problem. We are supposed to plot sphericity versus aspect ratio.
So, its 0, so its 1, 2, 3, 4, 5 and the y axis we will the values that we have thought ordering
is in the range of 0.7622, 0.9845 then we will take from 0.7. Let us say this is 0.7 and this
is 0.8, 0.9 and 1.
So, in x axis we have a aspect ratio and y axis we have sphericity. And for the first one the
aspect ratio was 1.3 and comes out to be here 1.3 for that the sphericity is around 1 which
is 0.9845 some something wrong there and the next one we have the aspect ratio 1.7. So,
there 1.7 and the value is 0.9397 we can take it as 0.94.
So, just around there and then for the aspect ratio of 2.1 we have got the sphericity to be
0.9179, which is 0.92 which is around there. And then point a 2.7 for the aspect ratio of
2.7 we have got the value to be 0.8714 something like that.
And the final one the aspect ratio is 4.2. And we got the sphericity to be 0.7672 which is
around; which is around that. So, the line looks something like that. So, this is the plot of
428
it need not to be a straight line, but if you plot this sphericity versus aspect ratio for
different part different shapes of particles this is how we have time.
Thank you.
429
Fluid Mechanics
Lecture - 39
Settling of colloidal aggregates - free settling
Of course we are not going to do that we are going to do a very simple case of we are going
to assume that we have looked at cases where you know I have a particle that was settling
right. Now, we are going to take a case, I am going to say that you know now I am going
to assume the aggregates to be still spherical. And now this spherical aggregate is it has
particles and we said that you know the properties of the aggregate is typically in between
the properties of the fluid and the particle right. In that sense you know we basically
defined a quantity called rho aggregate which basically goes as rho of the particle into 1
minus epsilon plus rho of the fluid into epsilon right epsilon is something called as a liquid
fraction right.
If I have these particle that are solid you know here and there are some vacant spaces right.
So, epsilon is basically the volume of liquid in the aggregate divided by the total volume
of the aggregate ok, that is your epsilon that is a volume fraction of the liquid in the
aggregate.
430
Similarly, I can also define a solid fraction right, so phi p if I say solid fraction this is a
volume of the solids you know or the particles in the aggregate divided by the volume of
the aggregate right. I can define that and they are related by the fact that you know your
epsilon is 1 minus epsilon phi, because your phi p plus epsilon should be 1 right the total
fraction of the solid in the you know aggregate plus the total fraction of the liquid in the
aggregate should be 1 right.
Now, and we said that we are going to take a case of impermeable aggregate; we said we
are going to take a case of a impermeable aggregate. And for that so, basically there is no
exchange of fluid between the aggregate and the fluid right you know in terms of. So, what
that makes is that you know I can still assume if I as; so what it basically means is that if
I go back to this case right, if I assume that the aggregates are impermeable ok. So, I do
not have to worry about you know this fractal dimensions and stuff like that because of the
fact that you know your A P is a projected area right ok.
Now, if I instead of a solid sphere settling I am going to assume that there is an aggregate
that settling and if I have some way of measuring you know the diameter of the aggregate
and the projected area that you know it would still be pi D aggregate square divided by 4
right by pi D square by 4 right.
So, therefore, what I can do is I can actually whatever expression that we developed for
you know some of the settling regime that we talked about right g D p square rho p minus
431
rho divided by 18 mu this was for an isolated particle that was settling I can actually use
the exact same formulation that is ut aggregate. Now, is going to be g D aggregate square
into rho p minus rho divided by 18 mu again this is for the case of isolated aggregates
right.
And I gave an example as to you know why I could still go ahead and use the Stokes
settling expression because if I take like say 15 nanometer particle ok, I said if you make
even an aggregate which is say 500 nanometer in diameter. If I do that I said you know the
fact that you know the particles, the aggregate that you are getting is still in the you know
in the nanometer size range your Reynolds number if you calculate right R e P would still
be much much less than 1 therefore, I could still go ahead and use this you know. So, this
is where we had stopped yesterday right.
Of course what I can do of course this is not rho p right, this is rho aggregate right, this is
rho of it is not rho p anymore right. So, you this is a density of the particle in this case this
is going to be rho of the aggregate right and we said because I know that you know rho of
the aggregate can be expressed in terms of the density of the fluid and the particle you can
write this as ut aggregate is equal to g D p square into rho p minus rho into 1 minus epsilon
right divided by 18 mu that is where we had stopped yesterday right ok.
So, now I want to simplify this further again here this is aggregate right this is D of
aggregate. So, if I want to simplify this further what I can do is I can actually take. So,
432
your volume of the aggregate is the number of particles in the aggregate multiplied by the
volume of the particle plus the volume of the liquid right is that is it ok. So, all I am doing
is you have a you know aggregate of certain volume and that should be equal to the number
of particles that you have in the aggregate multiplied by volume of the aggregate this
basically gives you the total volume of the solids in the aggregate and this V l is the volume
of the liquid that you have in there right.
𝑉𝑎𝑔𝑔 = 𝑁𝑃 𝑉𝑃 + 𝑉𝑙
So, now what I can do is this is pi D aggregate cube by 6 right and that is number of
particles and volume of the particle is say pi D particle right. So, that is the you know the
dimension of the particle that make the aggregate right plus the volume of the liquid I
could write this as the volume fraction ok, I could write this as because I know can I
express this V l in terms of V aggregate, I can do that right. So, what I can do is I can
basically write this as if I say that 1 minus epsilon would 1 minus. So, epsilon is the liquid
fraction right 1 minus epsilon is the solid fraction right. So, therefore, if v agg multiplied
by epsilon would give me what is the volume of the liquid right, yes now yeah. So,
therefore what I can do is I can write this as pi by 6 into D p cube plus epsilon times pi D
p cube by 6 into sorry D aggregate cube into that is it right that is the let me just check
yeah is it ok.
3
𝜋𝐷𝑎𝑔𝑔 𝜋 𝜋 𝜋 3
= 𝑁𝑃 𝐷𝑃3 + 𝜖(1 − 𝜖) = 𝑁𝑃 𝐷𝑃3 + 𝜖 𝐷𝑎𝑔𝑔
6 6 6 6
So, therefore, what I can do is; so this gets cancelled right your pi by 6 gets cancelled
everywhere. So, therefore, I could write this as I can D aggregate power 3 into 1 minus
epsilon it is going to be N p into D p cube therefore, I could write this as D aggregate you
know is equal to N p to the N p by 1 minus epsilon to the power of one-third into D p is it
ok.
3 (1
𝐷𝑎𝑔𝑔 − 𝜖) = 𝑁𝑃 𝐷𝑃3
𝑁𝑃 1/3
𝐷𝑎𝑔𝑔 = ( ) 𝐷𝑃
1−𝜖
So, therefore, I am able to express the diameter of the aggregate in terms of the number of
particles that you have in the aggregate times you know the volume fraction of; so your
433
epsilon which is a void people also call it as you know fraction of the liquid, people have
called it as void and things like that and D p is that is a diameter of the particle right. Let
us think about the limiting case say right we just said that ut aggregate goes as you know
g D aggregate square into rho p minus rho into 1 minus epsilon divided by you know 18
mu right
Now, if I take the limiting case of you know your what can you say about limiting case of
epsilon going to 0 or your phi p going to 1, in that case if I take if I look at this expression
your V l is 0 right and your number of particle is going to be 1 right. So, what I am trying
to say is that this expression that we have it kind of you know becomes your single particle
setting velocity right because your epsilon has 0 therefore, this term goes right number of
particle is 1 here therefore, your D aggregate is basically same as D you know particle
right.
So, essentially I mean you know whatever formalism that we wrote up here you know that
is no extending whatever that we had for single particles to aggregate. It does seem you
know it does make sense because I am able to recover that limiting you know case of single
particle settling ok. Now what I can do is I can actually substitute for D aggregate in terms
of you know this quantity right into the settling velocity expression.
Therefore your ut aggregate would go as g into instead of D p; D aggregate square I am

basically going to put up N p divided by 1 minus epsilon right to the power of one-third
434
right that is what I had into D p whole square right. And of course, there was rho p minus
rho there was 1 minus epsilon divided by 18 mu right that is what you had. So, therefore,
this goes as g into N p to the power of two-third right and I have D p square right and there
is 1 minus epsilon and that is to the power of two-third right minus two-third plus 1 because
that is in the denominator right and there is a one-third power here and there is two here
therefore, if I take it up it becomes minus 2 by 3 plus 1 ok,
And of course, you have and that multiplied by your rho p minus rho divided by 18 mu
therefore, if you sum simplify this it becomes g D p square into rho p minus rho divided
by 18 mu into N p to the power of two-third into 1 minus epsilon to the power of one-third
ok. And therefore, that is a single particle settling velocity that is your ut itself times N p
power two-third into 1 minus epsilon to the power one-third ok.
1 2
𝑁𝑃 (𝜌𝑃 − 𝜌)(1 − 𝜖)
3
𝑢𝑡,𝑎𝑔𝑔 = 𝑔 [( ) 𝐷𝑃 ]
1−𝜖 18𝜇
2 1 (𝜌𝑃 − 𝜌)
= 𝑔𝑁𝑃3 𝐷𝑃2 (1 − 𝜖)3
18𝜇
𝑔𝐷𝑝2 (𝜌𝑃 − 𝜌) 23 1
= 𝑁𝑃 (1 − 𝜖)3
18𝜇
2 1
= 𝑢𝑡 𝑁𝑃3 (1 − 𝜖)3
So, what you are able to do is I mean of course, this is a very simple formulation, you are
able to express the settling velocity of the aggregate in terms of the settling velocity of the
individual particles. And you know and this expression does make sense because you know
that as a number of particle becomes larger and larger of course, your settling velocity has
to go up right that is what it basically captures and it also turns out that it basically also
depends on the porosity of the aggregate ok. If you have some way of measuring you know
the number of particles in that aggregate and if you have some way of measuring what is
the porosity of the aggregate you can basically go back and look at this equation you know
have a way of calculating your settling velocities ok.
435
But in terms of you know making particles of you know the aggregates of this kind and
then doing settling is very hard because first of all making aggregates of controlled size is
very difficult and even if you are able to make it can I make aggregates of that look exactly
identical right because somebody was asking a question yesterday, if you have aggregates
of different sizes then you know it again is more complex right. First of all working with
aggregate you know a single aggregate itself is tough because you know the your a p is
different, porosity is all of that right, but now if I have a collection of such aggregates of
different morphologies you know.
So, it becomes very hard to you know kind of solve this problem, but there is a lot of
interest in looking at settling of aggregates because this is you know very much important
in the case of application right. I said that you know in the case of you know water
purification or you know you look at any other lot of colloidal processing when people do
you add radius to make aggregates and then you know the way they behave in a fluid is
important in the context of applications ok.
So, that is about I just want to give you a kind of a heads up as to how does one think about
you know if you have instead of a nice spherical looking particle you know, if you have
complicated objects like this how does one look at aggregation that was the intention ok.
436
Now, we are going to look at any questions; any questions so far. So, we will look at start
looking at a multi particle systems ok. So, I am again going to draw the classic picture, so
I have a column or maybe like say I have a container and say that I have a 1 particle that
is in a fluid say that you know your fluid is stagnant right that is again I am going to write
the same formula right.
Now, say that I am going to start working with multi particle ok, I am going to put another
one more, but another one more, but another one right. If I go on increasing the
concentration of the particles in the system can I again use a similar formulism that we had
for you know single particle settling ok, any thoughts.
Do you think that whatever formalism that we have had for the single particle settling can
be used for a multi particle system. Yes, now it turns out that it can still use it, but under
certain conditions ok, if you are working under very dilute condition if the distance
between the particles in the fluid if it is approximately 10 times the particle diameter and
if the distance between the wall and the particle, if it is again approximately 10 times the
particle diameter. It turns out that you can assume the particles would still be kind of you
know flowing in a fluid in a very similar way as you know a single particle you know
flowing ok.
And this is something called as a free settling regime ok; something called as a free settling
regime ok. Even though there are multiple particles in the system the behavior of each
437
particle or one particle does not have any influence on how the other particles would settle
or the presence of the boundaries of the container does not affect, does not influence in
any way how each of the particles would you know would kind of behave in the in the
presence of an external field and that is what is called as a free settling regime right.
Now; however, if and as I said right a typical you look at any system right I talked about
you know a viscosity measurements right. When people talk about viscosity measurements
I said that there are something called as a rheometers and I said that you know there are
different geometries right I talked about in a parallel plate geometry, cone on plate and
you know co ed geometries right. When people and in all these measurements, what people
do is people typically maintain a gap right there is a gap between you know two like say
place for example, and one kind of comes up with a question is you know what is the gap
that I should maintain ok.
And even in such cases a typical rule of thumb is that you know you should ensure that
the gap between the particles is at least about 10 times the dimensions of the object that
you have in the fluid ok. That means, if I am working with like say 10 micrometer particle
in a fluid and if I were to put them between two plates and if you wanted to do rheological
measurements. The typical you know rule of thumb is that the distance between the you
know plates has to be at least about 10 times the particle diameter ok.
In that case I would at least maintain a gap that is more than a 100 micrometer because the
particle size that I am working with this 10 micrometer right ok. Therefore, but the similar
kind of things also kind of works in the case of your free settling cases as long as the
distance between the particles is you know larger than you know 10 times the particle
diameter that is when you have this free settle ok.
Now, when you go on adding more and more particles into your system ok, what will
happen is that now as the particles come down it is going to display some fluid right. So,
I you know I do not have this case if you know quiescent or static fluid anymore ok, the
liquid is going to go up right because as the particles come down it will display some fluid
and that fluid is going to go up and your drag for single particle does not apply anymore
because you know the fluid around the particle is also going to be in motion right.
So, therefore, then you would have to think about you know kind of modifying you know
whatever formalism that we had for the single particle case in terms of you know using it
438
for the multiple particle system ok. And so, when the motion of the particle is kind of
influenced by the wall or the other particles in the system people use a term called hindered
settling ok. When you have a large number of particles in your system ok, if one is coming
down if there are other particles which kind of you know influence the drag that the particle
you know experiences ok. So, in such cases you say that you know the settling is occurring
under what is called the hindered settling conditions.
And what people have done is, people have come up with some very simple you know
kind of empirical relationships what are you looking at is a case this is u s is something
called as a hindered settling velocity that is the settling velocity in the multiple particle
system ok. And it goes as the settling velocity of the single particle that is under this is
under free settling condition, this is under hindered settling condition and that depends on
epsilon which again is the fraction of liquid in the suspension or a slurry right ok.
𝑢𝑠 = 𝑢𝑡 (𝜖)𝑛
And this n is an exponent and that depends on the again the flow regime that people work
with ok. If I take like say limiting case of you know there is very low Reynolds number
typical value of n is about you know 4.6 that is you know for Stokes you know settling
regime. And if you look at you know a typical case for you know Reynolds number that is
above of the order of in a 1000, that is your Newton settling regime, it turns out you know
the value of n is of the order of 2.5 ok
439
People have kind of done a lot of experiments and they have kind of come up with this
empirical relationship, that the settling velocity under hinder settling condition basically
goes as the free settling velocity multiplied by you know some function epsilon to the
power of n where epsilon is the fraction of the liquid that you have in the system yeah.
What is that n that I calculate? Correct. So you are talking about here right ok. So, before
I comment on that, what can you say about the settling velocity under hinder settling
conditions, do you think it will be lower or larger than?
It is going to be lower right, yeah it is going to be lower therefore, you know I would still
be under the; so if I were to say that you know I am under the. So, typically if you look it
up let me just do a in one of the examples we have a case where we kind of calculate the
free settling and hindered settling velocities of course, it is slowed down, but not by a large
ok. So, so in that context you know I mean whether you take, so ut here or u s is not going
to shift your you know Reynolds number by a large. Maybe we will comment on it and
when we do that.
So, now, I just want to think a little bit about you know how do we get this ok. So, as I
said this is a people have done a lot of experiments and they have kind of come up with
this empirical relationship, it turns out know you can actually derive it ok. So, that is what
we are going to do for the rest of the class.
Again starting point would again be assuming a certain settling regime you know I can
work it out by assuming that to the particle settling in Newton’s regime or I can basically
consider the Stokes regime. So, starting is going to be again this equation right. Now, I
have this equation and I say that this is applicable for Stoke settling velocity, Stoke settling
regime ok.
Now, because I said that your settling velocities under hinder settling are going to be lower
than the settling velocity under you know free settling conditions. If I calculate the
Reynolds number it will turn out that you know your Reynolds number for the hinder
settling condition would still be in the Stoke settling regime right. So, if I say that look I
can take this expression and I can modify it for the case of hindered settling, what do you
think should be the modification?
440
So, if I go back and look at you know the settling of the aggregates right I said that look I
want to use the same expression, but I want to look at aggregates. I said that you know I
would have to change D p into D aggregate that is what we did right and I said you change
your rho p into rho aggregate that is a modification that we did right.
Now, if I want to again use a similar concept, but I would like to modify this for the settling
in the hinder settling regime. What kind of modifications should I do for this expression?
Yeah, one has to worry about these interactions, but you can still I mean I do not have to
go for you know I mean; so, what he is trying to say that you know if I have a dispersion
of particles say that case 1, case 2. In in case 1 you have like say neutral particles they are
not charged and I take case 2 where I have exactly identical particles same number of
particles in the fluid. In the case 2 there is a you know electrical double layer interaction,
there is a repulsion between the particles. You know the question is the settling in this case
and the second case are they exactly identical ok.
So, you can it turns out that you know I can club all of this into some effective parameters,
I do not have to really worry about you know the inter particle interactions I do not have
to worry about the charges and the Van der Waals force everything. I can still use a similar
expression, but instead of using the properties of the fluid, I would have to use the
properties of the suspending medium sorry let me just say what I mean ok.
All you have to do is now we said that your rho, I was going to take as rho of aggregate in
the settling of aggregates right. Now, that you know I have a fluid now it is not rho p right
your rho right your rho is the density of the fluid that does in which the particles are
embedded right. I mean now what the particle effectively sees is not just the fluid, but the
fluid plus the particles right. In that sense what I have to do is my now rho which is the
rho of the fluid right or the rho of the suspension ok.
441
So, what I mean by that is instead of using rho which is the density of the fluid you would
have to use the rho of the suspension ok. Therefore, one of the modification is your g D p
square into rho p minus rho of the suspension divided by 18. Now, your mu right in the
case of single particle settling mu is the viscosity is the fluid, but now I have a fluid plus
the particles, I would have to worry about not mu of the just the fluid, but mu of the a
particle filled system right. Therefore, this, the modification that I am suppose to be
making is two modifications; one is you replace rho of the fluid with rho of the suspension
or the slurry if you want to call it as ok. The other modification is that you replace your
mu with what is called as a mu effective which is the viscosity of the; not the just the fluid,
but the particle filled fluid right.
𝜌 → 𝜌𝑠𝑢𝑠𝑝𝑒𝑛𝑠𝑖𝑜𝑛 ; 𝜇 → 𝜇𝑒𝑓𝑓
And we kind of had an example in the previous class where we looked at you know how
does the viscosity changes the function of maybe like say shear stress or shear rate right
as a as a concentration of particle right. So, we said we had a case where you know the
particles the addition of the particle you know did increase the viscosity right, if you have
phi is equal to 0 add a little bit of particles the viscosity goes up right.
In general the viscosity is known to increase with the addition of the particle therefore,
your mu is should be replaced with mu effective where mu effective is the viscosity of the
particle filled fluid system right. So, therefore, your ut now you are going to call it as ut
442
relative, I will come back in a minute as to why we call it as ut relative is g D p square into
rho p minus rho of suspension 18 mu effective.
Right, because you know I do not have to worry about particle sizes right because you
know I am assuming that you know you have lot of particles you know I am, you know
when you have multiple particle system you know your D p is exactly same right. I do not
unless I am making the particles cluster together; you know I am, unless I am making the
particles larger by aggregation I would have to worry about you know changing this. But
the fact that you know the particles remain as individual units and they continue to settle.
So, I do not have to worry about this. So, I am going to keep it as the same and then I am
going to change you know this and this right.
And as I said this mu effective it is typically written as mu times some function and this
function f of epsilon is typically greater than, it should be less than 1 right because the mu
effective is typically larger than the viscosity of the fluid itself right.
So, therefore, and your rho which is the rho of suspension I can write it as rho of
suspension, I can write it as rho of the particle into again 1 minus epsilon plus rho of the
fluid into epsilon right I can again similar formalism right. The density of the dispersion
is basically you know some kind of a average of you know the kind of contribution that
comes from the particle and the contribution that comes from the fluid right.
𝜌𝑆 = 𝜌𝑃 (1 − 𝜖) + 𝜌𝜖
443
So, if I substitute these into this equation, it turns out that your ut relative would go as g D
p square into rho p minus rho ok, I had rho of suspension they divided by 18 your mu
effective it becomes f of epsilon here. If I work it out it becomes g D p square into rho p
minus rho into epsilon into f of epsilon divided by 18 mu that is your single particle settling
velocity that is your ut right, that is your sorry that is your ut into epsilon times f of epsilon
ok. I am going to put T here because that is going to be again terminal velocity ok.
𝑔𝐷𝑝2 (𝜌𝑃 − 𝜌𝑠 ) 𝑔𝐷𝑝2 (𝜌𝑃 − 𝜌)

(𝑢𝑡,𝑟𝑒𝑙 ) 𝑇 = 𝑓(𝜖) = 𝜖𝑓(𝜖) = 𝑢𝑡 𝜖𝑓(𝜖)
18𝜇 18𝜇
Now we are going to so, base therefore, your relative terminal velocity under the hindered
settling conditions when you have multiple particle of the system basically goes as ut
which is the velocity, you know terminal velocity under free settling conditions times
epsilon which is the fraction of the liquid that you have in the dispersion multiplied by
some function f of epsilon. So, we will talk a little bit about what this function is you know
it may be in the next class ok. I will just stop here; if you have any questions I will take
them otherwise we will meet tomorrow at 1 o clock you know.
444
Fluid Mechanics
Lecture - 40
Settling in multiple particles system
So, we briefly discussed a multi particle system yesterday.
So, we defined something called as you know free settling ok, something called as
hindered settling right. And we said that you know hindered settling is going to occur if
you have cases where the particles are close enough or if there is a kind of a coupling of
the way the particle settles with the wall of the container or with other particles in the
system ok, then that is when we call what is called as a hindered settling, where the motion
of the particle is kind of influenced by the other particles in the system plus the wall of the
container right.
445
And we said that you know, so basically people have kind of proposed you know some
empirical equations of this sort that is the settling velocity under you know hindered
settling condition goes as you know this is a free settling velocity multiplied by some term,
which is which depends on epsilon, which is the volume fraction of the liquid that you
have in the you know slurry or the suspension. And n is an exponent that takes the different
values depending upon what kind what kind of fluid particle system are you working with
ok.
And we said that you know if you take you know Stoke’s law regime, then you know n is
going to be of the order of 4.6. And if you take the Newton’s settling regime, your n is
going to be of the order of 2.5 right; that is what we said. And then we just want to set this
problem up.
446
And I said that you know if you really want to look at you know the settling velocity in
the multiple in the multiple particle system, I mentioned there any you can actually start
with you know assume a particular settling regime either Stoke’s or Newton’s ok. And
then you basically write down whatever is the working equation for settling for each of
these cases, so we just took g D p square into rho p minus rho divided by 18 mu. You take
that and you modify it to suit the multi particle system right; that is what we mentioned.
And in that connection we said you are going to make two modifications ok, modification
one is that your rho which is the density of the fluid is now you would have to worry about
replacing that with the density of the suspension or the slurry right. And we said that
basically goes as rho of particle into 1 minus epsilon where epsilon, you know 1 minus
epsilon basically gives you the fraction of the solid that you are in the system plus rho of
the fluid multiplied by epsilon right that is one modification.
And the second modification that we said is you should replace your mu with some sort of
effective viscosity ok. And we said this mu effective ok; mu effective you know is going
to be something like mu divided by f of epsilon ok. And we said that you know this
function f of epsilon is typically less than 1, that is because we have seen examples where
we know that you know the viscosity of a fluid with the particles present is higher than the
viscosity of the need fluid itself right.
447
And I was trying to answer a question yesterday where I am said that if you have like say
a liquid flowing through a container right through a pipe for example, and if you say that
you know there is some sort of you know there is a you know you can think about them as
streamlines for example right. Now, what will happen is if you have a similar case with
the particles what happens you know the presence of the particle it basically obstruct the
fluid flow right.
In that sense, you know, your liquid is going to be stopped. So, now, the fact that you know
the there is a slow down slowing down of the liquid. So, your average velocity gradient
basically reduces ok. And if you have this you know mu tau is you know mu times d v by
d y, the fact that you know the average velocity gradient goes down when you have the
particle you for compensating that you know you viscosity has to go up ok. This one way
of thinking why does the viscosity of the fluid increases if you have particles in the system
ok. So, therefore, your f of epsilon is a function which is less than 1 that some you know
you would have to come up with a way of working around it right.
And I also mentioned that I mean in the case of multi particle system, your relative velocity
right is typically related by that is the relative terminal velocity that is because of the fact
that you know you have in the case of single particle system, your fluid was stationary you
know you are U of the fluid is essentially 0 right, only thing that was settling was the only
thing the that was moving was a particle right. But if you have a multi particle system as
448
though as the fluid is displaced, the fluid is going to go up ok. There is going to be U f
component as well and this U f is actually U of particle minus U of the fluid right; that is
how it is. And then you know and we just derived that and it basically goes as U t which
is the terminal settling velocity under the free setting conditions multiplied by epsilon into
f of epsilon right.
If you go back to this expression substitute for you know rho in terms of rho suspension
right, and then replace mu with mu effective, so that is what you will end up ok. Your U
relative terminal is going to be settling velocity under free settling conditions multiplied
by epsilon times f of epsilon ok.
Now, so before we go on you know go further I just want to introduce a few terms ok.
Typically, when people do multiple particle settling right typically people carry out certain
type of experiment which what are called as a batch settling experiments ok. In batch
settling what basically happens is you know you have a container and that container is
basically filled with particles ok. And basically you watch this container with the fluid
particle system as a function of time that is what is typically done ok.
Now, if we if you define couple of quantities something called as a Q P something called

as QP, where QP is the volumetric flow rate of the solid particles in the system ok. The
particles are coming down or settling right, and there are some velocity associated with
the way the particles move in the fluid right ok. Based on the velocity and if there is an
area that is available for them to settle down right. So, your QP is what is called a
volumetric flow rate right, and that would depend on the some velocity right. And P stands
for the particles right.
I am going to use a subscript s sorry subscript s ok. And this s is what is called a superficial,
it is basically stands for superficial times A, where A is the cross sectional area that is
available for the particle to settle right. So, if you have a container, if you have a cylindrical
container, your cross sectional area is going to be you know your whatever circle area right
now. So, if you define ok, if you define the velocity that is associated with the particles by
considering A, which is the which is the entire cross section area of the container ok, then
this velocity that you are you have it here; it is what is called as a superficial particle
velocity ok.
𝑄𝑃 = 𝑈𝑃𝑠 𝐴
449
Similarly, if I define you know a volumetric flow rate for the fluid that is Q f ok, and if I
say that the velocity with which the fluid is basically you know moving is again f for fluid
and s for superficial. Again if I use the entire cross sectional area of the container that is
when these velocities are what are called as a superficial velocities.
𝑄𝑓 = 𝑈𝑓𝑠 𝐴
But however, we know that you know out of this entire cross section, certain area is
occupied by the particle and certain area is occupied by the fluid right. Therefore, if I
actually if I actually consider the actual area that is available for the you know respective
fluid and the particle ok.
Then you define what is called as actual velocities ok, which basically is if you say that
your Q p ok, is U particle superficial times A ok. A is equal to U of just the particle times
what is the area that is available for the particle to settle? It is going to be what is that?
So, if you have a container right ok, say that you know that the cross sectional area is A
right. And if I say that the cross if I want to have an idea about what is the cross sectional
area that is available for the particles to flow, it is A into 1 minus epsilon right. Because if
I if I say that you know the 1 minus epsilon gives you the solid fraction right, the only
thing that I am assuming here is you know volume fraction is typically ,it is a 3D quantity
right you know it is basically for the entire fluid particle system. If I assume that the area
450
fraction is equal to the you know in a 2D sense ok. What is the fractional area that is
available for the for the particle settle is this right. Is that or do you have some?
Now, similarly, for the fluid ok, it is U p sorry U fluid superficial times A and that should
be equal to U of f into A times epsilon right, because epsilon is the liquid fraction right.
So, therefore, in any fluid particle system, if you are basically considering the entire cross
sectional area that is available in the calculation of your velocities that is what is called as
a superficial velocity.
𝑄𝑃 = 𝑈𝑃𝑠 𝐴 = 𝑈𝑃 (1 − 𝜖)
𝑈𝑓 = 𝑈𝑓𝑠 𝐴𝜖
But if you only consider the respective area that is available for either the particle or the
fluid into consideration for getting the velocity of the fluid or the particle, you know these
are what are called as actual velocities. In this case, these are the actual velocities. And
these two terms are what are called as a super. Is that or any, any doubts?
Yeah, it is just a convention that people use that that is about it ok, whether you because
we know that you know the way to go from the volumetric flow rate to the velocities, you
know you would have to divide that by the area right cross sectional area that is available
for the flow ok. If I take the entire cross sectional area that is available both for the particles
in the fluid put together that is what is called as a superficial velocity. However, if I
separate out the fractional area that is available for the particle and the fluid ok; if I use
only respective area available for each of these species, then is what is called as a actual
velocity that is the you know convention ok.
Now, so now, when you do a batch settling experiment ok, because the way the as I said
right way the batch settling experiments are done is you basically take a container right fill
in your particles, and the fluid right and you just watch as the settling occurs right. So, in
such a case, if you I can write that because of the fact that I am not adding any fluid or the
particle into the system externally, I can say that you know your Q p plus Q f should be
equal to 0, because you know in the in a sense there is no net flow ok.
451
Therefore, you are you know the volumetric flow rate of the particles plus the volumetric
flow rate because of the fluid that should be equal to 0 right, because there is no net flow
as such now I am not adding anything to the system or I am not taking anything out ok.
So, now therefore, I can write Q p. So, now, I can define Q p in terms of you know your
actual velocities right. Q p I can write it as U p times A times 1 minus epsilon plus U f
times A times epsilon should be equal to 0 ok. Therefore, I can just cancel this throughout.
Therefore, your U f becomes minus of U p into 1 minus epsilon divided by epsilon right,
is that ok?
𝑄𝑃 + 𝑄𝑓 = 0
𝑈𝑃 𝐴(1 − 𝜖) + 𝑈𝑓 𝐴𝜖 = 0
(1 − 𝜖)
𝑈𝑓 = −𝑈𝑃
𝜖
Now, with this what I can do is I am going to use this, and then go back to the equation
that we developed ok, which was U T U relativity right, U relative T was U T times epsilon
times f of epsilon right we had this expression right. So, I am going to substitute this back
here ok.
(𝑈𝑟𝑒𝑙 ) 𝑇 = 𝑈𝑇 𝜖𝑓(𝜖)
452
So, what are we going to do? What I am going to do is I am going to write it as U p minus
U f right that is my I said U relative T is going to be your U p minus U f right that is what
I said because you know that because both moving here should be equal to epsilon square
into f of epsilon right. Now, I know that if you go back this expression, so I can I can
substitute for U f in terms of U p right, is that. So, I go here. So, this is U p plus U p into
1 minus epsilon divided by epsilon is equal to equal to, I am sorry about that that is going
to be, this is not correct now, just epsilon right that should be equal to your epsilon into f
of epsilon right. So, if I sum if I simplify this, I basically get U p terminal ok, I will say
terminal is going to be equal to epsilon square into f of epsilon right that is what you get
right. Because you know there is an f I can take U p out, so epsilon gets cancelled, one or
epsilon is what you have here basically that is what I end up with. Is it ok?
(𝑈𝑟𝑒𝑙 ) 𝑇 = 𝑈𝑃 − 𝑈𝑓
(𝑈𝑃 − 𝑈𝑓 ) = 𝜖 2 𝑓(𝜖)
𝑇
(1 − 𝜖)
[𝑈𝑃 + 𝑈𝑃 ] = 𝑈𝑇 𝜖𝑓(𝜖)
𝜖
𝑈𝑃,𝑇 = 𝑈𝑇 𝜖 2 𝑓(𝜖)
What is that?
Student: On the right hand side.
On the right hand side, no, no, because I see U f I am basically replacing U f in terms of
U p into 1 minus epsilon divide by epsilon right ok, that is what I have done here.
Oh, yeah sorry there is U T here, I am sorry with that right that is your you know free
settling velocity right or the settling velocity under free settling conditions right. Now, if I
want to, now I still do not know what is f of epsilon right, because you know if I want to
go back and you know think about you know that the plot that we saw right you know you
had n which was a function of Reynolds’s number right. So, if I want to get that n I still
need to find out what is [vocalized- noise] what is this f of epsilon right.
So, what people have done is there are a lot of theories available for what this f of epsilon
is people have done a lot of theory as well as there are a lot of experiments available,
453
wherein we know something about how the viscosity of the fluid changes a function of
particle concentration ok. I am just going to put up some results from the literature.
So, this is a plot a the line that you see is a line for this expression that you have at the you
know on the slide, here eta is the viscosity of the particle filled system or is the viscosity
of the slurry or the suspension. And eta naught is the viscosity of the pure fluid without
any particles the suspending medium, and f is the volume fraction. And this line is actually
the fit to this equation, and the data point that you see here is a experimental data right.
And this is one of the theory that is developed by Einstein ok.
And, so therefore, what I can do is in terms of the effective viscosity function. Now if I
write this as eta is going to be eta naught into 1 minus phi into right minus 2.5 ok. And we
know that the we know that the viscosity of the particle fluid system is actually greater
than you know is larger right. So, therefore, your f of epsilon the function that we want it
should actually be 1 over 1 minus phi into minus 2.5 right; that is your f of epsilon ok.
So, what these relations which basically correlate the viscosity of the particle filled system
to the viscosity of the neat system and the volume fraction. What they will give you is they
will tell you, a way of actually getting this the f of f function which is what I need for me
to go back and you know kind of complete my derivation of you know finding out, you
know what the settling velocities are right. Let me just go back to right here.
454
So, what I can do is I can write this as U T into epsilon square right into 1 divided by 1
minus phi to the power of minus 2.5 right, I can write this as U T into epsilon square into
1 divided by 1 minus phi is your epsilon right that is the liquid fraction to the power of
minus 2.5. So, therefore, your U T basically becomes two point 4.5 right, your U T is U p
relative that is a the particle term in. So, this U p stands further the particles right that is
the terminal velocity of the particle which is settling in a multi particle system where the
hindered settling becomes important, basically goes as the free settling velocity multiplied
by your epsilon to the power 4.5 and of course, there are some conditions under which this
is applicable ok.
1 1
𝑈𝑃,𝑇 = 𝑈𝑇 𝜖 2 −2.5
= 𝑈𝑇 𝜖 2 −2.5 = 𝑈𝑇 𝜖 4.5
(1 − 𝜙) 𝜖
Typically this is typically used where the concentration of your particles is about up to
about 10 percentage or 0.1 ok. But of course, depending upon the epsilon that you are
looking at you know the very liquid fraction or the phi that you are looking at depending
upon the Reynolds’s number of the particle you are working with ok. You will have
different functional form for f of epsilon, you would have to use appropriate functional
form and then substitute in an equation like this ok.
And as we said right this is typically this is only applicable if I am working with the
particles of smaller size where the Reynolds’s number comes in to the Stoke settling
regime right because we started off with you know the Stoke’s settling regime right.
However, if you want to work with other you know settling regimes you would have to
appropriately modify you know the expressions properly yeah ok. So, therefore, basically
the idea of you know talking about this you know it kind of tells you something about how
do people think about multi particle system, and then yeah go ahead, yeah.
That is because see these all these expression that we develop right you know the Newton’s
regimes the Stoke’s regime everything, typically this is applicable for cases where you
have a stagnant fluid ok, and say they do not you let a particle into motion either by you
know inducing gravity all right. In that sense you know your fluid is still stationary ok,
there is no net flow of the fluid as such ok. Though this is U T, it should actually be U T
minus U f fluid ok. This is for the particle and this is for the fluid, but the fact that the fluid
is stationary you know it basically turns out to be you know the terminus is the particle
455
itself yeah. Epsilon should be? Yeah, I mean as I said right people have kind of developed
like for example, so this is typically valid for about 10 percent.
But what are that dilute systems, yes, yes. It turns out that you know the f of the functional
form that you have ok, people have different functional form. For example, if you go for
concentrations which is typically up to about say zero point o three that is 3 percentage by
volume ok. This f of epsilon is typically 1 divided by 1 plus 2.5 times 5 which is basically
a special case of this right. If I take this and then you know if I expand it into you know
higher order terms right if I neglect the higher order terms basically I recover this right.
So, therefore, this is f of epsilon that you choose it depends on you know the kind of
concentration that you are working with you know. So, in that sense you know so you
would have to choose appropriate values of f of epsilon yeah ok. Any more questions?
No?
So, so what I am going to do is. So, I am going to talk a little bit about in experimentally
what people do is this ok. When people are given a slurry ok, and you want to find out you
know their settling behavior. People do what is called as a batch sedimentation test ok; it
is called a batch settling test or a batch sedimentation test. These experimental experiments
are as simple as you take a container, fill your slurry, and just watch the container as a
function of time ok.
456
So, what do you see are the images at three different instants of time ok, time t 1, t 2, t 3 if
you want to call it as right, I say t 0 to begin with and some other time t 1, t 2. And what
happens you know what you look at the case of picture corresponding to t 0, you have a
slurry which is a uniform slurry, the concentration of the slurry you know as a function of
height is exactly the same, that means, no matter from which pocket of the you know
container I draw the solution, I measure their volume fraction is exactly same ok. So,
uniform slurry and correspondingly, if you look at the concentration plot you know
concentration everywhere throughout the height is c b, there is the concentration of the
part that you start with and it is uniform throughout the height ok.
Now, with time what you would see is you get you know in the container, you will start
seeing different zones that are formed ok. There could be a case you know where you have
a zone with clear liquid that means there are no particles, in the system all the particles in
that upper layer have come down ok. And you have a zone B, where the concentration of
the particles in the zone you know in the second zone is exactly same as the concentration
of the particle that you started off with ok. And then of course, at the bottom you have
something called as S that is a sediment and where the particle concentration is going to
be very high right.
So, therefore, if you are going to look at the particle the concentration is a function of
height, the concentration is 0, there because there are no particles; it is a pure fluid ok. And
then in the zone B, the concentration is going to be exactly same as you know your initial
concentration as C B. And then of course, in the sediment, you are going to have a higher
concentration right that is a concentration profile as a function of height ok. And then if
you wait for long enough time what will happen is that you know, so you are you only
your zone B completely disappears, you do not have B anymore ok, that means, all the
particles have settled there, there they are all at the bottom and then you have a clear fluid
ok.
This is a typical kind of a behavior that one sees you know when people do experiment
like this. And what people do from this experiment is you know I can basically track down
the interface right. If I look at the case you know the picture in the center, at some point in
time, I start seeing an interface between A and B right. There is a clear interface ok. There
is an interface that basically separates A and B right that forms at some instant in time
right. And then of course, there is a there is an interface between B and S that could start
457
forming at some instant in time or people doing such batch settling experiment is there you
know you basically follow the interface position as a function of time ok.
And what you do is you can basically plot how the; this is the height of the interface as a
function of time ok. Initially you only so in the beginning of your experiment, what you
have is you basically have a clear interface right, I mean you have you do not have any
clear fluid as such at some point in time ok, you will start seeing a interface between A
and B formed is typically a start with some height. And as the time progresses this interface
is going to descend right, it is going to come down. Therefore, you know you are going to
see a decrease in the height of the AB interface as a function of time.
And BS interface there is an interface between you know, the zone where the concentration
is same as initial concentration, and your solid that is not there initially there is no you
know initially the height of that interface is 0 right. And as its as the you know interface
forms this interface is going to move up right because of which I am going to see a increase
in the height of this you know BS interface is a function of time. And the point where those
two meet is when you know you have an interface between A and S ok. And it you could
have a case where you know your A and S could be flat like this that is going to happen
for cases where the solids that are present in the in the zone S, they are kind of they can
support their own weight ok.
458
What, what may happen in some cases is that you know if there is a compaction that may
happen even after the formation of this A and S interface what could happen is you could
have, so of course, s we will have some fluid and the particle as well right. S is there is a
lot of concentration of particles in the bottom there is also be some fluid ok. Now, imagine
that you may have a case where the particles forms some kind of clusters in the in the in
the bottom ok.
Now, with time you could you could still evolve. So, this A, A means you know in practice
a make it may reduce slowly, but however if the particles that are present in this S can
support its own weight typically you will see a very constant you know AS interface and
that is going to remain as a function of time ok. So, that is the case where you work with
some you know dispersion, and you see the formation of AB interface and AS interface
and finally, you will have AS interface which does not evolve anymore with time right.
Yeah.
No, no, this is actually height as a function of time ok. What I am doing is imagine that I
am taking a photograph of this right at different time ok. Say that you know the height of
the container is say h for example, ok. All I do is I start noting so when say that you know
at some instant in time, you know this is where the A, B interfaces ok. Now, what I do is,
I take this point and I plot it here ok, that is my height of the BA interface right. So, you
get the point.
So, all you are doing is the fact that you know, so the moment you start seeing a clear fluid
in the in your experiment, at that point you have there is a creation of BA interface right
or AB interface ok. All I do is I basically mark the position of that interface and I am
basically following that as a function of time that is this line ok. And the moment I see the
formation of a BS interface, I start noting down the position, and I basically follow the
height of that as a function of time that is basically this line ok.
And the point where they meet is when the AS interface forms right ok, and that is your
this line. As I said this could be remaining constant or this could evolve all depends on the
kind of particle that you have in the system ok. If you have typically if you have a rigid
particle ok, it remains constant. If you have like say fluid like particle you know the
emulsions or droplets or something that or if you have particles that are squishy ok, people
459
see a some slow evolution, but however that you know evolution could be very negligible
in the case of hard particles yeah.
Now, so this is a another experiment where what people do is a similar to the previous case
ok. You have zone A - the clear fluid. Zone B, where the concentration is same as you
know the initial concentration. You of course, have the sediment at the bottom. There is a
zone E, where the concentration fluctuations may happen ok.
What I mean by that is if I look at if I look at the concentration plot right, the concentration
in the top zone is exactly you know 0, there is no particles there, and B is same as the
initial concentration. And you have a zone E, where the concentration basically varies as
a function of height ok; at the top the concentration is as same as you know what do you
have in B; at the bottom the concentration is as same as what you have in S ok. However,
across the height the concentration does change ok.
And you know again depending upon you know what you know so what will happen is
your B could completely disappear, and then you know you only have an interface between
A and E, and an interface between E and S. Again you follow that as a function of time
ultimately you again end up with A and B in the, so A and sorry this is this should be S
right. This should be S right that is a sediment at the bottom ok.
460
Now, whether the type 1 sedimentation or the type 2 ok, where you the formation you have
the formation of this zone, where the concentration raised across the height, it typically
depends on you know what kind of initial solid concentration that you are working with
ok. Typically people have seen that if you are working with the a slurry where the
concentrations are lower typically of the order of you know about 20 percentage or less
ok, you see type 1 sedimentation ok. However, if you go for higher concentration you
know initially that is when you see a type 2 sedimentation.
The reason why I wanted to talk about such kind of test is there you know you have a lab
in the next semester where you want to look at settling of slurries. You have some
experiments where you are going to look at containers which have slurries of different
concentration. You are again going to look at you know monitoring the interface position
as a function of time, and that will you know tell you that will help you in terms of
obtaining the settling velocities ok. We will talk a little bit about that in the, you know in
the in the in the next few minutes or so ok.
Now, any questions? So, now, let us look at, so we did this right, your U p you know goes
as U T times epsilon to the power of 4.5 is what we did ok, but however in general it is
typically epsilon to the power of n right, and I said and depends on what kind of settling
that you are working with right. Now, from this what I can do is I can actually define U p
s which is the which is what is called as a particle settling flux U p s is what is called as a
461
particle settling flux, which is basically your U p times 1 minus epsilon right. I am basically
going back to the superficial you know concept right. So, if you want to get the superficial
settling velocity U p s should go as U p into 1 minus epsilon.
Therefore, I can write this as U T into 1 minus epsilon times epsilon to the power of n ok.
And I can talk about the dimensional flux ok, which is U p s divide by U t would go as 1
minus epsilon into epsilon any, any, any issues that is right. So, I am basically going back
to the superficial velocity concept ok.
𝑈𝑃𝑠
= (1 − 𝜖)𝜖 𝑛
𝑈𝑇
Now, it turns out that if I basically plot you know this U p s divided by U T as a function
of epsilon ok, it turns out you know it basically goes through a maxima ok, and then it
starts decreasing and then there is an inflection point here and then it again goes on
decreasing further. You can you can do that ok.
What you do is you basically take this expression vary epsilon from 0 to maximum is going
to be 1 right. If you do that ok, you basically get a plot like this. So, this the maxima and
the inflection point, I can get it by basically differentiating and equating to 0, and then I
can double derivative right. So, if I do that, this maxima seems to appear at a concentration
where your epsilon goes as n minus 1 divided by n plus 1 sorry that is going to be n divided
by ok. And this inflection point seems to occur at n minus 1 divided by n plus one ok.
So, so this plot is basically the dimensional settling flux where U p s is your superficial
particle velocity right that has a velocity which is based on the entire cross sectional area
divided by U T, which is the settling velocity under the free settling conditions right that
is U p s by U T as a function of epsilon ok. If I do that the maxima appears to be at n
divided by n plus 1, and your inflection point appears to be n minus 1 divided by n plus 1.
If I substitute for n for the case where we put n is equal to 4.5 right; if you do that it turns
out on to this maxima corresponds to 0.177 ok. And this inflection point corresponds to
0.35 epsilon is 0.35 ok. So, the reason why I want to do this is because the type of settling
that we talked about ok, whether you have a sedimentation type sedimentation 1 or
sedimentation type 2 that basically depends on your initial concentration in the dispersion.
462
Therefore, so these numbers they basically define some kind of limits where people are
going to see a very different kind of settling behavior depending upon the initial
concentration that you have in the dispersion ok, yeah. Is it ok, any questions? So, all that
we have done is we took this you know the general form of you know U p goes as U T
power epsilon power n, so we took that ok. And from that we basically obtain what is the
superficial settling velocity of the particles ok, which basically is U p times 1 minus epsilon
right. And I substitute for U p from here which is U T times epsilon power n. I get this
expression.
All I have done is I have done the first and the second derivatives equate that to 0, it gets
me access to this at this point. From that I can basically get some values of epsilon that
basically tells you something about the different settling behavior that we saw in type 1,
type 2, ok, it basically corresponds to concentrations where the concentration is typically
less than point 0.177, the type 1.
But whereas, at a higher concentration you would have you know a type 2 sedimentation
ok. So, that is the reverse side because you know this is the higher concentration right your
epsilon is smaller means your particle concentration is higher right. So, you will get type
2 sedimentation in this regime. And whereas, in this regime so basically from here to here
you will have a type 1 sedimentation ok, whether you have a zone of variable concentration
formed or not ok, it would depend on the concentration that you work with right ok.
So, now if you go back and take a look at these let me just go back to ok. So, if you look
at if you look at typical sedimentation test right, you it so happens you know you will
always come across cases where you have a sharp interface ok, where you have a sharp
interface, and that sharp interface separates zones of different particle concentrations ok.
463
So, what I mean by that is let us take a simple case I say that you know I have a case where
I start with you know a certain number of particles here ok. Say that at some point in time,
you have a clear interface that is my interface ok. Now, you have particles of some
concentration here; say that the concentration here is C 1 that basically is going to be 1
minus epsilon 1 right you. I am going to define concentration as in terms of your epsilon
is your liquid fraction 1 minus epsilon 1 will give you the solid fraction. I am going to
denote that as C1 in this case.
So, there is going to be some particles here say that the particle setting velocity is here is
U p 1 say that. So, you have started with a dispersion of some concentration. And then as
a as I observed the you know the contents ok, I see a case where there is a interface form
that is a clear interface between your two regions, where in one region the concentration
is C 1, the in the other region the concentration is C 2, which again is 1 minus epsilon 2
ok. And say that you know your interface is basically moving down with the velocity U
int that is an interface velocity right.
Now, if the particle settling velocity is U p 1 in the above region, and if the particle velocity
is U p 2 in the below region it turns out that I can think about a simple mass balance ok.
That mass balance typically reads something like this. U p 1 which is the particle settling
velocity in the region where the concentration is c 1 minus U interface that will give me
the relative velocity of the particles with respect to the interface multiplied by the
464
concentration of the particle that you have in the above zone that should be equal to the U
p 2 minus U interface which again is the relative velocity of the particles in the bottom
zone as a function you know related to the interface velocity multiplied by C 2 ok.
[𝑈𝑃1 − 𝑈𝑖𝑛𝑡 ]𝐶1 = [𝑈𝑃2 − 𝑈𝑖𝑛𝑡 ]𝐶2
So, all that you are doing is for the condition is for you to get a clear interface ok, for you
to get a perfectly sharp clear interface what is the concentration of the particles that go that
that reach the interface from the above should be same as the concentrations of the particle
that leave the interface from the two different zones ok. This is basically if you talk in
terms of this is basically the relative velocity right times if you look at this concentration
right, it is effectively a mass balance ok. Mass flux that is coming down should be same
as the mass flux that is you know going down ok. Now, from this what I can do is I can
actually get an expression for U int ok that basically goes as U p 1. So, U int goes as U p
1 C 1 minus U p 2 C 2 divided by C 1 minus C 2 ok.
𝑈𝑃1 𝐶1 − 𝑈𝑃2 𝐶2
𝑈𝑖𝑛𝑡 =
𝐶1 − 𝐶2
Now, if you go back to the cases where I had a region between the clear fluid and the zone
of some concentration ok, I know that one of the concentration is 0 right. I know that the
constant one of the concentration is 0 right. Therefore, therefore, your U int would be
proportional to your U p 1, because if I have cases where I have a zone or an interface
465
which separates a fluid with no particles, and a fluid with particles ok. In such a case my
one of the concentration is 0. Therefore, if you just monitor U int that is the interface
velocity I can actually directly calculate what is the particle setting velocity.
𝑈𝑖𝑛𝑡 ∝ 𝑈𝑃1
So, so therefore, basically you have if you have cases where you know there is a clear
sorry about that. So, so if you have cases where there is a interface between the clear fluid
and zone where you have particles of some concentration ok, I can use this formalism that
you know the settling flux or the mass balance if I do if I do it, if I get an expression for U
int right, I can actually calculate what is the settling velocity just by monitoring the
interface velocities right.
You know if you look at the case right I said that the interface was basically decreasing.
All I do is I take this slope that d h by d t will give me the velocity with which the interface
is basically falling that is exactly same as you know I can relate that to the velocity of the
particles are basically settling ok. Is it ok? Yeah; so, yeah.
Yeah this is only when C 2 is 0 ok. But what people do is now I can I can actually write
this in terms of, but of course, you know if you so when we did this U p s by U t right as
a function of epsilon then when we did ok. And epsilon if I you know I just use epsilon
here, but you know epsilon would be changing as a function of you know depending upon
you know the your sedimentation that is a occurring right, your epsilon would be changing
all the time right.
Therefore, what people do is you know we did that that the flux plot right, then you know
you can actually use the I can take two points I can get a tangent. And from the slope there
are ways of actually calculating what is your U interface and relating that to interface
velocity ok. We will not going to the details of that. But the point that I want to make is
that you know whenever in the experiment you are going to do in the next semester, you
basically follow the interface between the clear fluid and the you know and the bottom
fluid with some particle concentration. And because you are basically measuring U
interface that will be a nice way to back calculate your settling velocities ok.
466
This is going to be useful because see if you are working with like say system where I can
observe the particles and looking look at their you know position and find their velocities
right. If I have a way of doing that you know my job is easy, but you know if you have
cases where you have very fine particles, if the concentration is really large in such cases
you know how do I calculate the particle setting velocities ok. So, the only way of doing
that would be basically by monitoring this U interface and then from that you back
calculate what is your the settling velocities ok.
Maybe I will stop here and the I think with this we are going to stop with you know the
concept of so we have basically looked at three concepts settling of single particles ok, and
implications of that in terms of applications right, and then we looked at settling of
aggregates ok, and then we look at multi particle systems right that is what we have done
so far ok.
So, what I am going to do in the next class is to look at a case where what you do is you
basically have a like say a pipe for example, ok, and what we are going to do is I am going
to have a support plate, and then I am going to fill this pipe with particles ok. And then
you are going to look at how does you know the fluid flow occurs in cases where I have a
container which is filled with particles, and we would like to look at you know the flow
through what are called as packed beds right is what we are going to look at in the next
class ok. So, we will do that on Friday, yeah.
467
Fluid Mechanics
Lecture - 41
Flow through packed bed
So, actually today I would like to start with a new topic which is on Flow through packed
bed; is it clear, as to in the case of multi particle systems ok.
So, we kind of derived this in U T is, sorry U P that is the particle settling velocity under
you know hindered settling condition goes as U T power epsilon power n right. Is that
clear, is that clear how we got this right? We basically said that U relative terminal goes
as U T times epsilon times f of epsilon right, we derived that. Then we said that you know
U T, U particle under hinder settling, it goes as epsilon square into f of epsilon we derive
that.
This was basically derived by bringing in the concept of you know the volumetric flow
rate of the particles plus volumetric flow rate of the fluid should be equal to 0, because
there is no net flow in the system right. And, then we define couple of terms called as
actual velocities and the superficial velocities ok. You can actually think about writing a
very simple continuity equation.
468
So, your Q P which is the volumetric flow rate of the solids goes as U P superficial; S for
superficial times A which is the area of cross section there is available for the entire fluid
particle system to flow or that is equal to U P times 1 minus epsilon, because you know
times A of course, right there is A as well. Only thing I have done is I have only taken the
area that is available for the particles to flow right.
Similarly, I can write U Q f which is the volumetric flow rate of the liquid as U fluid
superficial times A that is equal to U fluid times A into epsilon right. This basically comes
from the couple of definitions of you know the superficial velocities and the actual
velocities right. Then we went on to say that you know this f of epsilon is a function which
is typically less than 1 and there are different models that people have proposed, and
according to one of the you know models you have your f of epsilon was you know epsilon
to the power 2.5 right we derived that.
Therefore, your U P comes out to be epsilon to the power of 4.5 and we said you know
this is typically applicable for you know your epsilon which is of the order of 0.9 or you
know your phi of the order of about 10 percentage you know order of 0.1 percentage right.
And, there are people have a lot of experiments to kind of verify some of these things and
I said that this value of n depends on the kind of condition that you are working with right,
Newton’s regime you know Stokes regime and things like that.
There is a very famous equation which is what is called as a Richardson Zaki, I think
equation in which what people they did, they did a lot of experiments on you know
hindered settling conditions and they did they worked with lot of particles of different
types. And, then they kind of showed that this experiment n is 4.65 ok, though according
to theory it should be about 4.5 and experimentally people have shown that you know this
value is typically slightly higher and which is about 4.65 ok.
469
And, other thing that we did was we started talking about U P S by U T that is a settling
flux non dimensional settling flux and the way we did that was your U P is you know
epsilon to the power of n times right U T right ok. And therefore, your U P S which is the,
the particle settling velocity, but the superficial velocity right goes as U P times 1 minus I
am just quickly do that ok, it goes as U P times 1 minus epsilon right therefore, your U P
S is going to be U T into 1 minus epsilon into epsilon power n right.
And, we said if you want to plot this ok. So, it would the function would go something like
that is what I had mentioned ok. This is actually plot of U P S by U T versus 1 minus
epsilon I did a mistake in the last class. It is not epsilon is 1 minus epsilon, it turns out I
can get this point and this point just by differentiating this 1 you know a couple of times
right. So, if I do that, so, if I take this U P S by U T and if I differentiate it once d by d
epsilon if you do it ok.
What I will do is I will get this as 1 minus epsilon into n into n minus you know epsilon
power n minus 1 plus of minus 1 into epsilon power n is equal to 0 if I equate that to 0.
Therefore, if I work this out I would basically get an expression for epsilon as n divided
by n plus 1 that is your this point and if I do the double derivative and if I do it you basically
come you know this particular point comes across to be n minus 1 divided by n plus 1.
And, this is this particular plot is you know what I have drawn is for epsilon sorry n is
equal to 4.65 that is the Richardson Zaki equation ok.
470
And it turns out you know this particular point corresponds to 0.177 and this 0.35; that
means, if I have you know this is 1 minus epsilon or its in its a concentration right or you
can think about this is a particle fraction. And, if your particle concentration is less than
0.177 you know you will see some kind of settling behavior in your system and then you
know of course, there is some region where you see some other settling region and you
know.
So, basically depending upon the initial concentration that you work with ok; so, the
settling if you were to do this batch settling experiments you will see different behavior
that I discussed in the last class right there is a type 1 settling or type 2 settling or something
very different from the these two, everything depends on the initial concentration of the
you know particle that you work with.
And in the case of so the other point that I wanted to mention is what is called as you know
there are when you do the settling experiments you will always have sharp interfaces that
is what I kind of showed in those batch settling you know experiments right. So; that
means, I could have a case where you know I have a uniform dispersion to begin with.
Everywhere the particle concentration is the same, but then you know at some point in
time you would have a case where I have an interface and I could have a case where I have
a clear fluid at the top and you know all the particles are bottom or in general I could have
471
a case where the particle concentration is less here. Concentration could be C 1 here and
the particle concentration could be bit more here right.
So, in such a case I can have a settling velocity associated with each of these phases ok.
This could be U P 1 you can take this regime 1 or 2 and in this case you know the settling
velocity is going to be U P 2 and if your interface is moving down with a velocity U int
that is the interface velocity. And, I said you can think about writing a simple mass balance
which basically is U P 1 minus U interface multiplied by your concentration, concentration
C 1 in this case is 1 minus epsilon 1 you are basically expressing this in terms of the you
know void fractions right should be equal to U P 2 minus U int into C 2 right. If you work
it out it turns out you know this basically will have units of mass per second ok.
Because this is you know meter per second right and this is your concentration volume
fraction of course, I can have the area multiplied on both sides right. So, basically this is a
mass balance which basically is telling you that if you want to have a clear interface
whatever the number of particles or the concentrations of particle that that will reach the
interface the same number of particles have to go ok. So, if you work it out your U int
becomes U P 1 C 1 minus U P 2 C 2 divided by C 1 minus C 2. If I have a case where I
have a clear fluid here and then all the particles are in the you know in the bottom region
regime. So, one of the concentration is going to be 0 right. If you say the C 1 is 0 that is a
in the top regime. So, basically this gets canceled you basically get U P 2 itself right.
So, therefore, when you do this settling experiment; so, when you are following the
interface velocity right which is basically you know I said that right. If you have different
interfaces there could be a case where some interfaces are basically you know following
as a function of time, there could be cases where the interface is you know rising as a
function of time all I need to do is get this slope that as you are plotting you know height
versus time I get this dh by dt that is basically my interface velocity and from that I have
a way of calculating the settling velocity ok.
So, we will stop here at this point. So, we have talked about you know different things in
the previous lectures right. We talked about settling of single particles then we talked about
how do we extend that to an aggregate of particles and then again extension of the similar
concept to cases where we have multiple particle systems right.
472
So, now we will talk about a concept called flow through packed beds ok.
And, in a sense packed are packed beds are kind of it is a simple concept you know what
do you basically mean by packed bed is you basically have a column ok. It could be a pipe,
it could be you know a container right and this column is basically filled with this is your
column ok. This is your packing material ok.
Now, I had mentioned in you know few lectures ago that the reason why the packed beds
are used are because they want you know they provide as a way of improving the contact
473
between two or more phases ok. I could have a case you know for example, say that you
know have a say a simple pipe hollow pipe and in which you know I am want to send 2
liquids like say you know maybe some methane or you know some liquid on a gas for
example, if I am going to ensure a proper contact between the two if I were to have a
simple pipe without any of these abstraction you know you know obstruction. So, there is
no proper contact between the two phases right two fluid phases ok.
So, therefore, the one of the main advantages of using this packed beds is you know they
essentially ensure that if you have a two phase or three phase you know systems they give
you a way of improving the contact between different phases. And, people typically use
this for cases where there is a gas liquid operation or a liquid liquid or a liquid solid,
depending upon you know there are a lot of cases in chemical industries where you come
across you know different you know fluids that are going to be you know you want them
to be in contact either liquid liquid, liquid solid or gas solid. So, in such cases people make
use of these devices to ensure that you know there is a proper contact between the different
between different phases.
Now so, this is again a schematic that I picked up this is an example of a case where people
have this is an absorption column. Again typically what you will have is you know some
packing material in the pipe, there is some liquid is coming in there could be some gas that
is coming out you know if your objective is to either you know take out if you have a if
474
there is a gas for example, if the gas has some trace amount of chemical for example, mist
you know.
Or if there is some for example, if I want to have a clean gather that I want to come out if
there is some unwanted gas that you want to separate out one of the best weight of doing
this would be use of this packed bed ok. Let it pass through the pass up through the column
and then you know use an appropriate liquid which basically takes out this unwanted you
know material and it comes out of the bottom and again you can you know subject to
whatever purification that you want to do ok.
And people use them for the scrubbing or degassing operations which is to remove traces
of gas, vapors or chemical mist. Absorption column people use it for you know catalytic
reactors ok. If you really want to if you want to have a particular reaction that I want to
occur over a solid surface the best way of doing this would be again you pack them into
these columns and as you let either a liquid or a gas through this column on the surface of
this catalyst you know some reaction may occur right.
So, yeah as I said right they are used for distillation operations a lot of mass transfer
operation in absorption is one example I talked about chemical reactors separators ok. So,
these have very wide applications and you will have couple of experiments in the in the
next semester to look at you know different aspects of flow through packed beds ok.
So, now before we go to so the reason why we are interested in the packed column is
because like we talked about pressure drop in a pipe right. Now that now if you have a
pipe that is filled with you know particles right. Again now the fact that you know there is
much more obstruction for the liquid to flow through your delta P is going to be much
much larger compared to the you know case of flow through pipe right.
So, the ultimate aim of some other thing that we are going to discuss is to look at how do
we think about pressure drop through this such systems and we are going to think about
some ways of analyzing you know flow through the packed bed. And, you know how do
we kind of correlate the pressure drop to different parameters like you know the structure
of the packing in a flow of the flow rate of the liquid and you know a few things like that
is what we are going to do ok.
475
So, it turns out that you know. So, you know of a particular operation called filtration
which is some kind of a its a special case of flow through packed bed right. Again filtration
one of the simplest thing that you can think about it you know when people do this coffee
or tea at home right when you when you strain now the coffee or the tea on the you know
strainer right you have some solid bed that is initially its a its a clean you know cloth or
something like that at some point in time you know as you saw start straining it you have
a bed of solid that form right.
So, ultimately you know the liquid starts flowing through the bed of solid that is formed
on the surface right. So, therefore, we are going to talk about filtration at some point ok.
We are going to say that you know whatever you know working equation that we
developed for packed beds we can actually modify it to you know suit the need of filtration
processes we will talk a little bit about that.
476
Before we do that I am going to talk a little bit about what are called as packed bed
internals. So, let us look at the construction ok. So, what you are seeing is a as a picture
where there is a packed column. So, this is two regions where there is particles are kind of
packed right. Right this is solid particle which are put in the column and the bottom you
have a support grid because I mean this grid is basically required because I want to hold
the packing right. If I do not have this grid all this particles would fall off right.
So, they typically there is a support grid is typically a porous plate and again people use
different kinds of design to ensure that you know there is a proper flow of fluid through
these you know fluid through these columns. At the top somewhere here and somewhere
here you have what is called as a liquid distributor or liquid the redistributor. And, the job
of this liquid distributor is to ensure that there is a proper distribution of liquid across the
entire packed bed that is the job of these liquid distributors.
And, if you look at somewhere here they have what is called as a liquid collector and these
the job of this liquid collector is you know if you have a really long column what may
happen is there you know as it flows through what typically fluid does is, it wants to go
through the least resistance path right. It will find out you know some least resistance path
you know after sometime of course, if you look at the top of the bed there could be a proper
distribution of the fluid because of this liquid distributors ok.
477
However, if you go down the column if the column is really really high you know you
know this liquid take some preferential paths and these are the least resistance path ok.
One of the way to avoid that would be you put in this liquid collector ok. The job of this
liquid collector is we basically collect the liquid as coming out down the packed column
and then you know it kind of feeds it to a liquid redistributor ok. So, these operations you
know these kind of internals are really required to ensure that the flow rate across the
different cross sections of the bed is kind of uniform. So, there you know there is a proper
contact between the fluids that you are trying to work with ok.
So, this is a different kind of liquid distributor as I was mentioning. Liquid distributor
basically provide uniform distribution of liquid and as I said if the height of the packed
bed is very very large ok, there may be something called as a channeling ok. The
channeling is what I said you know when the liquid starts taking some preferential paths;
that means, there is a larger flow in one region of the packed bed versus the other region
that is when you have what is called as a channeling and channeling can be avoided if you
use these redistributors somethings like that ok.
So, these are four different types of distributor people use. These are some kind of a porous
pipes ok. This is a pipe through which the liquid is flowing, but you know there are regions
where you know I let the liquid into the column. These two this and this, these what is
called as a notch type distributor if you see this right. There is a wedged kind of thing right
478
if you look at these things. So, there is the construction is something like that its a parallel
channel like that channel like that and then there is a notch and basically you fill the liquid
in this you know this you know channels and then through the notch liquid basically comes
down that is the case here and here ok.
This is an example of what is called as a orifice you know type of liquid distributor which
basically has porous you know porous openings right and then through that the liquid
basically flows ok. So, these are what people typically you know use and as I said right
the objective of this liquid distributor is to ensure that to avoid channeling and to ensure
that there is a improved efficiency right that is the whole point.
Now, I just want to define a few terms, first term that we are going to do is something
called as a porosity or if you already know this concept, but I still want to do it quickly ok.
479
So, porosity it is also called as a void fraction void fraction or porosity or people also called
as voidage as defined as volume of the voids in the bed divided by the volume of the
packed bed and volume of the voids in the bed I can basically write it as volume of the bed
minus volume of the solids right. And therefore, I can write this as 1 minus volume of the
solids divided by volume of the a packed bed and volume of the solids divided by volume
of the packed bed we know that you know its the packing fraction of the solid right. So,
that is a typical definition of you know the voidage
Now, so, how do we find voidage in the packed bed ok? So, I say that you know I have a
column ok. I am going to say that hey look you know I have a column that is filled with
this my packed bed and I want to find out what is the voidage ok. What do you think is a
you know, can you think of some ways of by which I can measure the void fraction ok?
Experimentally you can think of a couple of different simple ways of doing it ok. What
you do is so you take a empty column and with a packed bed what you do is you start
going on filling the fluid ok. You start going on filling the fluid. So, what fluid is going to
do is if I have a support plate and it will start occupying all the pores in the bed right. The
liquid that I am is going to add is going to basically occupy all the pores in the fluid.
Now, what you do is if you have a some way of collecting that fluid and then measure its
volume ok. If I know if the is a cylindrical column if I know the length and the cross
sectional area right. I can get the volume of the bed and if I measure what is the volume
480
that I have collected right that is one way of doing it. Other ways of doing this would be
you finish up your packed bed experiment and ensure that you know your liquid is
basically kind of filled up to the height of the packed bed and will basically drain the fluid
right. So, you can different ways of doing that.
Now, it turns out the porosity is one of the really important parameters that is you know
kind of helping the you know kind of achieving. So, when I talked about this efficiency
right, efficiency when I mean by what I mean by efficiency is that you know, if you are
able to ensure a better contact between the different phases that is when you know I can
say, you know the efficiency of the operation that I am looking at that its you know
reactions or whether it is in a mass transfer operation that is when that is what I am looking
for right.
To ensure you know improved efficiency one of the parameter that you can play with this
is porosity ok. Can you think of ways by which I can play with the porosity of the bed?
So, I have a packed column is filled with solids ok. Now what are the different ways by
which I can manipulate the porosity in the bed? You can change the. You can change the
dimensions of the particle ok. That is one way of doing it.
So, he is saying what he is trying to say let me just go back ok. So, this is a so, what you
seeing here is D p by D t where D p is the a particle diameter and D t is the you are you
know if you have a packed bed you can think about it as a diameter of the you know
column in which the particles are put right. Now so, what you are looking at is two columns
481
epsilon for spheres ok. I am already giving the second end this is epsilon for cylinders ok.
Now, when you say D p by D t is 0 what does it mean?
That, basically means you know I have a really large column ok, whose diameter is much
much larger than the diameter of the particle that are basically being used for making the
packed bed right ok. In such a case your epsilon is 0.0 you know is point 0.34 right. Now
this value is very close to, very close to if you have a spherical particle yeah. So, there was
a maximum random packing density right; that means, if I have a column I pour the
particles into the column. If all the particles in the bed right, if they are arranged in a
random you know configuration right ok.
So, it turns out that you know that maximum random packing density roughly of the order
of 0.64 right. Therefore, you know the closest that you can get if you really work with
ideally this should be 0.36 ok, but the closest that you get if you really take a very large
diameter column. And, if you put in the particle such that you know the particle diameter
is much much smaller compared to the diameter of the column right.
So, you can achieve 0.34, but you can clearly see, but if I go on you know if this if I go on
increasing the diameter of the particle it turns out you know this porosity basically
increases right. Does it make sense? Because you know let us think about a simple case
right. If I have a column I am going to fill in the column with exactly the same you know
diameter particles right. I am sure another in this case you know whatever volume there is
available for the liquid is very very little right ok. So, one of the way of doing this would
be basically I can play with the diameter of you know particle to the diameter of the
column.
482
But the other ways of doing this would be by using packing material different you know
types either. So, either you can work with regular particles, you can work with like say
spherical particles or cylindrical particles for example, or ellipsoids ok. You can work with
particles of you know different shapes or people have kind of designed different types of
packing material itself ok. What you are looking at is this a, a is the case of what is called
as the Raschig rings. These are basically you know yeah go ahead.
Now, when you say volume of the packed bed that is the volume of the so, if I have a like
say that I have a container a cylindrical pipe say that you know my the packing that I have
filled it is up to say length l for example. And, if the diameter is d right that you know pi
D square by 4 times L is your volume right that is the volume of the packed bed for me
ok. It is not just the volume of the just the packing itself when I say volume of the packed
bed.
So, therefore so in the example that we saw so, what I can do is I can take a column of
some dimension and I can fill it up with part of the different sizes that is one way of doing
it ok. So, other way I mean the same you know one of the ways of expressing exactly the
same experiment that I said right. I fix the column I take particles of different sizes right.
One of the way of expressing that would be a I basically express it as you know a ratio of
D P by D L right that is what was basically done yes yeah it will.
483
I think so, basically no what this basically tells you know as long as if you were to do you
work with any combination of you know your particle and the you know the column. So,
as long as you know this is actually I have taken from one of the books ok. So, the idea is
as long as you maintain this ratio you are the value would be typically close to 0.64. I mean
you know or the other way of doing this would be say you know I have like say a column
which is like say 100 centimeter diameter see that I have basically taking 0.1 centimeter
packing for example, ok.
Now, if I do an experiment where I you know kind of increase the size of this by twice.
So, I say I take 200 centimeter and if I also take you know so in both the cases you would
get a similar packing density ok. So, yeah so this something was a Raschig ring. So, you
can think about them as you know hollow cylinders right this is something called as a
Lessing rings ok
So, now, if you look at so, for example, if I look at this and this right in terms of the shape
they are very similar what do you think is in between I basically have another you know
basically kind of a rectangular this thing right. So, now if I look at compared these two,
this would definitely offer me a you know much more solid surface there is available for
any contact right. If I look at this and this case the fact that you know I have an additional
you know a region there. So, I basically have more solid surface available for any operation
that I want to do.
So, people have using Raschig rings, Lessing ring this something was a Berl saddle
something was a Berl saddles and this is what is called as a Pall ring, this is Pall ring these
two are Pall, Pall rings right and this is one something called as a Nutter ring and Nutter
ok. These are different names that people have you know these are used. These are all
industrially used packings of different shapes and as you can clearly see that you know it
is becoming more complex shape like if you go from a to f for example, the shape becomes
much more complicated right. And, the reason you know is that you know that will again
help you in you know playing with your you know solid you know fluid surface you know
areas right.
So, another way of so like we talked about you know this D p by D t as a parameter for
playing with your porosity ok. The other way of playing with the porosity is basically use
packing material of different you know types the Raschig rings, Lessing rings ok. So, if
484
you play with some of these things right. So, I have a table which is not the porosity I just
wanted I do not seem to have it ok. So, maybe I had something here I thought no I have it
here right.
So, if you look at Raschig rings right, it turns out the porosity that I can reach is larger ok.
So, if you take Berl saddles the typical porosity that you get is about 0.60 and again if you
use Berl saddles the typical porosity that you will be getting it will be in this range you
know compare it with you know spheres for example, you see for the case of spheres if I
really take a very small diameter particles ok.
The porosity that I can have is about 0.34, even if I increase the, if I play with the this
dimension ok. I can maximum I can go is up to 0.55, but; however, if you use different
shaped in a packing material I can see there you know basically there is a more porosity
right. A more porosity would be good for you know a better efficiency of the operation or
a better fluid you know particle contact yeah. So, these are the Raschig rings and the Berl
saddles are the shapes. So, basically if you go back here right this is your Raschig ring this
is your Berl saddles. Correct yeah.
Correct yeah. So, what you do is when you when you make a packed bed right as I said
packed bed is basically a column filled with a particles right. Now, the particles I have a
freedom of choosing whatever particles I am working with of course, if somebody is
working with like say a catalytic reactor, I would have to use a packing material there is a
a catalyst that you want to use for a particular reaction right. Or, if I want to basically use
an inert packing; that means, you know these are not participating in any reaction, if the
objective is only to improve the liquid fluid you know liquid gas contact for example, I
would like to have a much more surface area that is available right or more porosity for
example
So, in such cases what you do is you use one of these packing material. So, you know I
can basically play with the a porosity that will also help you in playing regulating the
pressure drop through the packed bed plus it will also ensure a efficiency increase in the
efficiency of the operation. So, any other ways of doing this so, I said you can play with
the particle in that and the column diameters and I can play with the porosity. I can play
with different objects of different shapes or you know different you know complex
geometries and I can play with epsilon that way any other yeah you have a question
485
So, these are basically let me go back here I showed that you know in this image it looks
like a sphere right. I have put in a spherical particle in the column instead of that I can use
a Raschig ring right. I can use a Lessing ring right. So, that was actually your column like
that, but with the right I can use that I can use Berl saddles ok. Any of the packing you
know shape that you saw I can actually replace that the spherical particle that you are
seeing here I can replace with any of these things. And, that will help me in regulating
your epsilon that will help me in regulating the solid fluid you know contact and it will
also help me in regulating the efficiency of the operation yeah. So, any other methods of
regulating epsilon? Yeah. No. Yeah.
No, when you say more porosity what does it mean to you? So, essentially the volume that
is available for the fluid is more right. Now, when the volume that is available for the
liquid is more ok, if I say then you know I am using a liquid gas operation you know of
course, the part of this you know the fractional voidage that is available there will be liquid
and the gas right. The more the voidage there is available there is more contact between
the gas in the liquid right. In that sense you know regulating porosity is a way of I mean
people go for more porosity because you know if you want to really improve the efficiency
of the operation I would definitely like to have more porous. Correct.
No, like I give an example where if I have a pipe right ok. I let the liquid and the gas flow
through the column. When people talk about this 2 phase flow in the columns people talk
about let me just go back when you talk about 2 phase flow let me just go back here ok.
So, when you talk about 2 phase flow to for example, say then I have a pipe. I want to have
liquid one or I say let us say fluid 1 and fluid 2 that is flowing through the column through
the pipe for example, ok.
What do you think would be the how the flow look like? Or, let us say two immiscible
liquids; liquid 1, liquid 2 there 2 immiscible say, but I would like to have a proper contact
would contact between the two. But, people talk about something was a stratified flow in
which I have liquid 1 flowing above maybe and liquid 2 flowing below depends on the
density of fluid. So, there is it two right that is one way of doing think about it. The other
way of doing this would be people talk about slug flow ok, in which you have you know
your liquid 1 could be here and you know liquid 2 will be outside right, this is liquid 1 ok.
So, it basically goes as right.
486
Now, if you look at these two cases the liquid liquid contact actually is happens here right.
You are only at the interfaces when there is liquid contact. If I look at this slug flow there
is a little bit better contact because you know I am able to break the stream into you know
maybe you know these slugs ok. But however, you know your liquid liquid contact is still
limited.
Now, if I you know what is the best way to ensure that you know there is a larger you
know kind of a interaction between the two fluids ok. When I say interaction not the you
know the interaction in terms of the molecular interactions, but I would like to have a
proper contact between the two ok. The best way of doing this would be you know using
these devices where you know you can ensure that you know there is a better interactions.
Why do you say the less porosity more than break? So, I mean say you cannot think of the;
so, in this case are you talking about porosity in this case. Why would you talk about
porosity in this case? Because if I look at this you know your entire column is filled with
a liquid, so, in that sense you know there is no pores as such right.
I would not say what do you really call it as break the fluids no I mean you know you see
you have just like you know just like a liquid flowing through a column you should
imagine that you know there is also a liquid flowing through a bed of solids ok. There are
channels available for the liquid to flow through ok. Now, as it flows through if I have 1
or 2 different fluids and you know it will have a you know it will basically give more
chance for the liquids to interact let it put the for in that that way yeah ok.
487
So, the other way of doing this would be looking at something called as a random packing
versus structured packing ok. So, if you take a column you just pour your you know
particles into the column ok. They pack in a particular way and typically if you again there
are ways there are actually kind of prescription as to how you should fill the packing into
your you know column. And, if you follow some procedures you will basically end up
with a truly random packing and other way of doing this would be what is called as a
structured packing in which I have a some predefined way in which you know the solid
material is basically arranged ok.
So, I am basically you know showing you the random packing of spherical particles here
and a structured packing of spherical particles here. And, as you can clearly see there you
know in this case the typically the porosities that you can achieve would be of the order of
0.36. Now, if you do a you know a random packing here, but however, if you look at this
case you know packing fraction is more if I assume a particular kind of packing you know
right your porosity is going to be right. So, basically the other way of you know playing
with your you know a porosity is go for random versus structured packing. People typically
do not use particles and then do a structure packing.
488
So, what is done in the ok so, I will come back to this point.
So, these are some kind of structured packing the people use. So, these are what you have
is you basically have structure which are either made of some polymer or it could be made
of some metal in which you basically you are if you look at these pictures right the solid
part right, the solid material that make the packed bed. There is a you know particular way
in which it is constructed right ok.
489
And, that forms the solid in the packed bed and whatever pores that you see here right that
there is a pore spaces through which the liquid basically flows through ok. So, and the idea
of use it going for random versus the structured packing is that you know the surface area
available for the liquid and the void space I can basically manipulate right like just the
point that I made earlier. So, I can basically play with the voidage plus also I can play with
the surface area that are available for the fluid contact by using these devices.
So, we have defined this parameter for sphericity and it turns out you know if you look at
when you are going to talk about this delta P is a pressure drop through the packed bed the
sphericity also becomes important parameter ok. We define more it is already it is defined
as s p by v p of a s p by v p of a sphere divided by s p by v p of the particle right. And, s p
by v p for the sphere is 6 by D p and s p by v p that is for the particle that is your sphericity.
And, if you see here if you look at spherical particles or you know or cubes or short
cylinders your sphericity is 1.
And, if you go for Raschig rings this is L is equal to D p that is your you know your if you
have this Raschig rings right that is your cylinder kind of thing that is your length and that
is your D p right. So, if L is equal to D p if it is a short cylinder sorry Raschig ring. So, if
you I can play with these dimensions are actually play with the porosity right 0.53 in 1.8
in one case you know 0.33 in the other case. Berl saddles typically have a sphericity of 0.3
and these are the kind of industrial packing the people use. And, people also work with for
490
example, you could you can have a column in which you know you may be working with
some kind of sand particles or you could be working with coal dust for example, in such
case again you know your sphericity would varying. And, there are ways by which as we
will discuss in the class there are ways by which I can basically calculate sphericity right.
Other term similar to the multi particle system there is a concept called superficial velocity
which is basically defined as the volumetric flow rate of the fluid through the column
divided by the cross sectional area that is your Q by A. It is also called as a EMPTY
TOWER velocity ok. Superficial velocity is also called as a EMPTY TOWER velocity.
This is the velocity of the fluid through the column when there is no packing that is your
Qth I know if have a column if have a way of measuring what is the fluid through the
column.
And, if I know what is the cross sectional area that Q by A is what gives you the superficial
velocity right also called as a EMPTY TOWER velocity. But, the average velocity of the
fluid through the pores is basically your superficial velocity divided by epsilon right,
because epsilon will tell you what is the cross sectional area that is available for the liquid
to flow through right that is the actual velocity or the average velocity of the fluid through
the pores yeah.
So, that is about it ok. I just wanted to briefly introduce you to the concept of packed beds
today. We are talked about what a packed beds where are they used. We talked a little bit
491
about packed bed internals that there is a liquid distributors, liquid collectors you know
and you know some concepts like that we defined a few parameters voidage the sphericity.
We talked a little bit about superficial velocity and we also talked about different types of
packing and different types of packing material. And, what are the different methods by
which people manipulate your porosities and you know the fluid solid contact area that is
available for them ok.
We will stop here. On Monday we will have we talk start a talk, we will start talking a
little bit about pressure drop through the bed ok. We will what we do is we are going to
take the knowledge of flow through pipes and you know we kind of apply that to the flow
through pack that is what we will do on Monday yeah.
Thanks.
492
Fluid Mechanics
Lecture - 42
Pressure Drop through Packed Bed
We will continue with flow through packed beds ok. So, any questions from the previous
class? Any doubts that you have anything that you did not understand? So, just to sum up
the previous class what we did was, we basically defined a few things in a packed bed
right. We said what is a packed bed, what are the applications where they are used, we
talked a little bit about what is called as a packed bed internals what does if you look inside
a packed column what does it have, we talked a little bit about you know what are called
as liquid distributors ok.
Which are which ensure that you know the channeling is kind of minimized and then also
what are called as liquid collectors which are basically placed at different locations you
know across the height of the packed bed. And the job of this liquid collectors is to ensure
that you know you collect all the liquid that is coming down the packed bed at a particular
location and then feed that into the distributor and distributor, does the job of uniformly
distributing the fluid across the cross section of the bed right.
493
We talked a little bit about different types of distributors as well something called as a
notch type, vein type you know and perforated you know channels can be used as
distributors. Then we went on to talk a little bit a little bit about definitions right. We talked
about what is a porosity ok. And then we talked about different ways of manipulating
porosity right. We said you could use a packing material of different shapes you know; in
general people use spherical particles or regular particles as packing.
Of course, you know you can go for more complex shapes we talked about using
something called as a raschig rings right and lessing rings pall rings berl saddles you know
things like that right. And what these different packing gives you is, it gives you a way of
manipulating the porosity within the packed bed, it will also gives you a you know way of
manipulating what is the total solid you know fluid area that is available for any you know
process that you are trying to do.
Then we talked a little bit about you know other definitions like sphericity, you know and
we talked about what is the definition you know and how does the you know sphericity
depend on different you know shaped objects that is where we had stopped right.
In today’s class what we are going to do is to look at. So, we are interested in looking at
pressure drop pressur drop through packed beds that is what we want to do you know for
today’s class.
494
So, that is a schematic of the packed bed say that this is a column whose length is like say
L and say that you know it could be any cross section. So, let us say for the simple case let
us take a cylindrical column, let us say that the you know the surface area or the cross
section area of the channel is something like s0that is a cross sectional area of the column
in which the packing is being placed let us say the let us write the packing ok. So, the
packing is something like this right some spherical particles, there are put in a fluid sorry
put in a column and you want to find out what is the pressure drop through the packed bed.
People typically think about two concepts, one is if I have a way of calculating what is the
frictional losses come that come because of individual particles in the packed column. If I
have a way of calculating that and sum that up for all the particles that form the packed
bed you have you can come up with a way of finding out what is the frictional losses
because of the presence of particles in the column. That is one way of doing it.
The other way of doing this is that you know actually flow the liquid is flowing through
the column. And it is flowing through the space that is available for it and basic it
essentially flows over the solid surface right you have a particles that form the bed and
there is a space that is available and when the liquid is flowing through it, basically flows
over a solid surface ok. If you have a way of calculating what is the total drag because of
the fact that you know the part the liquid is flowing over this solid surface that is another
way of you know thinking about you know calculating the pressure drop ok.
So, what we are going to do is to look at a very famous model which is what is called as a
channel model it is also called as a capillary model for you know which is which have
been developed to basically predict what is the pressure drop for the flow through the
packed bed that is what we are going to do ok.
The concept is fairly simple so, what we are going to do is we are going to say that you
have this packed bed right. Now if you imagine that the liquid is flowing through this
column and for a minute forget about the particles ok. Let us say that the liquid is flowing
through this column and of course, we said that you know the let us put it this way ok. So,
can I say that so, because of the fact that you have a packed bed. So, can I say that the
liquid is flowing through some channels is that a fair thing to say?
Right because you know of course, so wherever the liquid is flowing through so, I could
if I were to maybe trace like say let us say I am going to trace the. So, let us say there it
495
was it was wherever you know the fluid is flowing through right that could be the you
know it could be flowing like that can I say that. Of course, it is flowing through the gaps
between the particles right it is flowing through the gap between the particles right and of
course, it could so, happen there you know some other stream from some other location
may come and join right. It is it try to say that?
So, I can say that the liquid is essentially flowing through a channel; of course, these
channels are not straight channels right; these are not straight channels right. This is a what
are called as tortuous. These are tortuous channels because you know that is you know
liquid is basically not taking a straight path it is through basically going through a column
which can you know turn you know you know there could be change in the direction of
the fluid as the liquid flows to the column ok.
Now, the way we are going to work out this channel model is that, we are going to say that
I am going to replace a complicated packed bed with a very simple configuration where I
am going to say that there is a channel for the liquid to flow through and you have this the
hatched region that I am that I am drawing these are the solids that formed the packed bed;
of course, is a very simplified model right.
So, all I have to do is I would of course, I would have to correct for the fact that pipes or
channels or not straight. Of course, I would have to correct for this thing, but I can assume
that for the for the simplest case that I can say that you know I can assume a packed bed
496
to be consisting of channels through which the liquid is flowing. And I am going to assume
that they are straight channels and of course, later on I am going to correct it for the fact
that you know the channels are tortuous ok; that means, they are not straight that is what
we are going to do yeah.
Now whenever you want to do that ok. So, again one of the concept that we are going to
follow is we are going to go back and look up what we are done in the previous lectures
ok. Do you all remember this formula delta P is equal to? Sorry it is not 13 is a 32 then we
just check, if that is right yeah I remember this formula? That you guys are done what is
that any thoughts what is that? What is that? I am going to write down the name of this
equation Hagen Poiseuille equation right you remember this, what do you right?
Δ𝑃 32𝜇𝑉̅
=
𝐿 𝐷2
So, when we looked at flow through conduits; flow through circular pipes, you had
developed an expression for you know the average velocity through the packed bed. You
had developed expressions for what is the pressure drop ok. The pressure drop basically
goes as 32 times you know mu which is the viscosity of the fluid which is flowing through
the conduit and v bar is what is called as a average velocity with which the liquid is flowing
through the column and D is the diameter of the column through which the liquid is
flowing right.
Whenever you want to kind of look at analogy of flow between you know pipes of different
dimensions or different kinds what you do is, you would have to come up with a way of
actually relating the parameter that you that are involved to the actual quantities in a
particular flow problem
So, for example, say that you have you know flow through a circular pipe ok. Say that you
know I would like to use you know whatever equations that you are developed for a flow
through a circular pipe, if you want to extend that to flow through a rectangular channel.
So, one of the concept that people do is, they instead of so, wherever you have dimension
D which is the diameter of the column they replace that with what is called as a equivalent
diameter.
497
You replace D which is the dimensions of the pipe through which the liquid is flowing
through in this case to what is called as a equivalent diameter that is one thing you would
have worry about and this v bar is the average velocity with which the liquid is flowing
through the pipe.
In the case of packed bed we said that you know they something called as a v 0 bar which
is what is called as a superficial velocity or the empty tower velocity that is the velocity
before the liquid enters the column. Accordingly the velocity with which the liquid is
flowing through the column is v 0 bar by epsilon right that is the actual velocity right.
So, therefore, what I am going to do is, I am going to take this expression that has been
developed for pressure drop for flow through pipes and we know that this is only valid for
laminar flow conditions ok. I am going to use that, I am going to modify it and then yeah
you have somebody has a question yeah. What is that?
Oh yeah delta P by L I am sorry with that yeah that is true right. So, this is delta P by L
that is the pressure drop per unit length right that is equal to 32 mu v bar divided by D
square right yeah that is right.
Now, how do we go about calculating the equivalent diameter? So, if you have a simple
geometry like you know going from a circular pipe to a you know a square conduit or a
rectangular conduit, you know you can there are simple ways of getting what is the
equivalent diameter you know. But; however, for this case it is a little bit complicated
right.
What do you think that one should be doing to get what is the equivalent diameter? So, I
am just going to say that you know the length of the tubes is same as the length of the
packed bed and I am going to say that this diameter of the pipe through which the liquid
is flowing through is say D equivalent.
498
So, what do we do? So, for doing that we do two things one is we do what is called as a
surface area balance and number 2, what is called as a volume balance ok.
Now, let us say that there are n conduits or n tubes through which the fluid is flowing right
and let us say that is diameter of each of tube is D equivalent. Can I write an expression
for the total surface area of all the tubes that make up the column that make up the packed
bed ok? That is total surface area of all the channels or tubes or capillaries that make up
the packed bed. What would that be? That is going to be n times pi D equivalent times L
is it because you have tubes you are right your perimeter right times the length it is going
to give me what is the total surface area that is available for the fluid to flow through that
multiplied by n is going to give me what is the total surface area of all the channels put
together is that, yeah.
𝑇𝑜𝑡𝑎𝑙 𝑆. 𝐴 = 𝑛𝜋𝐷𝑒𝑞 𝐿
Now, let us think about expressing. What do you what can you say about the volume of
the solid particles are present in the packed bed? What can I say about the volume of the
ok?
499
Let me just go back here what is the volume of solid particles in the column, what is that?
It is going to be S 0 which is the cross section area times L that is going to give me the
volume of the packed bed that multiplied by 1 minus epsilon is going to give me the
volume of the solids in the is that. That is going to give me what is the total volume of all
the solid particle that you have in the packed bed.
𝑉𝑜𝑙𝑢𝑚𝑒 𝑜𝑓 𝑠𝑜𝑙𝑖𝑑 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒𝑠 = 𝑆0 𝐿(1 − 𝜖)
Now, if I divide this right you have you have a doubt somebody has a doubt? If I divide
this by V P which is the volume of the single particle, what is it going to give me? It is
going to give me the number of particles right is going to give me the number of particles
that are present in the bed, is it ok?
𝑆0 𝐿(1 − 𝜖)
𝑁𝑃 =
𝑉𝑃
So, now if I ask you what is the total surface area of all the particles in the bed is S 0 into
L into 1 minus epsilon divided by V P multiplied by S P right where S P is the surface area
of every particle that you have is it okay? Now, you know that when we talk about
sphericity phi s right it is 6 by D P divided by S P by V P right that is a definition of
sphericity therefore, I can write this S P by V P as 6 by phi s into D P. I can as a substitute
here S 0 into L into 1 minus epsilon multiplied by 6 by phi s into D P right yeah ok.
500
𝑆0 𝐿(1 − 𝜖) 6
𝑇𝑜𝑡𝑎𝑙 𝑆. 𝐴 𝑜𝑓 𝑎𝑙𝑙 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒𝑠 = 𝑆𝑃 = 𝑆0 𝐿(1 − 𝜖)
𝑉𝑃 𝜙𝑠 𝐷𝑃
Now, the fact that you know the channels are basically made up of the walls right. So, you
have these channels right these channels through which the liquid is flowing ok. Now the
fact that you know these channels are basically made up of the walls that arise because of
the particles present in the fluid ok. So, what I am going to do is, I am going to say that
the total surface area of all the channels in the column is equal to the total surface area of
all the particles that I have in the system ok. Therefore, what I can do is I can write n times
pi D equivalent times L that is the total surface area of all the channels in the column is
equal to S 0 into L into 1 minus epsilon into 6 by phi S into D P that is the total surface
area of all the particles that I have in the column ok.
6
𝑛𝜋𝐷𝑒𝑞 𝐿 = 𝑆0 𝐿(1 − 𝜖)
𝜙𝑠 𝐷𝑃
Therefore this is basically the surface area balance ok. All I am doing is I am calculating
what is the total surface area of all the columns that I have in the system and that has to be
equal to the total surface area of all the particles in the system ok. This equation is yeah if
you have any doubts we can discuss that the only ideally there is some small correction as
well right because you know what is happening is that you know if you have particles there
are in contact right there is a small contact area that is available I am basically neglecting
that ok.
But if you leave that you know if you assume that you know these are point contacts, if
you assume that you know there is no significant reduction in the surface area because of
these contacts. This is a good equation or a good approximation to say that the total surface
area of all the columns that you have in the you know in the in the bed is equal to the total
surface area of all the particles that you have the column right is that right.
All that we did in obtaining this right hand side was that we said that the total volume of
the bed that or the total volume of the particles in the bed is S 0 into L into 1 minus epsilon,
epsilon is the void fraction right. That gives me the total volume of solid divided by the
volume of the particles will give me the number of particles in the bed multiplied by S P
which is a surface area of every particle that basically give me gives me the total surface
area and all we are saying is the total surface area because of the column you know I should
501
be equal to the total surface area of them you know all the particles right yeah. No this S
P by V P this is we know that this is a characteristics of the particles that we using right
what is that?
D P is basically diameter of the particle right if you have a packed column if I am saying
that you know I have you know particle of some dimension ok. If I were to be working
with spherical particles if I say that you know your diameter of the particle is D P that is
your the D P here that is the diameter of the particle that are that make up the column that
is D P.
No the fact that I am using sphericity concept, it need not be for a spherical particle right.
It could be for any particle as long as you know the dimensions of the particle is D P that
is all yeah I mean all we are trying to do is. So, if I take this to be for a spherical particle
we know there is basically 6 by D P itself right, but; however, you know your S P by V P
which is 6 by phi s into D P it is going to be just 6 by D P for the case of spherical particle;
however, if you are working with particle of any other dimension you would have to work
with have in this pre factor which is sphericity that is all yeah. Now, what we will do is we
will look at the volume balance.
Now what we are going to do is, we are going to say that we talked about the total surface
here that is available for the liquid to flow right what about the total volume? Volume of
all the conduits put together what is the volume of all the conduits put together that is going
502
to be equal to n times right D equivalent square into pi divided by 4 that is you know that
is the right multiplied by your L right there is a total volume of. So, that is the volume of
the cylinder right you pi D equivalent square divided by 4 multiplied by L is going to be
the volume under the pipes itself multiplied by n which is the number of pipes that I have
that basically is the total volume of all the conduits put together right.
And of course, that has to be equal to the total volume of the liquid that you have in the
system right that is going to be S 0 into L into epsilon right S 0 is your cross section area
L is your length and epsilon is the porosity and this is the volume balance ok. All that we
are doing is we are saying that the total volume of the liquid you know in the system or
the liquid is flowing through the system is basically equal to the volume that is available
for them to flow in this case, the total volume that is available for them to flow is going to
be the number of channels multiplied by the volume of the each channel right ok.
2
𝑛𝜋𝐷𝑒𝑞 𝐿
𝑉= = 𝑆0 𝐿𝜖
4
Now, what you are going to do is you are going to use this equation and the equation that
we wrote up here that is 1 and 2 you are going to use that, and you are going to work out
and then get what is your D equivalent ok. So, let us do that.
So, n times pi D equivalent times L was equal to S 0 into L into 1 minus epsilon divided
by phi s into D P into 6 right something like that that is your yeah can you work with 1
and 2 and let me know what is your D equivalent? Can you do it in your notebook tell me
what is the d equivalent? Ok.
So, what you what you can do is, I have this equation that relates the this volume imbalance
right or I can do is I can substitute for n from equation one I can substitute for n from
equation 1. So, what I have done is. So, n I am going to write it as S 0 into L into 1 minus
epsilon into 6 divided by phi s into D P multiply I have right yeah divided by pi into D
equivalent into L right that is your n right multiplied by you have this term that is pi D
equivalent square divided by 4 into l should be equal to S 0 into L into epsilon say it is not
gets cancelled on both the sides right.
So, if you work it out it turns out that you know your this is D equivalent it turns out that
you know your D equivalent is going to be two by third into phi s into D P into epsilon
503
divided by 1 minus epsilon ok. You can work it out right therefore, your D equivalent is 2
by 3 into phi s into D P into epsilon divided by 1 minus epsilon.
2 𝜖
𝐷𝑒𝑞 = 𝜙𝑠 𝐷𝑃
3 1−𝜖
So, you have been able to get hold of a expression for D equivalent which I can further use
and then substitute in the Hagen Poiseuille the equation and calculate what is the pressure
drop right. So, let us go back to that equation right.
So, your delta P by L was equal to 32 mu V bar divided by D now D equivalent square
right. Can you substitute for D equivalent square and then just ? So, what is done is then
you know. So, we of course, we said that you know this is actually applicable only for
straight conduits right or the straight pipes. So, because of the fact that you know your
channels are not straight, you would have to introduce a factor called lambda you know
lambda 1 which is what is called as a tortuosity factor ok.
So, therefore, I am going to multiply this everything by lambda 1; lambda 1 is a parameter

which is called as a tortuosity factor this basically takes into account the fact that the pipes
are not straight they are tortuous in nature ok. And there are we are going to people have
done experiments to determine what this lambda is we will talk a little bit about that in the
you know in the next few minutes ok. As mu and of course, this is not going to be is V bar
I would have to replace that by v 0 divided by epsilon right that is your the average velocity
504
right and I would have to substitute for what is D equivalent ok. So, what you can do is
that is 2 by 3 into phi s D P square into epsilon divided by 1 minus epsilon right that is
your not D square right that is just a D P right. And of course, there is a lambda 1 that is
the tortuosity factor ok. So, if you work it out what you actually get is something like this.
You get 72 times lambda 1 times v 0 bar times mu into 1 minus epsilon square divided by
epsilon cube into phi is into phi s square into D P square yeah that is what you are going
to get ok.
Δ𝑃 32𝜇𝑉̅ 𝑉̅0 𝜆1 72𝜆1 𝑉̅0 𝜇(1 − 𝜖)2

= [ 2 ] 𝜆1 = 32𝜇 [ ] 2 =
𝐿 𝐷𝑒𝑞 𝜖 2 𝜖 𝜖 3 𝜙𝑠2 𝐷𝑃2
[3 𝜙𝑠 𝐷𝑃 (1 − 𝜖 )]
All that you have done is you basically simplified it further right you substitute it for what
is D equivalent in terms of 2 by 3 phi s D P into epsilon divided by 1 minus epsilon square
because you have 1 epsilon that epsilon squared here and there is 1 epsilon here. So,
therefore, you have this epsilon cube with the denominator. And of course, you have phi s
square into D P square and mu and v 0 bar that comes from here right and one minus
epsilon square that comes because of the fact when you have 1 minus epsilon in the
denominator right and of course, if you work it out. So, that is going to be your 72 times
lambda 1 ok.
And what people have done is they have done experiments in which they have they have
a way that is delta P by L right. If you take columns of you know in which you maintain
different delta P is you know and then you vary your delta P by L or if you take columns
where you know keep your delta P constant. And if you vary L right you vary this when
people have done a lot of experiments by using packed beds of different materials and they
have kind of found that you know this 72 times the lambda 1 which is a pre factor, it
apparently it is about 150 which leads to lambda 1 of the order of 2.1 and this 2.1.
So, one way to think about this what is called the tortuosity factor one way to think about
this would be that the fact that you know the channels are not straight of course, the fluid
is going to spend more time in the packed bed because you know it has to pass through a
greater length of the tube ok.
And the fact that you know this is slightly greater than 1, it tells you that you know the of
course, they are a tortuous and you can think about you know this being the typical time
505
that the fluid spends in the column would be roughly of about 2 times you know what it
would spend if the channels are straight ok. This is roughly is what you can see and of
course, this depends on the kind of flow conditions ok.
So, so, far what we have done is we have assumed Hagen Poiseuille equation which is
valid only for laminar flow conditions ok. So, now, similarly what we are going to do is,
we are going to develop a similar expression for the case where you have a turbulent flow
condition ok. Can we do that any doubts with this? So, this tortuosity factor there is no
way of directly measuring what it is ok. The only thing that you can do is you formulate
the problem in such a way that you know you are under this laminar flow conditions, that
is no your Reynolds number is less than one for the flow through packed bed problems
right.
And then what you do is you carry out experiments if you have a way of measuring delta
P and if you have a way of varying say v 0 bar if I know what is epsilon and you know
your D P and everything. Therefore, what you do is, if you were to plot how this delta P
by L varies as a function of v 0 bar what you would do is, you will get a straight line that
is passing through the origin right and from this slope they basically back calculate what
this lambda 1 is that is how it is done ok.
And this is based on a set of experiments done on columns of different you know packing
material you know different length columns and stuff like that and this tells you something
about lambda tells you something about how torturous the channels are ok; it turns out that
you know the lambda that you get for this case is 2.1. And however, if you go to the
turbulent condition the lambda that you get is much higher ok. So, case of the order of
something like 68 or something like that ok. We looked at a number a little later therefore,
this lambda 1 correction is necessary because of the fact that you know the flow is not
through straight channels any questions?
So, it is almost time up we will stop here. So, I will take a few questions if you have some
ok. We will continue with you know doing something very similar for the cases under
which your flow through you know the tubes are under laminar flow conditions any
questions you have a question both of you yeah yes any questions yeah no. You can tell
me you know let us discuss that now I mean if you have any question that you have you
should you should just raise and then figure out we do you know understand that no is the
506
concept clear of the of the capillary model or the channel model. It is a simple concept that
you know you are basically assuming that you know the flow through packed bed indeed
can be thought in terms of flow through channels of course, these channels are not straight
they are going to be different you know streams you know joining from different you know
locations of the packed bed ok
However for the sake of simplicity we assume that there are straight channels we said there
are n channels of exactly the same length as the length of the packed bed. And we did very
simple volume balance and the surface area balance that gave us a way of estimating what
is the equivalent diameter or you can think about this as a equivalent diameter of the
conduits through which the fluid is flowing and then we substitute that in the Hagen
Poiseuille the equation. And then came up with the expression for what is the pressure
drop through the packed bed it turns out to be one fifty times mu 0 v 0 bar into mu one
minus epsilon square divided by epsilon cube into phi s square into D P square that is the
that is where we stop here today we will continue with the tomorrow.
507
Lecture - 43
Tutorial - 06
Hello, everyone. Today, we are going to solve a problem on settling of particles.
So, the problem is we have spherical particles of density. Particle density given as 2000
kg per meter cube and these are in the size range of 20 to 100 micrometers and we have a
stream of water flowing upwards in a vertical pipe of diameter. So, this is the schematic.
We have water flowing upwards and we have particles and the density of water is 1000 kg
per meter cube with viscosity 0.001 Pascal second.
Now, we are asked to find the maximum water velocity which is required to ensure that
no particles of diameter of size greater than 50 micrometers are carried upwards with the
water which means that particles of size greater than 50 micrometers have to settle down.
So, the first approach to solve this problem is to find the Reynolds number of the particle.
So, the Reynolds number of the particle is given as diameter of the particle density velocity
by viscosity so, but the unknown here is the velocity of the particle.
So, based on the Reynolds number of the particle we can divide the setting of the particles
into three regions. The first region is the Stokes region or laminar setting and the second
508
region is the transition and the third region is the Newton’s; Newton’s regime. So, in this
region the Reynolds number of the particle is less than 2, here the Reynolds number lies
between 2 to 500, in Newton’s region the Reynolds number of the particle is greater than
500.
So, we assume that particle settles in Stokes regime. So, in Stokes regime the terminal
velocity of the particle is given as. So, now, we have to find the settling velocity of the 50
micrometer particle. g is the acceleration due to gravity which is 9.81 meter per second
square and the diameter of the particle is 50 micron, rho p is the density of the particle
which is 2000 kg per meter cube and the rho f is the density of the fluid which is 1000 kg
per meter cube divided by 18 times viscosity of the fluid which is 0.001 Pascal second.
So, the terminal velocity for the 50 micrometer particle comes out to be 1.36 into 10 power
minus 3 meter per second this is for a particle of 50 micrometer size. So, since we assume
that the particle settles in a Newton regime, the Reynolds number of the particle should be
less than 2. So, we will check if the Reynolds number of the particle is less than 2 to make
sure that our assumption is correct.
So, now, we will calculate the Reynolds number of the particle for a 50 micrometer
diameter of the particle. So, 50 into 10 power minus 6 multiplied with V is the settling
velocity which is 1.36 into 10 power minus 3 and density is 2000 viscosity is 10 power
minus 3 sorry, this is the density of the fluid. So, it is 1000 kg per meter cube and we get
509
Reynolds number as 0.068 which is less than 2. So, our assumption of Stokes regime is
correct.
So, now, in the question we are asked to calculate the maximum water velocity with which
we should pump the water, so that particles of size 50 and less than 50 should settle down
should sorry, it particles of size 50 and less than 50 should be carried up with the water.
So, if we pump the water with this velocity which is the settling velocity settling velocity
for 50 micrometer particle. So, if we pump the water with this velocity all the particles of
size less than 50 will be carried upwards with the water and particles of size greater than
50 micrometers will be settled down because this is the terminal velocity of the 50
micrometer particle.
And, if you see the terminal velocity is directly proportional to the square of the particle
diameter. So, the particles of greater the particles of diameter greater than 50 micrometers
will have a settling velocity of higher than settling velocity greater than settling velocity
of 50 micrometer particles. So, that when we pump the fluid with this velocity all the
particles of less than 50 micrometers will be carried upwards with the water and those
particles which are heavier in size which are greater than 50 micrometer in diameter will
be settled down.
Thank you.
510
Fluid Mechanics
Lecture - 44
Pressure drop through packed bed continue
So what were you looking at in the last class is, we looked at a concept called capillary
model or tube model right which is a methodology, which has been developed to basically
correlate delta P by L which is the pressure drop per unit length of the packed bed to
properties of the packed bed ok. In terms of V 0 bar which is the superficial velocity right
of the liquid that is flowing through the packed bed mu which is the viscosity, 1 minus
epsilon whole square divided by phi s square into D p squared times epsilon cube right.
That is what we had developed was that?
Yeah. So, what we said is that you know when we the way this was done as we and ah.
So, we are basically looking at the case where you are looking at laminar flow conditions
right laminar flow conditions and the starting equation was delta P by L is equal to 32 mu
V bar divided by D square right and then what we did is, we use this capillary model and
then we did the surface area balance and the volume balance and we said your you would
have to replace your D by D equivalent which was something like 2 by 3 into phi s into D
p into epsilon divided by 1 minus epsilon right.
511
Yeah and then this V bar which is the average velocity with which the liquid has flowing
through the conduit right that we would have to replace by V 0 bar divided by epsilon right
that is the modification that we did. And if when we work it out this the constant here it
came out to be something like 72 times lambda right lambda 1 and we said that you know
people have done a lot of empirical lot of you know experiments and they found out that
you know this you know if you do an experiment of delta P by L versus you know if you
have a data of delta P by well L versus V 0 bar right and if you are you know in the laminar
flow condition, that is if you ensure that your Reynolds numbers are less than 1 you would
get a straight line that is passing through the origin and if you know the other factors right
this is basically your slope is going to be you know this time some constant right and
people have found that you know that is of the order of 150 which basically gives you your
lambda 1 as 2.1.
So, I was trying to discuss that yesterday said that this 2.1 accounts for the fact that the
liquid is not flowing through straight pipe, but the pipes are tortuous ok. In essence how
high this number is you know if it is much larger than you know 2.1 then will be much
more tortuous ok, if it is closer to one you know you can say there it is like not too tortuous
that is the you know you can think of. Now what you can do is I can actually recast this
equation into something like this delta P by L is equal to I have this V 0 bar and mu at the
top and at the bottom you can write it as kappa sorry K and K essentially from if you look
at look this up K would be something like phi s square into D p square into epsilon cube
divided by 150 into 1 minus epsilon whole square right.
So, I am basically I have taken this expression I just forgot to mention that this equation is
what is called as a Kozany Carman equation that is because these people developed this
equation. See these are the developments that happened in you know 1930s and 40s. The
fact that you know these models are still used in predicting you know pressure drop
through packed bed it is a testimony to the fact that these are very well thought out models
and typically when people do this modeling in chemical engineering; that means, you take
a particular process and then you kind of if you want to put across a very simple
methodology to evaluate like say in this case the pressure drop per unit length the concept
of you know the fact that you know this packed bed consists of you know tubes, you know
and the concept of equivalent diameter ok.
512
So, this not easy to come up with right and the fact that they were able to do it you know
in the 1930s and 40s and people are using these correlations even to date in the industry
that is because the fact is a very well thought out model and any model that kind of you
know stands for over 10 years is considered to be really good model, but the fact this is
been there for more than you know 50- 60 years; that means, you know this is one of the
really nicely developed a methodology for basically getting this delta P by L ok. And this
expression where delta P by L is equal to V 0 bar into mu divided by K and this K is what
is cause of permeability ok. And if you look at this expression this K bar has units of meter
square right because you know your phi s is a is non dimensional D p square has units of
meter square and other things are all you know non dimensional right.
Δ𝑃 𝑉̅0 𝜇
=
𝐿 𝐾
So, so, and this equation is also called as Darcy’s law which essentially states that you
know you are the velocity with which a liquid is flowing through a porous media is
basically proportional to delta P which is the pressure drop and it is inversely proportional
to mu which is the viscosity of the fluid ok. And if you go back and look up people in other
disciplines especially if you look at civil engineering you will always come across cases
where you know people are looking at porous rocks, you know people are looking at flow
through you know you know sand for example, ok.
So, everywhere you know people kind of use these the flow through packed bed concept
really you know extract some meaningful parameters. We also want to look at cases where
we are going to say that these expression that we have developed this Kozany Carman
equation you can also use it for basic characterization of the particles itself which we are
going to look at a little later ok. That is for the laminar flow conditions if you want to look
at the you know the turbulent flow conditions.
513
So, this is the Kozany Carman equation that we would have developed as I said its only
for laminar flow conditions. So, if you want to develop a similar kind of you know
expression for delta P by L for the turbulent flow conditions again we like we started with
you know Hagen Poiseuille equation you will again have to start with a similar expression,
but now for the turbulent flow conditions right.
So, it turns out you know delta P by L for the flow of liquid through conduit in the turbulent
flow condition it goes something like f that is a friction factor times rho that is the density
of the fluid and V 0 bar or say V 0 just put as V 0 at this point V 0 square divided by just
want to where is that anyone remembers that is D right ok. That is your expression for a
pressure drop per unit length through the pipe turbulent flow conditions that is a friction
factor density of the fluid, average velocity with which the fluid is flowing through the
packed bed and D is your diameter of the conduit right.
Δ𝑃 𝑓𝜌𝑣̅ 2
=
𝐿 𝐷
So, again we like into exactly the same exercise you replace your V bar by V 0 bar divided
by epsilon that is the velocity with which the liquid is flowing through the packed bed right
that is the actual velocity right that is the average velocity which are the critical onto the
pipe and then D we are going to replace by D equivalent right can you just do that and let
me know what you get for delta P by L for the turbulent flow conditions. It is for turbulent
514
flow conditions right. So, it is going to be delta P by L you are going to have f there rho
and your V 0 bar is V 0 bar square divided by epsilon square at the bottom and your t
equivalent is going to be 2 by 3 into phi s into D p into epsilon divided by 1 minus epsilon
right.
Therefore that becomes f into rho into V 0 bar square and this 1 minus epsilon goes up
divided by you have phi s D p into epsilon cube at the bottom and that is going to be 3 by
2 right is that. That basically goes something as I have missing 2 somewhere oh there is a
2 here this its actually 2 times f rho V bar square by D ok. So, there is a 2 here. So, this
this gets cancels. So, you essentially get 3 times f rho V 0 bar square 1 minus epsilon
divided by phi s D p into epsilon cube.
2
Δ𝑃 3𝑓𝜌𝑉̅0 (1 − 𝜖)
=
𝐿 𝜙𝑠 𝐷𝑃 𝜖 3
So, that is and again you know we do not know what this f is right again f depends on your
Reynolds number and stuff like that ok. Again people have done a lot of experiments and
they have found that you know this 3 times f factor is of the order of 1.75 again this is
based on several experiments that people have done on packed beds of you know different
materials they fill ah cylinders in the packed bed, they fill bull saddles you know spherical
balls and then you make sure that you know your Reynolds number conditions such that
your Reynolds number is greater than 1000 right.
So, we said for the flow of fluid flow of fluid around particle, we said you know anything
greater than 1000 up to about 200,000 was the last number range for which the conditions
are turbulent right, but therefore, ah. So, now, this 1.75 if it turns out that if you go back
and look it up.
515
So, say that you know I have a Reynolds number condition of the order of like say 10000
ok. Now what you do is now I know what is your D equivalent you use that and you back
calculate what is the Reynolds number that corresponds to flow of the fluid through the
pipe ok.
Now, it turns out you know the Reynolds number would be of the order of 4000 right.
Now from this you can go back to what is called as a Moody’s chart right m o o d y
Moody’s chart right you can go back and look up you know corresponding to a Reynolds
number of 4000 what is your friction factor and that friction factor comes up to be
something about 0.01 all of you know Moody’s chart right its basically your Reynolds
number your friction factor versus Reynolds number plot right. I think you are going to
have an assignment where you are supposed to write something about this if I am not
wrong ok.
So, so, the point is now if f I can go back to f and I can say that your lambda 2 which is
the tortuosity factor now it is actually 3 times 1.75 that is a the factor that I have got from
experiments divided by 3 times f. So, therefore, if you do it with 3 times 0.01 this comes
to be 58 much to0 larger right. We character this lambda 1 to be 2.1 for the laminar flow
conditions, lambda 2 which is the tortuosity factor for the turbulent flow conditions it is
about 58.
516
So, therefore, so, the fact now this lambda two is much much larger; that means, to say
that you know it is not that the pipes are more tortuous it is just that the pipes are as tortuous
as the laminar flow conditions; however, because of the fact that there is a turbulent flow
condition what is going to happen is there is going to be a lot of changes in the area you
know; that means, you know in some channels the fluid is going to be you know flowing
in a channel of particular area and there could be expansion of the liquid at some locations
plus the fact that you know liquid is going to turn the direction in which it is flowing all
that leads to much more larger kinetic energy losses that leads to a larger you know you
can think about that in terms of the larger value of tortuosity yeah go ahead.
I just said that you know. So, basically ok. So, when we looked up here right. So, this is
when I when we did this right 3 times f rho V 0 bar square 1 minus epsilon divided by phi
s D p and epsilon cube that is for a straight pipe right, but of course, you know I would
have to have a tortuosity factor lambda 2. So, what I mean these are all based on some of
the experiment that people have done look after this these you know development in terms
of you know correlating your delta P by L two different parameters were developed now
you know what people did is, they did a series of experiments under different Reynolds
number conditions and then they matched this theory with the experiment and then that is
how this 1.75 and you know 150 numbers have been obtained.
Yeah. So, that is what 3 f times lambda 2 is 1.75 and therefore, that is why you know in
the next this thing what we did is I want to estimate what is this lambda 2 value therefore,
you know it is 1.75. So, if you assume a particular you know Reynolds number you know
it will be different right depends on you know what is the Reynolds number condition that
you have in the system; that means, you know.
So, therefore, if I were to take a you know Reynolds number of the order of like say 1200
ok. Of course, you know your lambda 2 that you all have got to have been lower right that
is because you know again the Reynolds numbers conditions are such that you know
maybe kinetic energy losses or in that cases are much lesser compared to the case of ten
to the power four Reynolds number so on.
So that is the this expression is what is called as a I will go back to the next expression.
So, this delta P by L which is 1.75 into rho V 0 bar square into 1 minus epsilon divided by
epsilon cube into phi s into D p right that is the expression that we developed right in the
517
previous; in the previous slide right that is here rho V bar square 1 minus epsilon phi s D
p epsilon cube and multiplied by 1.75, this is what is called as a burke b u r k e plummer
p l u m m e r equation again this is the two folks which came up with this expression you
know again in the 1940s . Now so, we have kind of developed expressions for delta P by
L for laminar flow conditions and turbulent flow conditions and someone called ERGUN
e r g e n or g u n Ergun’s equation.
So, what he did? What he did is he said the delta P by L for the entire you know range it
basically is delta P by L for the laminar flow conditions plus delta P by L for the turbulent
flow conditions ok. Again the concept is this right you know if you look at a Reynolds
number right this is the you know inertial force divided by viscous force right that is your
Reynolds number.
Now, the pressure drop that you have when the Reynolds number is lower it is kind of
dictated more by the viscous flow viscous term and in the case where in a where you have
you know high Reynolds number, it is kind of dictated by the inertia term right you know.
So, therefore, all he did was he clubbed this Kozany Carman equation and Burke Plummer
equation and he said that you know if you really are interested in the pressure drop per unit
length for the entire you know regime of flow all you have to do is sum up these two
equations ok.
518
So, therefore, Burke Plummer’s equation go something like it. So, 150 into mu V 0 bar 1
minus epsilon whole square divided by phi s square into D p square into epsilon cube plus
1.75 rho V 0 bar square right 1 minus epsilon divided by phi s D p into epsilon cube that
is the overall the you know that is the Ergun equation which basically captures the pressure
drop per unit length for the entire flow regime going from all the way from laminar to
turbulent.
Δ𝑃 Δ𝑃 Δ𝑃
=( ) +( )
𝐿 𝐿 𝐿𝑎𝑚𝑖𝑛𝑎𝑟 𝐿 𝑇𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡
2
Δ𝑃 150𝜇𝑉̅0 (1 − 𝜖)2 1.75𝜌𝑉̅0 (1 − 𝜖)
= +
𝐿 𝜙𝑠2 𝐷𝑃2 𝜖 3 𝜙𝑠 𝐷𝑃 𝜖 3
So, the point that I want to say is that you know all this basically assumes you know if you
look at laminar flow conditions, this is the term which basically dominates the pressure
drop, if you go to the turbulent flow condition this is a term that dominates the pressure
loss. However, if you have any you know Reynolds number in between, you know it there
is a contribution of both is the both of these terms to the pressure drop. Ah Now if I look
at this expression right yeah go ahead.
So, I was I mean I just took a some case that I was actually given in the book ah, but the
idea is this right you assume a. So, if I want to get what is the f right which is the friction
factor, if I want to use a Moody’s chart I should know what is the Reynolds number right
for the flow through pipe right. Now what you do is I want to get this Reynolds number.
So, what you do is, if I say that you know the flow is occurring in turbulent flow conditions
in the packed bed I find out what is the Reynolds number for that condition ok.
From that I go to this you know the pipe concept and I back calculate what is the Reynolds
number is, you know if the if the fluid at similar Reynolds number would to would flow
through the pipe and that turns out to be know in this case about 4000 and I went to the
Moody chart Moody’s chart and for the Reynolds number you know again you know it
depends on your smoothness and stuff like that ok. Typical value of f for that case you
know is about 0.1 that is how it was right.
Now, if you look at this Burke Plummer equation right what you can do is I can actually
rearrange this a little bit ok. So, if I say delta P by L right I have delta P by L is equal to
519
you know 1.75 rho V 0 bar square into 1 minus epsilon divided by phi s into D p into
epsilon cube what I can do is I can rearrange this a little bit ok.
What I will do is I will have delta P here I am going to get this V 0 bar you know to the
other side and let us say V 0 bar square right and then what I am going to do is I have rho
here I am going to get rho was to this side and that is equal to 1.75 into 1 minus epsilon
divided by phi s, I am going to take l to the other side I am going to write as L by D p and
so, I have what I can do it let me just do this way I am I have taken also epsilon square to
the other side I am going to take V 0 bar square divided by epsilon to this side right.
So, therefore, I am going to end up with something like this 1.75 into 1 minus epsilon into
phi s into epsilon into L by D p ok. Now this is the pressure losses and this is the can you
identify this head losses right the correction head losses right. So, therefore, this basically
now that is divide by 2 right.
So, now, what you do is if I substitute for typical values of you know ah. So, what can you
say about L by D p? L is the length of the packed bed and D p is the diameter right and I
can say it what is that? It can it can tell you something about the number of layers of
particle that you have right if you had if I have a total bed of height L and D p is your
diameter, it will give you typically it will give you the number of layers of particle that
you have right.
Now, if I. So, now, if I take some value of epsilon for example, if I take epsilon to be
something like say 0.4 and if I substitute in this. So, what I get is a value which is
something like have it or not 5.25 this is for L by D p is equal to 1; that means, this 5.25
is the. So, basically what you are doing is. So, in basically turns out to be delta P by rho V
0 bar square by 2 right is of the order of 5.25 ok that is for every layer of for every layer
because I am taking L by D p as 1 it will basically telling you what for every layer of the
particle that I have in the you know packed bed, what is the pressure losses compared to
the head losses right. And the fact that you know this pressure drop is 5.2 times you know
this head losses; that means, the kinetic energy losses in the case of you know the turbulent
flow conditions are much much larger.
So, if you were to do a similar kind of analysis for flow through pipes, you will you know
you will see that you know this delta P this ratio will be much less than 1; 1 or less than 1
the fact that it is a much larger number that basically tells us that the kinetic energy losses
520
in the packed bed because of the fact that there is a change in the cross section area of the
conduits through which the liquids are flowing plus the fact that you know there is a change
in the direction of the fluid that leads to much higher kinetic energy losses in the case of
flow through packed beds. So, that is that is all what I wanted to say.
Now, it turns out that any questions so, far. So, we you can also think about doing a now
that we know what is this D equivalent right which is you know 2 by 3 into phi s into D p
into epsilon divided by 1 minus epsilon right. Now if you take this expression let us say
that you know I have a packed bed and see there is no I have filled it up with spherical
particles and say that there is a random you know packing and we know that you know the
maximum no random packing density is of the order of 0.64 right.
Now, if I substitute that value back here right and if I say they know I am working with
spherical particle your phi s is 1. So, you are 2 by 3 into you know D p times 0.36 divided
by 0.64 right that is your 1 minus you know that is a packing density. So, it turns out there
is the number that you get is of the order of 0.375 D p. This is a good kind of exercise to
do what it basically tells you is there you know that the diameter of the channels through
which the liquid is flowing these are much smaller than the dimensions of the particle itself
ok. In this case is of the order of 0.34 0.38 or something like that right that is the; that
means, the diameter of the channels through which the liquid is flowing through in the
521
packed bed it is you know typically it is always less than the diameter of the particle that
make up the column.
So, that is the it does make sense right because it you know of course, it has to be less than
because you know the liquid is basically flowing through a gap that is available between
the particles right therefore, it is expected that you know if this has to be less than the
diameter of the particle. So, maybe I am just going to end whatever we were discussed on
the packed bed looking at some applications.
Of course I mean in terms of the applications, we have said you know they are kind of
used in you know all kinds of chemical you know unit operations plus they are used in
chemical industries, they are used in you know as adsorption columns, you know they are
used in distillation, you know they used in you know reaction engineering in terms of
catalytic bed reactors and things like that right.
But flow through packed bed concept also has been exploited for actually what is called
as a particle characterization itself. So, couple of applications I am going to talk about one
is a determination of particle size other one is determination of what is called as a specific
surface area of the particle right. Again you have a assignment coming up right the next
assignment is going to be on particle characterization and stuff like that. So, you will have
to write up something about you know different methods of particle characterization. So,
this could be one of the method.
522
So, basically what you. So, in doing these particle characterization, what is actually
exploited is that if you look at you know this Ergun equation right if you look at Ergun
equation. So, this is the Korzony Carman equation and that is the Burke Plummer.
And if you look at this expression your delta P by L right it goes linearly you know with
the superficial velocity and it goes as the square of the velocity you know in the turbulent
case, but if you work with you know if you work with conditions where Reynolds number
is less than 1, I can actually neglect this right therefore, if you were to plot therefore, if
you were to plot delta P by L if you were to do an experiment where I measure the pressure
drop per unit length across the packed bed as a function of the superficial velocity, what I
should get is I should get a straight line passing through the origin and the slope of this
line is going to be 150 times mu eps 1 minus epsilon square divided by phi s square D p
square into epsilon cube.
So, if you work with you know conditions where your Reynolds number is less than 1. So,
this is what should happen. Now if I have you know a way of you know measuring delta
P by L in my experiment and of course, I can also if I have a way of measuring what is the
flow rate typically what is done is if you have a packed bed filled with packing material
what you do is, you basically have two pressure tappings one at the top one at the bottom
you connect it to a manometer that basically gives you what is the pressure drop right you
can get this. I know what is the length of the packed bed that is your L.
523
I have a way of getting this and once the liquid actually comes out of the packed bed I can
basically collect it right I can measure Q which is the volumetric flow rate and volumetric
flow rate divided by the cross sectional area will give you what is the superficial velocity.
So, you have a way of measuring this and this and then once you estimate the slope if you
have some way of measuring your epsilon we kind of discussed up a couple of methods
right and of course, if you are working with the spherical particles your phi s is going to
be 1, from the slope I can actually directly get what is the diameter of the particle ok. That
is one of the applications of you know Korzany Carman equation where if you do
experiments under laminar flow conditions I can actually use this delta P by L versus V 0
bar plot to get the dimension of the particle.
Now, again if you again will go with again the same expression 150 times mu V 0 bar
sorry that is wrong right that is your that is one minus epsilon whole square. Now if you
look at this right your phi s square into D p I will go back to the definition of sphericity
where we said your phi s 6 by D p divided by you know S p by v p right that is what we
did we said therefore, your phi s into D p is going to be 6 divided by S p by V p right where
s p is the surface area of the particle and V p is the volume of the particle right.
Now, I can write this as 6 divide by S p into volume I can write it in terms of the mass the
density. Therefore, it becomes multiplied by density of the particle divided by mass of the
524
particle whole square right that and your S p by m which is the this surface area of the
particle per unit mass that is what we define it as a specific surface area right.
So, therefore, it goes as 6 divided by say s may be specific you know surface area times
rho p whole square right is it ok? No its no there is no whole square right sorry I was
talking about this. So, there is no right is it ok? So, basically I have kind of rework this.
So, essentially again it is the same concept right. So, all you are doing is you are instead
of phi s square into D p square I am basically expressing that in terms of the density of the
particle and the specific surface area again I can plug it back into this equation.
𝜙𝑆 = (6/𝐷𝑃 )/(𝑆𝑃 /𝑉𝑃 )
6 6
𝜙𝑠 𝐷𝑃 = =
𝑆 𝜌
( 𝑃𝑚 𝑃 ) 𝑆𝑆𝑆𝐴 𝜌𝑃
So, once I do that if I were to do a again a similar experiment of course, I can think about
you know this V 0. So, what people do is the moment you do experiments in a column of
a particular diameter right say this is d right what you do is I can basically express this V
0 bar also in terms of the volumetric flow rates divided by the area right. So, pi d square
by 4.
So, basically you know I can subsequent I can basically you know recast this expression
in terms of the volumetric flow rate of the liquid that you know that is flowing through a
packed column I can recast it in terms of the diameter of the column you know that makes
the packed bed and also this phi s and D p in terms of you know your specific surface area
and density of the particle if I have a measure if I have a way of measuring, if I know that
I am using a fluid of a particular velocity if I have a measuring way of measuring rho p
again we discussed how to get rho p right in one of the classes using what is the solution
densitometry technique.
So, again a similar plot you know k delta P by L as a function of either V 0 V bar or you
know Q if you want to do it. So, again from the slope I can actually back calculate what is
a specific surface area again which is one of the useful quantity if especially if you are
looking at you know applications where you know catalysis or something where you know
the surface area becomes important one of the parameter that for you know for which you
would have to characterize the particle is for this specific surface area and flow through
525
packed bed concept gives you a way of you know calculating what is the specific surface
area.
So, now commercially people there are a lot of instruments, there is something was a blains
apparatus; blains apparatus which is a commercially available instrument which is for
measuring specific surface area where in the concept that we discussed flow through
packed bed under laminar flow condition that is what is basically used for getting the
specific surface area ok. You have an experiment in the next semester in which you are
going to use this concept for measuring the specific surface of a particle. So, any questions
with that I am going to stop with flow through packed beds. We are going to look at some
examples maybe the tutorial on Friday to look at a few problems where in we are going to
look at flow through packed bed.
So, the next concept that I am going to look at is something called as a fluidized bed. Any
questions that you have? Yeah.
No when you say see V 0 bar was the superficial velocity. Correct. So, all you do is you
have a you know packed bed you know this if Q is a volumetric flow rate and if is a circular
conduit you know you have your cross sectional area going to be a circle right and Q by
A is your you know superficial velocity right. Now the fact that you know your the entire
cross sectional area is not available for flow, but only part of the you know area is
available. So, therefore, what do you do that is V 0 by epsilon is what how it defined right.
526
So, because of the fact that there is a less area available I can it just basically comes from
the continuity equation right.
So, your Q is equal to you know some velocity times you know A and the same has to
again go through the this is before it enters the column the same has to go through the
packed column that is going to be some you know say if the s is V 0 here right say that
sees this is V 0 now if I say the velocity through the column is V 0 bar that has to be epsilon
times A right because the volumetric flow rate has to be you know same across the column.
No. So, V 0 bar. So, the epsilon is the; is a void fraction right. So, what is the. So, now,
this epsilon can go from 0 to 1 right. When you see epsilon is equal to 1 that is when you
do not have a packed bed right both the velocity is exactly the same right. Now the moment
I have some porosity the average velocity is always going to be higher than the velocity
with which the liquid is coming you know just above a packed bed right. So, basically
epsilon captures the factor in the velocity goes up with a liquid you know enters the packed
bed that is the.
527
Fluid Mechanics
Lecture – 45
Fluidized bed
So, we are going to talk about a new topic which is; what is called as Fluidization, ok, or
what I call as Fluidized beds.
Again, like packed bed call you know packed columns fluidized bed concept is again is
something that is again used for you know solid liquid or solid fluid contact, and it has a
lot of applications in the industry. One of the example that I showed in the previous class
was looking at what is called as a fluidized bed coating, ok. If you have say some particles
in the fluid. So, I was talking about the application in coatings where if you really want to
have a like say you have a solid particle.
Say, it could be say for example, a drug particle or something was a API, which is a active
pharmaceutical ingredient, and if you want to coat it with a thin layer of you know some
other material. And that some other material could be it can be a material which is kind of
taste masking because you know a lot of you know like say paracetamol if you take it as a
little sour like now it is a little bitter, ok.
528
So, I can have a coating of some other material which is a inert material which you can
consume like maybe a starch layer for example, right. It could be used in the case of taste
masking or you could also have a case where I would like to have a layer of something on
the particle surface which may help in what is called as a controlled release, ok. That means
somebody takes a pill and you would like to have that drug released into the body over
extended period of time, ok.
What people do is they make this coating and you know you can play with a coating
material you know I can maybe work with you know again the concept that permeability
we talked about, right, yesterday, that comes into picture, depending upon you know how
porous the membranes are you know I can basically control the amount of release that
happens over a period of time, ok. So, there are a lot of applications of you know
fluidization and one application you know is in coating industry where people would like
to look at coating of you know some material on a particle for example, ok.
Now, so we have looked at a packed bed, right. So, packed bed essentially is a long column
we said now, there is you know support plate at the bottom something like that, right and
you have particles are basically filled in the column, ok. Typically, what happens in the
packed bed you know, there is actually a something at the top which basically you know
restrains the solid from going, you know if you are using fluid a very very high velocity it
could so happen that the particles can be carried away, right. So, to avoid that test basically
529
some structures at the top to continue to have the particles in a packed bed state, ok. Now,
if you remove that for the time being, ok, let us say that you know removing that, and what
we do is you are letting either air you know or some liquid, into the packed column, what
do you think would happen?.
So, I am going to have a uniform you know flow rate through which I am going to let the
fluid into the column, again we can talk about you know this either you know the
superficial velocity or the actual velocity, right. So, as and when I increase the flow rate
of course, your pressure drop is going to go up, right. If you go to Ergun’s equation, right,
so you have delta P by L which is basically proportional to a term which depends on the
velocity plus a term which depends on the velocity square, right. Your; you know the
pressure drop is going to go on increasing, ok.
Now, what will happen if you continue to increase the you know fluid flow rate? Right.
At some point you are going to reach a state where the particles are going to break contact
with each other and they are going to, go to a state where the all the particles are you know
a nicely suspended state, ok. So, that is a packed bed, ok. When all the particles go into a
suspended state that is when you said you know I have basically what is what I have done
is I have basically fluidized the bed, ok. So, fluidization essentially refers to you know
start with a, you know packed bed.
So, you know you basically increase your velocity you know enough that all the particles
in the bed they essentially break the contact and they kind of you know, so in this state you
know the bed essentially behaves like a fluid, ok. The entire fluid particle mixture, it
essentially you know behaves like a fluid or you know you if you want to use a more
accurate term it will behave as a dense fluid, ok. Dense fluid because you know the of
course, that your density is going to be much larger and the fact that you know your
particles are there, your viscosity is going to be again higher than the you know the fluid
in which they are dispersed, ok.
And so, where I said for all practical purposes it behaves like a fluid you can actually
transport you know such a you know fluid particle mixture in a pipe, you can actually use
all these you know all the fluid operations that you do, right you know draining of these
fluids through valves everything can be done, ok. For all practical purposes this you know
fluid particles in a fluidized state behave like a liquid, ok.
530
Now, one of the questions that people ask you know when you are working with fluidized
bed is what should be the minimum velocity, what should be the minimum velocity that I
should be using for me to go from a packed state to a fluidized state, ok. That is one of the
you know important question. So, we are going to look at some aspects of that, ok.
Again, we will take the same picture. So, I have a column, there is a support plane at the
bottom you know I have a packed bed, ok. There is some pressure drop tapings here, ok. I
am going to measure the pressure drop, and you know it is an open column. So, I do not
have the, you know the support plate at the top, right. And we can look at, so say that again
the cross sectional area of the you know column is S you know and the initial length that
you had was say L, ok, L is the length of the packed bed that you begin with, ok. What
will happen if I start increasing the volumetric flow rate, right, so or the not M anymore.
So, I am just going to write it as, right. So, I have a fluid that is coming with some you
know flow rate Q and there is a corresponding you know the superficial velocity through
which it is going into the column.
So, I would like to look at you know a couple of parameters, ok. One is how does the
height of the packed bed change as a function of you know V 0 bar that is the superficial
velocity and I would also like to look at how does and I would also like to look at what is
the how does a delta P change is a function of you know the V 0 bar, ok. So, what do you
think would happen? So, I have you know packed bed. Some you know height or some
531
length L, ok. I have a fluid; it could be air or liquid coming into the column, ok. I slowly
go on increasing the flow rate or the superficial velocity I would like to look at how does
delta P vary as a function of V 0 bar and h vary as a function of again V 0 bar.
So, of course, you know your it will depend on you know under what conditions I am
doing the experiment, right. Do I have you know what is a Reynolds number is it, is the
Reynolds number less than 1 or is the Reynolds number more than 1000, depending upon
that I would have to either look at Kozeny-Carman equation, or a Burke Plummer equation
or a Ergun’s equation and I can basically plot how does the pressure drop way is a function
of you know V 0 bar, right.
If you assume that you are working with very fine particles, ok; that means, you know the
size of the particle that you have in the column is smaller. Of course, you know I would
have to be I will be working in the range where you Reynolds number is less than 1,
therefore, your pressure variation is basically captured by the Kozeny-Carman equation
during which you know if I go on increasing V 0 bar of course, my pressure drop is going
to go linearly, right, because you know we this pressure drop is proportional to you know
it is, right its it goes linearly as V 0, yeah, correct.
Yeah, top is free to move, but this will happen as long as this continues to remain as a
packed bed, ok. I am only looking at the initial stage where the in spite of the fact that I
am slowly changing the flow rate, all the particles in the bed are intact there is practically
no change in their rearrangement, there is no rearrangement as such. The fluid velocity is
so low that I can assume it to continue to behave like a packed bed, ok. That is the state I
am talking about hm.
Yes, yeah. So, of course, yeah I would have to overcome gravity if I really want to. So, I
am still talking about cases where, I am still talking about case where you know the
whatever pressure that, whatever you know the force that comes because of the fact that I
am applying a particular flow rate that is still lower than you know the whatever the gravity
that you know because of which is be you know. So, I can say you know there is a
gravitational force that is acting on the entire bed, right.
Now, I would have to offset that, ok. So, I am still in a range where you know your the if
I say delta P, right multiplied by area will give me the pressure force, and I would have to
worry about what is the strength of that this versus the gravitational force, right. So, if I
532
am lower, right if this is more, right, it will still continue to you know, increase linearly,
ok.
So, initially what you will do is you know with the increase in V bar, the your the pressure
drop is going to go on increasing linearly, right that is a straight line passing through the
origin, ok. Now, if I were to look at the height of the bed, ok, the fact that you know there
is no rearrangement in the packed bed, ok, the particles are essentially not moving.
Essentially, the height would remain constant, right. There is no essentially, there is no
change in the height of the column, right or height of the packed bed you know in sense,
right. Now, what will happen? At some point you are going to basically reach a case where
the pressure force that is delta P times A is equal to the gravitational force that is acting on
the entire bed, ok. And at that point there is going to be some rearrangement and if you
increase the velocity past that velocity, there is going to be the some rearrangement
because of which the height may change a little bit, ok.
This is a region where you know the particles are going to kind of loosen up, this is a some
minor change in the porosity of the bed, ok. And however, the change in the porosity, is
kind of offset by the fact that you know there is a small change in the velocity. So, if I
what to look at the delta P as a function of V 0 bar beyond that point, it will continue to
remain constant, ok. The reason why it appears that you know it is remaining constant is
because you are really talking about a very large change in the pressure drop when it is in
a fixed bed state, ok.
Of course we know that you know with the increase in the velocity of the you know the
fluid with which I am letting the fluid into the column, the frictional force are going to
increase, right. However, that change, that small increase that you would ideally see, it is
not seen because you know the in this range the pressure drop is much larger. Therefore,
essentially if you were to plot delta P versus V 0 bar you will essentially get a straight line
in the initial stage and for you know it will remain more or less constant beyond a particular
stage yeah, right.
Yeah. So, of course, you know when it goes to the fluidized state you know you do not
have any contact between the particles anymore, ok. So, whatever losses that comes
because of the fact you know there are of course, there is a fluid particle contact area
because of which there is definitely going to be some losses, ok. But those losses are
533
basically very negligible compared to the loss that would happen when they are in a packed
state, ok.
Now, beyond this point if you go on increasing the V 0, you know the superficial velocity
your height would go linearly with V 0 bar, ok. So, that means, beyond this point your bed
is going to continue to expand, and there is going to be a linear increase in the height of
the bed as a function of time, ok. Now, when I say height of the bed, I am talking about
height of the fluidized bed, right, not the packed bed anymore because all the particles
have lost contact, ok. Then in a fluidized state and beyond this point is when I said you
know there is a fluidization has happened, ok.
Now, when people reverse the flow, ok; so, now, this is the point where I said this is the
point where your delta P multiplied by A is equal to your Fg, ok. That is the point where
you know, where the pressure drop across the bed, ok, the pressure force basically
counterbalanced by the net gravitational force there is acting on the bed, ok.
Now, if you reverse the flow, what is going to happen is it is going to trace this path back,
ok, but however, this the drop in the pressure drop starts to happen a little earlier than what
happen you know, what happened earlier, so your pressure drop for the decreasing case is
going to be something like this. So, if I were to redraw this let me just do that quickly, ok.
So, I am going to draw that that is your a linear variation in the beginning, ok, that is when
you are increasing. When you are decreasing, so the decrease starts happening from here
and then it becomes 0,
And if you look at the way the height changes as a function of you know V 0 bar, so that
constant height you basically get here. That is because of the fact you know there is a
subtle variation in the porosity that will happen you know in the case of you know when
you are expanding the bed, ok. So, all I am trying to say is you know the porosity, at in the
beginning, or the porosity of the packed bed when I started off with. And the porosity of
the bed, when I trace back there is a subtle change in the porosity, that subtle change in the
porosity is what leads to these minor changes, but however, otherwise the behavior is
exactly very similar, yeah, right. What is that? This is the height of the fluidized bed. So.
So, what is that?
I mean so, whether it is a, ok; so, his question is that you know I have a column here, right.
Now, I have fluidized the bed, so there are going to be particles everywhere, right. Now,
534
you know he is asking a question as to you know is there a range over which you know
there is a you know, basically I am talking about the interface between you know the
fluidized bed and you know the clear fluid, right. If you look at in that context is that very
sharp, it will depend on you know what kind of particle are you working with, ok.
If you really have a perfectly mono dispersed particle I would expect it to be a very sharp
interface, ok. But however, if the moment you have a size range, or a poly dispersed sample
or you know bi-dispersed sample of course, you know this you know deviation this you
know very clear interface is going to be a little bit you know blurred for example, yeah.
So, I am; so, when I said this h, I am in the, so I am basically talking about what is the
height up to which the particles are basically suspended in the fluid, that is that height
corresponds to, ok.
Now, what I can do is I can actually get an expression for the minimum fluidization
velocity that you know that V 0 M bar, which is the minimum fluidization velocity. And
typically what people do is when they do experiments, so this vertical line that I am
drawing, the velocity be corresponding to that is what is called as a minimum fluidization
velocity, ok. That is if you are increasing the you know your V 0 bar which is the
superficial velocity, right, so at some point in the your delta P versus sorry delta P versus
V 0 bar, right.
535
So, at the point where you start decreasing at the point where you are seeing a decrease in
the delta P in the second cycle, or at the point where you are seeing a constant height versus
V 0 bar in the second cycle, right. Here the first case you start with a fluidized packed bed
you go into a fluidized state, ok. In the second cycle you start with a fluidized state and
you go back into the packed state, ok. So, this the velocity that correspond to this particular
line is what is called as a minimum fluidization velocity.
And one of the way of getting you know expression for that minimum fluidization velocity
is that you basically equate the pressure forces, right. So, your delta P times S which is the
pressure drop across the packed bed, multiplied by S which is the surface area or the cross
section area of the bed, right, that should be equal to the gravitational force. And because
of the fact that you know you only have as a fraction of the bed with solids, so 1 minus
epsilon is going to give me the fraction of the solid that I have in the bed, right multiplied
by rho P minus rho which is the density difference, right. Because you know I would have
to worry this is the net gravitational force acting, right. I would have to worry about the
rho P minus rho that is the rho of the fluid, right because there is a buoyancy as well, right
times g, right.
And so, now, if you look at this; so, if I and times L as well, right your L which is, so if I
want to talk in terms of, right is that ok. So, this is there is also S here, right, ok. That is
your surface is your cross section area multiplied by length that will give me the volume,
ok, right. Multiplied by 1 minus epsilon will give me the volume of the solid that I have in
the bed, ok. Multiplied by rho P minus rho is what gives you the net gravitational force,
right multiplied by of course, the gravitational constant, right. So, I can get this out, right
therefore, I can write delta P by L as g times 1 minus epsilon into rho P minus rho, ok.
Δ𝑃𝑆 = 𝑆𝑔(1 − 𝜖)(𝜌𝑃 − 𝜌)𝐿
Δ𝑃
= 𝑔(1 − 𝜖)(𝜌𝑃 − 𝜌)
𝐿
So, when this happens that is when you know you start seeing a variation in the height of
the column, right. So, height of the column in you know it was initially constant, the point
where I start seeing a you know the increase in the height of the bed, ok, at that point is
when you know you have this the pressure force equal to the gravitational force, ok. That
is what this expression this expression corresponds to, ok. Now, yeah, right. What is that?
536
Yeah, no. At the before it is not that is L, right. Correct.
Yeah. So, but you know as I said, right there is only a subtle change in the height, ok. I
mean you know you are basically talking about you know a very small fraction, you know.
So, practically you can neglect that small difference in the height you know L to be, ok.
You are talking about this delta P versus V 0 bar or the other one. Correct.
So, this line, ok. Let me make it clear, right. So, let me just rub this here, let us do that, ok.
Your height was constant, right to begin with, ok. There is a small change here, and when
you are reversing it back its going to come back like that, right that is your height, ok. So,
that point here, and the point where the pressure drop starts falling in the second cycle that
should coincide with that and the velocity corresponding to that is what is called as a
minimum fluidization velocity.
So, I mean, so people have come up with the different procedures for doing it, ok. So, this
is one of the guidelines for doing it. But you know we are going to get an expression for
V 0 bar M, ok. Let us look. So, one of the thing that you can do is again you know is there
a large variation in the velocity across this, not much, ok. So, I mean you know I would
worry about you know whether you choose this or this, if the difference between you know
V 0 bar that you would get would be really large you know, but, ok. So, I am basically
talking to you about a procedure that is used experimentally for obtaining the you know
the minimum fluidization velocity, ok.
So, now, at the minimum fluidization velocity, the epsilon that is a porosity of the bed is
typically denoted by epsilon M which is the porosity of the bed at the point of fluidization,
ok. And the point when the fluidization starts to occur is typically kind of you know
referred by a term called incipient fluidization, ok. That is the point at which the
fluidization starts happening, ok.
Now, what I can do is, I can write an expression for delta P by L from Ergun’s equation,
ok. I can equate that to this you know the right hand side. So, therefore, what I can do is I
can write one fifty times mu V 0 bar, into one minus epsilon M whole square, right divided
by phi S square into Dp square into epsilon M cube that was the first term in the Ergun’s
equation, ok. And 1.75 into rho times V 0, of course, now it is going to be M because it is
a minimum fluidization velocity, right. Again M here into 1 minus epsilon M divided by
537
phi S into Dp into epsilon M to the power of 3 should be equal to g times 1 minus epsilon
M into rho P minus rho, ok.
2
̅̅̅̅̅
150 𝜇𝑉 0,𝑀 (1 − 𝜖𝑀 )
2 ̅̅̅̅̅
1.75𝜌𝑉 0,𝑀 (1 − 𝜖𝑀 )
3 + 3 = 𝑔(1 − 𝜖)(𝜌𝑃 − 𝜌)
𝜙𝑆 𝐷𝑃2 𝜖𝑀 𝜙𝑠 𝐷𝑃 𝜖𝑀
Yes, so that is the, so if you can; so you can solve this and you can actually obtain an
expression for you know it is a it is quadratic in V 0 bar M, ok. And there are ways by
which I can actually calculate what is V and epsilon M you know, if you know the
properties of the part that you are working with, basically I can substitute all of that. If you
know the you know the fluid that you are using for your operation I can actually calculate
what is the V 0 bar M, that is the minimum fluidization velocity. That is the velocity with
which I should be operating to achieve a fluidization state, right. Now, it turns out that.
Yeah, go ahead. It is a general equation that is applicable for the entire.
No, no, I you can do that, right its, ok. I mean with this I can actually go back and I can
also look at two limiting cases as well, right. All I am trying to say here is you know you
can look at a general expression, ok. It is not necessary that you know. So, when I talked
about you know the experimental way of you know when I talked about this, right. I just
took a simple case of you know how. So, otherwise I would have to maybe you know, so
if I did not consider the like say Kozeny-Carman equation, I could still draw this plot, how
does the pressure drop change is a function of you know V 0 bar I can you know work it
out, right, ok.
I just took a simple case for which you know you have a linear variation. So, you know
now, it makes the point clear, ok. It is not necessary that you know I would have to be
working with the Reynolds number less than 1, ok. So, now, any more questions? Yeah,
ok. Now, what I can do is I can actually go back and take a look at the limiting cases, ok.
And I can actually one of the question that people ask is that you know, what should be
the typical velocity, ok; what should be the typical value of V 0 bar M compared to the
settling velocity?
So, obviously, your V 0 bar M has to be less than the settling velocity, right. If you have a
packed bed made up of particles, if I want to fluidize it the velocity that I maintain, ok, the
velocity of the fluid that I maintain to have them in a fluidized state it has to be less than
538
the a settling velocity, right otherwise all the particles are going to be carried away in the
fluid, right, ok.
We talked about this alliteration, right, ok. If you maintain a fluid velocity such that the
superficial velocity, or the velocity with which you are sending the fluid into the column,
the moment it becomes larger than the settling velocity the individual particles that make
up the column or the fluidized bed, all the particles are going to be carried away, right.
In fact, we can actually get a estimate of you know how high or how low you know this V
0 bar M should be compared to ut by a very simple analysis. What we are going to do is
we are going to look at the first part of you know your. So, this is 150 V 0 bar M into mu
1 minus M whole square divided by your phi s square into Dp square into epsilon M cube,
right. That is your Kozeny-Carman equation, right at the point of you know fluidization.
So, therefore, what I can do is I can actually get what is V 0 bar M, right. What I can do is
basically if I work this out, what will happen? I have g here, and rho P minus rho, right
and then I have this 1 minus epsilon M here, so that gets cancelled with this, ok. I have a
term which is 1 minus epsilon M on the right hand side, right that basically comes down.
So, let me just do that very quickly, ok.
You are going to have epsilon M cube here multiplied by 1 minus epsilon M, right, ok.
And your phi a square into there is going to be Dp square here, right, there will be phi a
539
square, ok. Thus I have got it the other side and there is a mu here. So, mu is going to come
to the denominator, right and of course, it is going to be 1 over 150, right. That is going to
come to the other side. I can multiply this by 18 and divide by 18, and if I look at this part
that is essentially is your settling velocity, right. Your g you know rho P minus rho into
Dp square divided by 18 mu that is your ut which is the settling velocity that goes as right
your V 0 bar M, there is a minimum fluidization velocity is 18 divided by 150 into ut time
epsilon M cube into phi s square divided by 1 minus epsilon M, ok. Therefore, I can say
your ut by V 0 bar M, is equal to 150 divided by 18 into, right 1 minus epsilon M divided
by phi a square into epsilon M cube, ok, right. This factor sum you know 8.33 or something
like that, ok.
𝑢𝑡 (1 − 𝜖𝑀 )
= 8.33
̅̅̅̅̅
𝑉 0,𝑀
3
𝜙𝑆2 𝜖𝑀
Now, if you are working with spherical particles you are, you know your phi S is 1, right.
And again there are a lot of experimental data where people have shown that if you are
working with spherical particles, it turns out you know the value of epsilon M that people
have measured is typically in the range from 0.4 to 0.45 that is a the porosity at the incipient
fluidization at the point of fluidization for spherical particles. If I substitute that value your
ut by V 0 M comes out to be of the order of 50, ok.
If I substitute for phi S is equal to 1 here you have 8.33 multiplied by 1 minus epsilon M
divided by epsilon M cube it turns out that the number that you get is of the order of 50.
That means, if I were to work with you know the fluidization I can use any velocity, from
the minimum fluidization velocity all the way up to 50 times the minimum fluidization
velocity, right.
The moment the velocity becomes more than 50 times the minimum fluidization velocity
that is when all the particles are going to be carried along with the fluid and when that
happens people call it as what you call the entrainment, ok. When you have a fluid stream
that is coming into the packed bed, or into a fluidized bed and if that stream is basically
carrying, you know carrying the particles also along with the fluid, that particular you
know operation is what is called as a entrainment. And typically you would have to work
under conditions where your entrainment is very much not there, right, you know you do
540
not want the particles to be carried out no along, right, ok. So, that is for ut by V 0 bar M
of 50, ok.
But if you work around this and if you do exactly a similar analysis for you know the
turbulent flow conditions, it turns out that you know ut by V 0 bar M that you get would
be of the order of it goes as 2.32 divided by epsilon M to the power of 3 by 2, ok. The
similar ratio that ut by V 0 bar M goes as 2.32 divided by epsilon M to the power 3 by 2
you can work it out, right.
All you have to do is you know take out the Burke Plummer part, and then rearrange it in
terms of the settling velocity, that is what it comes out to be. And if I do a similar analysis
you know for the case of epsilon M to the power of to be 0.45 this ratio comes out to be
7.7, ok. That means, the operating window for the fluidization, you know I can work over
a much wider you know range of velocities for the case of laminar flow conditions.
However, if I work in a turbulent flow conditions the range of V 0 bar M, you know the
velocity that I can use you know without the entrainment happening is basically lower,
right. That is what it is going to be yeah.
And we will stop here. We will solve some problems on you know all these concepts on
Friday as well as on Monday, yeah.
541
Prof. Basavaraj M. Gurappa
Lecture – 46
Fluidized beds
So, yes I think an important concept that we learned about Fluidized bed was something
called minimum fluidization velocity right, minimum fluidization velocity.
So, I was just trying to explain it based on a typical experiment that everyone performs ok,
in which what you do is you basically monitor, how does the bed height change as a
function of the superficial velocity. In a similar experiment I can also follow how does the
pressure drop change as a function of superficial velocity, right. And, we said that you
know in the initial region right, with the increase in the flow rate, there is a increase in the
pressure drop, ok. Because, which there is a linear change in the or linear increase in the
pressure drop.
However during that period your bed height remains constant because, you know the
upward force that the fluid is basically imparting onto the bed of solids, is not enough to
lift the particles that form the bed, right. And, we said you know there is a small change in
the bed height, that is where you know all the particles kind of loosen up, there is a small
change in the epsilon, ok.
542
However, your pressure drop still continues to remain constant and that is what we are
done ok. We are done this for people typically do it for a cycle you know wherein you start
with a packed bed, you go on increasing the flow rate and then you know and then you
come back to the original state, right. And, actually Praveen had asked a question as to
why should we you know why do we take you know the point B, right.
There is this particular line, that is what corresponds to the minimum fluidization velocity
right and there is a question as to why that is the case, ok. It has to be that point and it
should not be this point that is because of the fact that; so, when you make a packed bed,
ok. So, when you make a packed bed, what do you think this these lines right, the initial
line. What do you think it will depend on?
So, in the initial line is kind of varying in this way right, that is that is your initial increase,
right. So, why does not it very; well, why is not the slope a little steeper or a little smaller?
What does that depend on? It depends on the way you pack things in the packed bed, right.
It depends on the porosity that you create by packing and it is known that I can actually
create a packed bed of different porosity, starting with the same column particle
combinations, ok; that means, if I take a vertical column, if I pour solids, ok. They pack in
a particular way and they give rise to a bed of particular epsilon.
Now, if I do exactly the same process, but in the end if I shake it up ok, they pack a little
differently, right. So, that initial line basically depends on you know the kind of operation
543
that you are trying to do in order to create the pack bed, ok. However, when you do the
reverse cycle; when you do the reverse cycle. What you do is, you take it to a fluidized
state and then you slowly reduce the flow rate, ok.
When you do that ok, the kind of packing factor or the porosity that you achieve is fairly
you know, kind of no matter what kind of salt that you work with. If I take spherical
particles, it turns out that you know this the porosity that you get when you decrease the
flow rate. It is typically around the same value and that is the reason why you know people
choose the second cycle ok, where you decrease the flow rate, ok.
And, that B where you start seeing the flat line which corresponds to the packed height,
right. That is the point where you should be choosing for the minimum fluidization
velocity and it is not this point because, the superficial velocity corresponding to this point
depends on the way you pack stuff ok.
But, however the return cycle, ok typically always gives you a bed of a porosity which is
very similar for whether I use smaller particles, larger particles as long as I work with the
same particle shape, ok. You will always end up with epsilon M that is the porosity at the
minimum fluidization which is very close to we discussed in the last class right, it typical
value of a 0.36 ok, that is the, ok.
544
And, what we had also done was; yeah, we said that you know when, you do; you basically
do a force balance right, you do a force balance wherein you equate the upward force, that
the fluid is imparting to the bed, ok. When that equals to the weight of the bed ok, that is
when you know the packed bed starts fluidizing, right.
And by equating the pressure force to the weight of the bed, we obtain expression for what
is called as a minimum fluidization velocity. And, for different conditions right when, the
Reynolds number is less than 1 and the when the Reynolds number is greater than 1, ok.
And, then from that we also calculated what is a typical ratio ok, what is the typical ratio
of the terminal velocity.
In the particulars you know flow regime again ok, divided by the minimum fluidization
velocity, it turns out that is of the order of 50 for spherical particle for the laminar flow
conditions, ok. However, if you take the higher Reynolds number case, that is the turbulent
flow conditions, it is number basically gives rise to 7.7, right. This is a typical values that
is obtained by assuming a particular value of epsilon M, that is the porosity at minimum
fluidization, ok.
At the take home from this particular you know analysis was that, you get up a larger you
know range of flow rates over which I can operate the bed, you know under in the fluidized
condition. So, that the entrainment does not happen right; that means, you know. So, we
said that for you to have the particles continue to remain in the column, I would have to
545
maintain some velocity, right. Of the fluid velocity, if the fluid velocity is much larger or
higher than the terminal velocity. What will happen is, it will take out all the particles also
along with the fluid, right.
And, we said that is something was a entrainment right when, some of the particles from
the in the bed or carried along with the fluid, that is called an entrainment and you do not
want entrainment to occur, ok. Of course, there are cases where entrainment may be
required right, you know if you really want to you know say that you know I have particles
of different sizes, I would like to take out the larger ones or smaller ones, right.
You know I can also exploit entrainment in some cases, but; however, if you have cases
where I do not want the particles to come along with the fluid, ok. Then, you know this is;
so, basically what I can do is I can actually work with a flow rate range ok, all the way
from whatever value that you are using up to about 50 times. The minimum fluidization
velocity without entrainment occurring in the case of low Reynolds number case or the
laminar flow conditions.
However, if you look at the high Reynolds number case, that range of flow rate that I can
operate is limited right. So, that is why we have stop with the fluidization, right. So, any
questions that you have about these concepts, if you have any questions we can discuss
otherwise, we want we will move on to the next problem, ok.
546
Next I want to talk a little bit about some variation of you know flow through packed beds
but, something called as a filtration.
So, I mean the simple definition is that you know the filtration is the process of removal
of solids from a fluid and this fluid could be either a liquid or a gas, ok. And, for you to
separate up the solids, you know you would have to pass this through a filter medium,
right. This should be some you know or this filter medium is also what is called as a septum
ok; that means, you have a some medium that has a capability of retaining solids, ok.
What you do is you have a fluid that contains particles and you would have to pass it
through this medium and, ultimately the particles get deposited on the medium and the
clear fluid comes out right. That is what is done in a in in filtration and the valuable product
of filtration ok. It could be just the solid particle that you want to recover or it could be the
fluid, it could be both or it could be neither ok. It depends on you know what kind of
operation that you are trying to deal with.
Can think of an example where a both the fluid and the solids after the filtration are not
useful. Can you think of an operation, where I am carrying out filtration using a fluid which
contains particles, where after the filtration I get the solids and the liquid but, none of these
products are useful?
547
Yeah.
We start is water treatment I would not I would call it, I would kind of classifies under
cases where the fluid is useful but, the solids are not right. Because, you know if you have
a wastewater treatment right, you have solids you know typically the solids are the ones
that you want to you know remove out and throw it out, right. And, I would like to use the
fluid that comes out right, the fresh water of the clean water for some use, right.
So, that will essentially come into this case right, where the valuable product is a fluid but,
the solid is not right. Of course you may have cases where, like say industry has a waste
and if this is a waste sludge, ok. Where neither the liquid nor the solid is useful to them,
they want to dump it into some you know, some you know they want to use kind of do
some of the, how should I say this, some of the after treatment methods, ok.
But, if that after treatment method requires, removal of both the liquid waste and the solid
waste and you know they may have a separate process for each of them to you know kind
of dispose right. So, when you are doing that neither the solid nor the fluid is useful right,
you know in the in the end, I they just want to dispose it off.
But however, they just they do not want to dispose it as a fluid particle combination but,
they want to separate it out because, there are certain after treatment methods, which are
specifically suited for solids; some after treatment method which is specifically suited for
the liquid, right. Can you think of an example where the solids are the useful product?
What is that?
Salt water; but, I do not think you can take salt water and then filter it right you know. The
salt is too small you know in size to be retained on an on, right. So, typically there are a
lot of people who work with this nano technology right, the nano particle synthesis,
synthesis of particles and stuff like that, right. Where what people do is they make particles
and then they want to take out these particles for some application, right.
And, typically when they make these particles, it is of course; there are several different
techniques available for making it. But, one of the simple technique that people use is a
kind of a chemical synthesis route in which you have the formation of particles in the
solution, ok. Now, if I want to take out particles from the solution, what is done is typically
548
you can use a centrifugal filtration or you know or some kind of a process where you retain
the solids, ok.
And, the solids are the useful product and then you know the liquid is you know, it could
be waste you know of course, you may also have cases where you know, if I if there are
some reagents in the liquids, I mean you know. I may be wanting to reuse it as well. But
however, if this if the particles are that your generator are useful, you know it that is the
case you know where the solids are the useful product and you know the liquid is not so
much right. Can think of an example where both are valuable products.
Student: Crude oil.
Crude oil, what about a crude oil? I do not think people use filtration in crude oil yeah,
maybe during their, you know extraction properly, but not ok. So, you could have a case
where is imagine a case, where I have a particulate dispersion and say that these particles
are kind of acting as a site for some reaction ok. And, say that you know there is a reaction
going from a to b, and it happens on the surface of the particle that are dispersed in the
fluid, right.
Now of course, the reaction products, if they are formed in solution, I would have I can
take out the solution and I can use it for you know I can the separate out the product right,
ok. So, therefore, the fluid or the or the liquid is useful. And, you can also what you can
do is after the reaction has happened, you know I can recycle the particles right. I can you
know take it out, you know I can you know use it back in next subsequent, you know
operations, ok.
So, essentially what I am trying to say is that whenever you are working with filtration
right, you can have you know any combination, ok. You can have a case where only the
solid is useful or the fluid is useful, could have a case either neither of them are useful or
case where both of them are useful, ok. It depends on the kind of filtration process that you
are going to work with or the kind of the fluid particle system that your you are dealing
with, ok. I have a question here ok; think of an example for filtration operation where, the
suspending fluid is a gas.
Yeah.
549
Yeah that is true, right. So, one of the example is in you know. So, he is talking about an
example where you have a polluted air and I would like to remove you know for example,
when people use this mask, right. You would have seen, you know these days, you know
traffic junction or somewhere. A lot of people use this masks right and the purpose of those
masks is essentially to filter out all the solid particles and then you know let the clean air
right. So, you can inhale the clean air right.
So, of course, the vehicular emissions; so, there is a like say a diesel engine or a or a petrol
engine you know combustion occurs, ok. And, typically whatever you know comes out of
the engine will have some you know gases for example, or some particulate matter. And,
before the gas is you know the exhaust is kind of let into the atmosphere, there will be
some filters; you know where you basically remove some particulate matter, ok.
That is a case where you may have there is an example where the filtration operation is
carried out to remove the suspended, you know the suspending medium is a fluid and what
you removing is a particle. Can you think of an example where the suspending particles
are liquid and the fluid medium is gas? Can you think of an example where I would like
to recover, the liquid particles you know or the liquid droplets present in a, you know a
fluid stream and the fluid is a gas.
Again industrial exhaust right for example, you will have cases where you know the flue
gases that come out of the industry right, will can have some mist right, some liquid
droplets and you would like to recover them, ok. So, essentially filtration is a very
important operation right, you know in industry and again you know all of you are exposed
to filtration when you in daily life as well, right. When you when people make coffee or
tea right, what you are essentially removing is, when you strain it right.
You have a filter medium that is your filter medium and then what you pour on to the filter
medium is your coffee plus some coffee particles and milk and everything, right. And,
what comes out of is a clear liquid right or the coffee which is the useful product in that
case and you basically dispose your solid, ok.
550
So, that is a some brief about you know filtration. I just want to spend a little bit of time
talking about classification of filters. And, whenever it somebody carries of the filtration
operation, typically the fluid flow through the filter medium occurs because of the pressure
difference that you maintain, ok. You know we know that you know for any fluid flow to
occur, you would have a, you need a pressure difference right.
Similarly, in the filtration as well because, you would like to have the fluid flow happening
across the filter medium, I would have to maintain a pressure difference across the filter
medium, right. And, so therefore, you can talk about what is called the upstream and a
downstream right. And, that is your filter medium and of course, you know the solids get
deposited on the filter medium, right. So, depending upon you know.
So, there are cases where what you do is, you maintain in the upstream, you maintain a
pressure, which is greater than the atmospheric pressure and your downstream is
maintained at the atmospheric pressure. So, this is an example of a what is called as a
pressure filtration, ok. This is an example of what is called as a pressure filtration, where
the upstream is maintained at a pressure above the atmospheric pressure and your
downstream is the atmospheric pressure.
Or you can have a case, where your upstream is maintained at atmospheric pressure;
however, and the downstream you put you apply a vacuum, ok. In both cases you are
essentially creating a you are applying a pressure difference and the flow of liquid through
551
the filter medium occurs because of this pressure differential ok. And of course, there are
different ways of maintaining the, you know applying a pressure for example ok. There
are cases where I can use a pump or a blower or a centrifuge force, ok. Or I can also have
a case where you know where I have a filter medium and I have a long column of liquid
above it right.
So, there are different ways by which I can apply your pressure, you know greater than
that atmospheric pressure ok. So, it depends on you know. So, therefore, in terms of the
present pressure differential across the filter medium, you can classify the filters into two
categories ok. One is the pressure filters other one is what is called is a vacuum filters, ok.
And, yeah; so, you can also classify filters based on what is called a filtration mechanism,
ok. In what is called as a cake filters; a cake filters are typically used when you are trying
to filter out, a large amount of solids from a slurry ok. I have slurry, which contains a large
concentration of particles in a fluid ok. And, what you do is there is a filter medium and
then that is your you know upstream, that is your downstream ok. And of course, all the
particles are kind of deposited on the filter medium, ok.
Now, when it comes to clarifying filters; so, clarifying filters are typically used when you
are required to remove a very small amount of solids, ok. Typically the input to such filters
is almost the clear liquid with very few fine particles, ok and, if your objective is to remove
even those small little, you know amount of solids are present.
552
So, you basically pass it through a clarifying filter and one distinction that you see between
the image in the top on the bottom is that, you know most of the solids are kind of deposited
over the filter medium in the cake filters right, that is what the schematic shows, ok.
However, if you look at the clarifying filter, there are lot of solids, that are deposited within
the pores, right. That is the pores that are ok that is the pores of the filter medium. You see
that the size of the particles are even smaller than the pores in the filter medium right. And,
typically the removal actually occurs because of the addition of these you know small
particles to the surface of the filter medium.
And, the only distinction between these two; the cake filters and the clarifying filters is
that, one is typically used for cases where you want to remove a large concentration of
solids, the second one is used for cases where the concentration of solids in the slurry is
very small and the other difference is that the pore size ok. The size of the pores in the case
of cake filters is smaller mostly smaller than the size of the particles. However, in the case
of the clarifying filters; the size of the pores is larger than the particle size that you have.
And, typically the particles adhere to the surface of the medium because, of some you
know some specific interaction between the particles in the medium that you use ok. In
what is called as a cross flow filters, what you do is you have a cross stream of suspension,
ok. That is being flown over a filter medium that is your the filter medium right, that is a
filter medium, ok and, there is a suspension which comes from the left in this case.
And, typically what you do is you maintain a specific velocity with which the suspension
is basically sent and the velocity should be sufficiently larger. So, that the particles do not
sediment right because, you know you would like to keep them floating over the, you know
the filter medium. And, what happens you know some of the thus the liquid basically
comes through the pores that are present in the filter medium and what you end up to the
right is a concentrated dispersion, ok.
So, therefore, cross flow filters are typically used when you want to convert slurry from a
dilute slurry into a concentrated slurry, ok. So, of course, you can do it by other methods,
you know I can take a dispersion which contains a smaller concentration of particles, I can
evaporate the liquid right. And, I can make it into a concentrated solution of course,;
however, that is much more time consuming, right.
553
If I even if I take like say a 10 micro liter fluid, it takes about you know a several minutes
to dry, right. Therefore, if you want to go from I dilute slurry to a constant slurry, it may
take much larger time if you want to use something like drying process. However, if you
use this cross flow filters, it is a you know it is a much quicker operation and I can actually
go from a dilute slurry into a concentrated slurry, you know a reasonably quick time, ok.
So, therefore, based on the filtration mechanism, you can classify the filters into cake
filters, clarifying filters and cross flow filters, ok; any questions that you have, ok.
So, yeah there is some right up so, you can go through it later.
554
Now, so I just want to introduce couple of so the filter media right. So, I said people use
septum you know or the filter medium for separation of particles right. And, of course,
you know the filter medium or the membrane that you are using for filtration and that
should have several requirements right. So, therefore, one of the requirement is that you
know, it must retain the solids to be filter giving a reasonably clear filter aid, right.
So, therefore, so the septum that you are using you should not let the solid particles also
along with the fluid right. The object is to you know filter out the solids; therefore, the
filter aid that you get ok. The filter aid is the fluid that comes in the downstream, right.
You have the upstream, the slurry comes in and whatever comes in the downstream is
called a filter aid right ok.
Therefore, this one of the requirement of the filter medium is that it should retain most of
the solids and it should give a reasonably clear filter aid, it must not plug or blind ok. What
it basically means is that, I have a filter medium with time, ok. With time what could
happen is you know depending upon the operation that you are trying to deal with, these
solid particles may get embedded in the filter medium, right.
Of course, that depends on several factors that depends on the type of the particle that you
are working with, it depends on the pressure difference and things like that ok. And, the
moment you have solid particles that kind of you know are incorporate into the filter
medium, that is what is called as the plugging or a blinding ok, of the medium and of
555
course, that should not happen ok. And of course, it should be chemically resistant ok; it
should be strong physically as well to withstand whatever operation condition that we were
dealing, right.
You may be doing up; you may be doing filtration at high temperature with you know
higher pressure differential, ok. So, you should be able to withstand, you know all these
operating conditions. In the other requirement is you know, it should permit the cake
formed to be discharged cleanly and completely, ok. So, what you do is, once the filtration
is done you are the solids are deposited onto the on the filter medium right. There should
be some easy way of discharging the you know the solids and of course, it should not be
expensive, ok.
So, that is the some of the requirements of filter medium. The other definition that I want
to touch upon is what is was a filter aids and as the name itself says, this is something that
helps in the filtration process, right. So, this is what typically happens is they know if your,
if the solid particle that are there in the slurry that you are trying to filter ok. If the particles
are too small and if the particles are what a called slimy or sticky then, what will happen
is doing the course of filtration.
So, what will happen is that the cake that is formed on the filter medium, it becomes very
dense, ok. When it becomes very dense, the porosity is much lower. If the porosity is lower
is difficult for the filter aid to come out, ok. Therefore, ultimately what will happen is the
liquid may not even come out ok. So, what do you do, I want to avoid that, ok. What people
do is, they add some additives, ok.
So, these additives are there inert ok, they do not you know do any other job right ok, they
do not chemically interact with the liquid neither the solid right. There are inert particles
and the moment you add them what will happen is, they will try to maintain a typical
porosity throughout the filtration process, ok. It will prevent the formation of this dense
impermeable quick cakes, ok.
Therefore, the filtration can continue for a longer period of time, people typically use you
know some cellulose particles, you know you can buy, you know cellulose particles in the
market. So, you can I can put it is inert, it does not really affect either the fluid or the
particle, right. And, then some silica so, basically what you do is. Before you do filtration,
you add these additives into the slurry and carry out filtration, ok.
556
Therefore, the cake that is formed has both the solids as well as these additives and it will
help you to maintain a typical porosity that is required for the filtration to occur. So, the,
that is by adding; the so, when you add things additionally into the slurry ok. That is one
way of doing it, the other way of doing it what is called as a pre coating. What you do is?
So, you have a filter medium, I apply some layer ok, some material on the top of the surface
right of the filter medium.
And, it will kind of prevent, it will prevent deposition of fine particles on the surface ok.
So, either I can use the filter aids or I can use pre coat ok. These are two different ways of
ensuring that you know the filtration occurs smoothly during the course of the operation
that it correctly, ok. Any questions, any questions so far ok.
So, that then will just move on to the. So, the next few minutes I am going to talk a little
bit about; so, the objective of the remaining class, you know is to look at principle of cake
filtration. So, basically we would like to work you know, you would like to develop a
working equation for filtration ok, that is the whole purpose, right. So, in essence, I can
think of you know filtration as a special case of flow through packed bed, right.
Can I say that, filtration is just a special case of flow through packed bed. But, the only
thing is in the conventional packed beds, the resistance to flow remains constant right. But
however, in the case of filtration, the resistance to flow, it changes during the course of
your operation, right. That is the only difference ok. So, therefore, what is what we are
557
going to do is basically; they are going to start with the working equation for flow through
packed bed and you are going to modify it to suit filtration process ok, that is what we are
going to do, ok.
So, in filtration the flow resistance increases with time as the filter medium becomes
clogged or the filter cake builds up ok, on the filter medium right. So therefore, so what
we are going to do is, we are going to start with the Ergun’s equation, which is which we
developed for pressure drop across the packed bed for the flow of a fluid through a packed
bed, right. And, then you are going to modify that to suit the filtration process, ok.
Now, before we do that I just have a question, right. So, whenever you are doing filtration
process, what do you think are the easily measurable quantities that change during the
course of filtration? So, I have I am carrying out a filtration operation right and I would
like to measure some parameters during this process, ok. Can you think of; first thing is
can we think of some quantities that we can measure during this process ok, the second
question is quantity that can be easily measured, ok. What is your answer? What is that?
We can measure this all though this have connected right. So, if I have some what I can
do is, I have a clean filter medium to begin with, ok and, there is a deposition of solids
during the process and I can actually measure it, right. That is of course, I can do that any
other parameter.
558
What is that?
Porosity, porosity how do we measure porous and during the filtration process. So, even
the measuring the solid at a deposit on the surface, it is a little tough to do of course, you
can put in some sensor and then see no. But, the quantity that you can easily measure are
the flow rate and the pressure drop, right. I have a filter medium ok, upstream and the
downstream, I can collect the rate at which the filter aid is coming out, right.
I can actually monitor that as a function of time, right. That is one quantity that I can
measure and the other quantity that I can measure is the pressure drop that I have either
that I have applied or if there is any change during the process. I can also measure, what
is the pressure drop that is across the right ok.
So, therefore; so, depend people talk about two different types of filtration processes ok,
when because of the fact you know, the quantity that I can measure are the flow rate of the
filter aid. And, the pressure drop I can talk about what is called as a constant pressure
filtration ok. And, what is called a constant rate filtration.
In the constant rate filtration what is typically done is; I am measuring, I am actually
collecting the filter aid right. Of course, if I look at a conventional filtration operation, the
flow rate should actually decrease right with time, right. Does that will also; that is what
had actually also happens in the in the coffee straining as well right. I have a, I have a
made a coffee and I am just trying to filter it. Of course, you know after you know us after
you pour out small amount of coffee, ok. Later on, it becomes difficult for the liquid to
come out right, that is because a of the cake formation.
And, what you do is if you want to maintain the constant flow rate of filtration, what you
do is? You start playing with the pressure drop. Therefore, what you do in the a constant
rate filtration is because, your objective is to maintain, the constant rate of filtration. You
go on increasing the pressure drop during the process of during the process. So, that the
filter aid continues to come at the constant rate, but at the expense of progressively
increasing the pressure drop across the filter medium ok, that is what is called as a constant
rate filtration.
In the case of constant pressure filtration what you do is? You maintain a you fix the
pressure drop right. I have an upstream pressure, under downstream pressure, I fixed that
559
ok. I maintain that constant during the course of filtration and I let the pressure let the flow
rate fall, ok. Typically when people work with this cake filtration ok, which involves
passing of a slurry over a filter medium and collecting of the filter aid at the downstream.
People typically work with either the constant pressure filtration mode or a constant rate
filtration mode and we are going to develop working equations for both of them, right.
So, the other thing that is also important is that. So, in the case of so, that is a schematic
that you have right. So, this is your ok, that is your you know the cake filtration, you know
a schematic, upstream the medium, there is a cake formation and that is downstream, right.
Now, if you look at this, if you want to think about the pressure drop right.
Now, there is two contribution to the pressure drop right. One pressure drop that comes
because of the cake that is formed on the filter medium, the other pressure drop comes
because of the medium itself, right. Because, medium is also a porous medium right, the
medium that you are using for filtration. It is of course, again another porous medium ok,
there is the when the liquid flows through the porous medium. There is a of course, a
pressure drop that corresponds to the flow through the medium itself, that is this part and,
delta Pc that corresponds to the pressure drop across the cake.
Therefore, the total pressure drop for the filtration process is the pressure drop across the
cake plus the pressure drop across the medium, right. And of course, that you know the
pressure drop across the medium is 0 to begin with, ok. And, it may remain constant
560
throughout the process, if the medium is mechanically robust, there is no blinding right
when ok. If the medium is not modified in any way during the course of filtration, your
delta P m is going to continue to remain constant throughout the filtration process. But,
however, if there is a blinding or if there is a higher pressure difference across the medium,
you know even delta P m could also change, ok.
However, delta P c would always be there and it could increase during the course of
filtration depending upon, how much of filter cake is deposited on the filter medium right.
So, we will think a little bit about some of these things and then try and develop a working
equation for them in the next class, ok.
So, I will stop here and if you have any questions I will be I will be happy to answer,
otherwise you can stop here, any questions that you have. Simple concept right, I mean
filtration is a fairly simple concept. So, we will try and you know work with the working
equations in the next class and then yeah. So, we will have class on Tuesday and
Wednesday and then a tutorial on the last day, which is Friday ok. So, if you do not want
a tutorial, we can also end it by Wednesday; otherwise, we will have a last class on
Thursday; Thursday, Friday sorry yeah ok.
Thanks.
561
Fluid and Particle mechanics
Lecture – 47
Filtration
So, in the last class, we learned that if you look at the cake filtration. So, there is a the
liquid that flows through two resistance there in series; that means, you know it. So, if
initially you have the fluid that is flowing through a filter cake, plus it is also flowing
through the filter medium. And we learnt that, the total pressure drop delta p is equal to
the summation of the pressure drop across the filter cake plus the pressure drop across the
filter medium right.
And, in this class what we are going to look at is, how do we think about expressing delta
p c, that is a pressure drop across the filter cake and delta p m which is the pressure drop
across the filter medium. In terms of some of the measurable properties such as you know
the flow through the filtrate, flow rate and the viscosity of the in the filtrate the properties
of the particles and things like that. So, first what we will do is we will start with thinking
about the pressure drop through the filter cake that is, you know we are going to talk about
how do we relate delta p c to the measurable parameters. What you are looking at is a
diagram ok.
562
Where you visit this is basically a section through a filter cake; and the filter medium at
some time t you know from the start of filtration ok. That is you know u at time t is equal
to 0, you do not have; you do not have any filter cake developed; that means, you know
your this part is not there right. So, only thing that you have is only the filter medium and
during the course of filtration, the filter cake builds up at time t is equal to 0. What you are
done is, you’ve taken a cross section and the figure basically represents that view ok.
And so in this diagram, so, that the direction of flow is basically marked there. So, that is
a direction of flow of slurry and that is the you know if you want to call it as that is a
upstream end right. And, then the cake builds up and you have a filtrate that is coming out
that is the downstream. L c is the thickness of the filter cake and that is measured from the
filter medium and A is the area available for filtration. So, what it essentially means is,
you know if I have like say if this is my filter medium right.
And, if the filtrate is flowing in this direction, the cross sectional area that is available for
you know for the filtrate to flow through that is A and what we are going to do is, we are
going to look at a very thin section. So, what is drawn here is small section that is I am
going to mark it now. So, that is a very small; thin section of dimension dL that is the dL
is the thickness of the; you know the layer of filter cake that I am considering. And, this is
basically located at a distance L for the distance from the filter medium to that layer is
capital L. And, we are going to say that the pressure you know at that particular point is p
ok. And, what is shown in the figure is the pressure gradients.
563
Now, if you look at this thin section ok, typically the velocity with which the filtrate flow
through the filter cake are so low that you can essentially assume the flow to be laminar
ok. So, therefore, you have so as I said right, we are basically borrowing the concept from
flow through packed bed and we are trying to apply it to the flow through the filter cake.
And, the pressure drop across the filter across the packed bed, we said you know it has two
terms right one is the laminar flow term and the turbulent flow term.
So, the laminar flow term basically goes with you know, which is proportional to v or u in
this case, which is the superficial velocity of the filtrate. And of course, there is going to
be a u squared term that we are basically neglecting it because, we are assuming the
laminar flow which is true because, the velocity with which the filtrate flows through the
filter cake are sufficiently low that you can assume the laminar flow conditions.
So, the other interesting thing to notice; if you look at the pressure gradient either it is not
linear right. So, it basically you know varies something like this right, that is the line that
is basically drawn that is how the pressure drop varies right. Now, if you look at the
conventional you know packed beds right so you would expect the pressure drop to vary
linearly with the length of the packed bed right. You know if you; if you know because,
you know if you are you know if you are using certain kind of packing so what happens is
that you know you are the porosities are constants typically and then your mu which is the
velocity, which is the viscosity of the fluid that is constant.
564
And, you know if you are working with rigid particles so essentially. So, all these terms
are essentially constants and that basically gives rise to a linear variation of pressure with
distance. However, that is need not that is need not be the case, in the case of you know
flow through filter cakes because of which you essentially have you know not a linear
variation, but a variation that looks more like you know quadratic ok. The reason for that
is that if you look at regions of the filter cake close to the filter medium they are kind of
subjected to greater compressive force compared to you know the region that is you know
the closer to the upstream.
Because of which there could be a spatial variation of porosity that is one of the reason
why there could be you know a non-linear variation. The other reasons could be there you
know your view which is the filtrate the velocity with which the filtrate is flowing through
the medium that also could be spatially varying you know depending upon you know what
kind of system that you are trying to look at ok. So, therefore, this non-linear variation that
you are seeing is essentially justified.
So, essentially what we are doing is, you know in this thin section, we are going to say that
you know the delta p by L that is the pressure drop across this thin section. You know how
does that vary with over the small distance dL and we are basically retaining only the first
part of the Ergun’s equation that is the Kozeny Carman equation right.
565
Now, what I can do is and as I said you know the u is essentially defined as the superficial
velocity of the filtrate, exactly in line with you know how the v 0 bar was defined for the
flow through packed bed. And what I can do is I can start with this expression, I can do
some manipulations ok.
So, one of the manipulations that I can do is that instead of working with you know phi s
squared D p square, where phi s is the sphericity and D p square is the; D p is the diameter
of the particle that are part of the slurry which is being filtered or the diameter of the
particle that constitute the cake, I can replace the sphericity in terms of s p by v p that is a
specific surface you know surface area of the particle divided by the volume of the particle.
And if I do that, I can replace phi s squared D p square you know as s p by v p whole
square divided by 6 square ok. And so, the 6 square 36 therefore, you know you have 150
divided by 6 square that leaves out you know 4.17 right. And the superficial velocity of
the filtrate is essentially defined as, the rate at which the filtrate is collected at the
downstream divided by the filtration area that is available.
𝑑𝑝 150𝜇𝑢 (1 − 𝜖)2
= 2 2
𝑑𝐿 Φ𝑠 𝐷𝑝 𝜖3
𝑠𝑝 2
4.17𝜇𝑢 (1 − 𝜖)2 (𝑣 )
𝑑𝑝 𝑝
=
𝑑𝐿 𝜖3
𝑠𝑝 6
𝑆𝑝ℎ𝑒𝑟𝑖𝑐𝑖𝑡𝑦: =
𝑣𝑝 Φ𝑠 𝐷𝑝
𝑑𝑉
)(
𝑢 = 𝑑𝑡
𝐴
566
So, in this definition u is equal to dV by dt divided by divided by A. V essentially refers

to the volume of the filtrate collected from time t is equal to 0 until time t is equal to you
know t. And, once a particular volume of filtrate enters the filter cake, essentially the same
volume you know flows through the entire filter cake.
That means you know your whatever dV that enters the same amount of volume is
essentially flowing through the entire you know if I basically spread them into different
strips right does to the. So, dV essentially is same across all the strips that basically
constitute the filter cake. Therefore, you know your V by A is same for all layers.
Therefore, I can assume that u is independent of the length across which the filter you
know the filtrate is essentially flowing. And the volume of the solids in the layer, I can
write this as A is the filtration area multiplied by delta L.
That gives me the volume of the solid plus the liquids you know in the filter cake right in
the thin cross section right that is so, essentially A multiplied by dL gives me the volume
of this thin section that multiplied by 1 minus epsilon will give me the volume of all the
solids in the inner thin section. And, if I want the mass that is dm, that is the mass of the
particle in the thin section; you know which is essentially this the volume of the solids in
the section multiplied by rho p; where rho p is the density of the particle that constitute
that make up the you know of the solids in the filter cake right.
567
So, therefore, I can again simplify this equation further. So, I have delta p here and I am
going to take dL to the right side and I can write dL ok, I had dL here so if I want to write
it as you know this term multiplied by dL here right. So, therefore, I can express dL as dm
divided by A times 1 minus epsilon into rho p right.
So, one of the epsilon essentially cancels with this. So, I am basically left with 1 minus
epsilon here, there is k 1 there is mu u here. So, I am basically taking this as a constant k
1 right. And of course, s p by v p remains here, and I have dm that comes in and there is
A you know that is here. And, essentially epsilon cube remains here and you know rho p.
So, essentially I am basically expressing delta p in terms of delta m right.
𝑑𝑚 = 𝐴 𝑑𝐿 (1 − 𝜖)𝜌𝑝
𝑠𝑝 2
(1
𝑘1 𝜇𝑢 − 𝜖) ( )
𝑣𝑝
𝑑𝑝 = 3
𝑑𝑚
𝜌𝑝 𝐴𝜖
568
Now so, if you work with; if you are doing filtration operation and if you ensure that you
know you are carrying out filtration operation under conditions where the pressure drops
are very very low. And, if you are working with a slurry which contains rigid uniform
particles ok; all the pre factor including k 1 mu u 1 minus epsilon s p by v p rho p you
know all of these are our constants ok. And the only variable that I have is dm ok.
And you could ask a question as to, what are the conditions under which you know you
could have you know this you know the each of the parameters that I am mentioning as
constants could vary, you can consider several cases. One of the example could be the you
know if I have like say, I am working with a filtration process in which I instead of having
rigid particles that constitute the you know the slurry, if I have like say emulsion droplets
which are essentially liquid droplets right.
And we know that you know as the filtration you know occurs. So, I have you know the
filter medium and once these liquid droplets start depositing on to the filter medium right.
When you know the further filtration happens you know, these droplets could deform right
instead of them being perfectly spherical you know they could take some shape like that
right. So, therefore, there could be a variation of s p by v p which we may have to worry
ok. If the particles that I am working with are not rigid ok, but similarly if you have you
know the slurry which contains loose flocks, you know for example, you know if the slurry
consists of you know some kind of aggregates of particles.
569
So, what could happen is, during the filtration process there could be some compaction of
these aggregates which could lead to changes in the porosity ok. And of course, you could
also have cases you know, where you could. So, essentially if you are working with rigid
particles you know, you could assume the rho p to be constant ok. However, if you are
working with you know something like micro gel particles which you know in such cases,
your rho p itself will not be a constant, you know you micro gel particles are particles
where you have a rigid core you know plus some hairy brush kind of a thing there could
be spatial you know variation of density.
So, therefore, there are cases where, you know you need not have the pre factor constant
ok. However, if you are working with slurries which contain rigid particles which do not
deform during the course of filtration, I can assume you know all the these terms could be
constant. Therefore, what I can do is I can get an expression for the pressure drop across
the filter cake just by integrating this equation starting from so, I my pressure at.
So, if I go back to the right. So, if you have. So, p prime is the pressure at the filter medium
right. So, therefore, I can integrate this expression going from p prime to p a. Where p a is
the pressure at the upstream end therefore, I can integrate from p prime to p a D p is equal
to so all of this is constant if I am working with slurries containing rigid particles. And the
mass of the cake that is that you know deposited to begin with the 0.
Then when the pressure drop is you know, when the pressure is reached and you know and
I have a total of mc mass deposited. Therefore, I am going to have the integrations from 0
to m c. Therefore, this essentially gives rise to p a minus p prime, which essentially is your
delta p c which is the pressure drop across the filter k and so therefore, this becomes mc
right.
𝑠𝑝 2
𝑝𝑎 𝑘1 𝜇𝑢 (1 − 𝜖) (𝑣 ) 𝑚𝑐
𝑝
∫ 𝑑𝑝 = ∫ 𝑑𝑚
𝑝′ 𝜌𝑝 𝐴𝜖 3 0
𝑠𝑝 2
𝑘1 𝜇𝑢 (1 − 𝜖) (𝑣 ) 𝑚𝑐
𝑝
𝑝𝑎 − 𝑝′ = = Δ𝑝𝑐
𝜌𝑝 𝐴𝜖 3
Now, so in this case I have assumed that the filter cake is incompressible ok. And as I said
that is only true, if I am working with rigid particles right.
570
Now, so, what I can do is, I can define a particular quantity, which is what is called as
specific cake resistance ok. Which essentially is I am basically manipulating this
expression ok. I club lot of constant together. So, that is k 1 there ok. s p by v p whole
square that is here ok, 1 minus epsilon that is here ok. And, rho p into epsilon cube all of
these things. So, if you look at these parameters carefully, these are all the parameters that
are specific to the kind of particles that your that are present in the slurry.
You know and essentially the these are these are properties which depend on the on the
packed bed right. So, you know epsilon is a porosity of the bed right, and Sp Vp and rho
p are the properties of the particle. So, all these constants to get put together are kind of
clubbed together and that is your alpha. And essentially you are left with delta pc there ok.
And your A from here goes to the numerator divided by you have mu u and mc.
𝑠𝑝 2
𝑘1 (𝑣 ) (1 − 𝜖)
Δ𝑝𝑐 𝐴 𝑝
𝛼= 𝑤ℎ𝑒𝑟𝑒 𝛼 =
𝜇𝑢𝑚𝑐 𝜌𝑝 𝜖 3
So, the other way to think about you know this specific cake resistance is that if you look
up the flow through packed beds right. So, we had developed this expression delta p by L
goes as 150 mu v 0 bar into 1 minus epsilon whole squared divided by epsilon cube into
phi s square D p square right. Now, I can plug a lot of these constant that are there right.
So, all these constants I can plug them into you know one constant, I can write this as delta
571
p by L is equal to mu v 0 bar divided by something called as k ok. k is what is called as a
permeability, k is what is called as permeability which basically tells you in a way one of
the way to think about this would be that you know how much permeable you know the
packed bed is for the flow of fluid ok.
Δ𝑃 𝜇𝑉0
=
𝐿 𝐾
And so, if you look at this expression, this is what is called as a Darcy’s law ok which
essentially states that, the flow or the superficial velocity essentially is directly
proportional to the pressure drop across the bed and it is inversely proportional to viscosity
right. Now, I can rework this I can say that your delta p basically is mu v 0 bar divided by
K by L rights or is a ok. So now, now what I can do is in flow through packed bed. So,
you know you define resistance ok, which basically goes something like this.
So, your L by L is the length of the packed bed. Divided by this permeability in this case
if because, we are talking about flow through you know the cake I can write this as K c
that is the permeability of the cake. That essentially is defined as some kind of resistance
ok. I am going to call it as resistance because of the filter cake R c ok. So, therefore, the
resistance because of the filter cake is defined as the length of the you know the packed
bed or in this case the length of the cake divided by K c, where K c is the permeability of
a cake right.
Now, so therefore, your R m which is L by K c basically goes as delta p divided by mu

into v 0 bar ok. So, just look at the analogy of this expression with this right. So, I have
delta p c there, I have mu into u ok. Therefore, I can write this as delta p c divided by mu
into u whole divided by mc divided by A ok. This is the cake resistance that is R c and
because, you have a flow through you know filter cake here. So, I would have to worry
about what is the mass of the cake that has you know that is kind of deposited per unit area
of the filter ok.
So, therefore, your alpha which is the specific cake resistance is essentially defined as R
c, which is the resistance of the cake which essentially is delta p divided by mu v 0 bar in
the case of flow through packed bed. So, in this case is going to be delta p c which is the,
the pressure drop across the filter cake divided by mu times u where u is the superficial
572
velocity divided by mc divided by A which is the mass of the cake developed per unit area
of the filter ok. That is you know the concept of you know the specific cake resistance.
Now of course, you know I can express you know alpha also in terms of you know phi a
squared D p squared that is basically going back to another starting equation only thing is
instead of k 1 I am going to have some other constant k 2 ok.
Now, So, therefore, so this is the expression that we have kind of developed right. And as
I said right also so this expression that we are looking at it may not be precise ok. If the
feed does not contain rigid particles ok; therefore your porosity the constant k s p by v p,
they may vary from layer to layer right as I said there you know if you look at you know
the region close to the filter medium ok. If there is a large cake that is developed, the region
close to the filter medium, there could be much more closer packing of particles compared
to the region closer to the upstream right.
Or, if you could have a case you know where you could have, if you have applied up a
larger pressure drop across the filter medium you know there could be compaction of the
cake which may lead to variation in the spatial variation in the porosity ok. Therefore, I
may have to worry about non-linearity in such cases; however, for the case of
incompressible cake I can assume that there is a linear variation of pressure drop across
the length.
573
So, similarly in analogy with you know the definition of the cake resistance, we can also
define the filter medium resistance ok. So, we worried about m c by A ok. That is because
you know we were talking about a specific cake resistance ok. Which ok; that means, I
know I would have to worry about what is the mass of the cake develop per unit area of
the fit you know filter, but; however, in the case of filter medium resistance I do not have
that term does because this is this is defined as the filter medium resistance ok.
Which is R m essentially goes as the length of the filter medium that is you know if you if
you have a you know if L m is the length of the filter medium, divided by what is the
permeability of the medium that essentially goes as delta p m which is the pressure drop
across the filter medium divided by mu and u alright.
𝑝′ − 𝑝𝑏 Δ𝑝𝑚
𝑅𝑚 = =
𝜇𝑢 𝜇𝑢
So, but of course, you know the delta p m if you look up you know the delta p m may also
varying right because, you know if you could have a case; if you have if you have if you
have applied a larger pressure drop across the filter during the filtration process.
What could happen is? The larger you know the pressure drop can may force some of the
particles through the filter medium some of the solids to the filter medium. And depending
upon the cleanliness ok, if you have like some of the pores that are going to be blocked
because of the you know presence of solids ok. So, these factors may change delta p m,
574
but you know if you maintain the cleanliness you know if you are working with a
reasonably smaller pressure drops, you can assume delta p m to be constant during the
entire course of filtration process. However, delta p c would varied in the course of
filtration ok
So, therefore, I can write the total pressure drop that is delta p is delta pc plus delta p m
ok. As the contribution from the pressure drop because of the cake and the contribution
that basically comes because of the presence of the filter medium. Now, we had defined u
which is the superficial velocity as dV by dt divided by A. You can relate mc which is the
mass of the filter; mass of the cake that is deposited that is mc. in terms of V which is the
total volume of the filtrate collected from time t is equal to 0 to time t ok. And c which
essentially is the mass of the solid particles deposited in the filter per unit volume of the
filtering.
If I know these two quantities, mc essentially is the product of V times c. So, this will have
units of like say meter cube ok. And c will have you know itself something like kg per
meter cube therefore, mc essentially will have units of mass that is the mass of the solid
particular and basically deposit onto the filter cake per unit volume of them right. So, now,
so therefore, instead of mc I can replace mc with c times V right. And so, now, what I have
done is here and of course, I have u here right. So, u I am going to write it as dV by dt
575
multiplied by 1over A right. And because I have delta p right delta p is equal to. So, I have
delta p is equal to dV by dt multiplied by mu into mc alpha A plus R m right I have this.
Therefore, I can get this you know dV by dt into the other side that becomes you know dt
by dV right. I am going to get delta p to the to the right hand side that will have delta p in
the denominator. And mu remains you know wherever it is and A also remains in the
denominator. So, essentially I am recasting this equation you know as dt by dV is equal to
mu divided by A delta p times alpha c V divided by A plus R m right.
𝑚𝑐 𝛼 Δ𝑝𝑚 Δ𝑝𝑐 𝐴
Δ𝑝 = Δ𝑝𝑐 + Δ𝑝𝑚 = 𝜇𝑢 ( + 𝑅𝑚 ) ; 𝑅𝑚 = ;𝛼 =
𝐴 𝜇𝑢 𝜇𝑢𝑚𝑐
𝑑𝑉
𝑢 = 𝑑𝑡 ; 𝑚𝑐 = 𝑉𝑐
𝐴
𝑑𝑡 𝜇 𝛼𝑐𝑉
= ( + 𝑅𝑚 )
𝑑𝑉 𝐴Δ𝑝 𝐴
Now, so now, now I think I have an equation, which is in a particular form which I can
further reduce to suit the need of you know so we were talking about the constant pressure
filtration and the constant you know rate filtration right. So, in the case of constant pressure
filtration, what is done is you basically maintain a constant pressure drop across you know
the filter filtration process. And then you basically look at how does the volume of the
576
filtered that I collect you know varies as a function of time ok. Therefore, I am basically
interested in to develop an expression in which t and V are my variables.
So, therefore, I would. So, what I have is this expression ok. So, what I know is you know
because, I am working at a for a constant pressure filtration. My delta p is going to be a
constant. Therefore, in this expressions only variables are V and t, but I know the initial
condition at time t is equal to 0. I do not have any filtrate coming down; that means, you
know your V is equal to 0. And the only pressure drop that I have is the pressure drop that
comes because of the presence of the medium. Therefore, my delta p is essentially is equal
to delta p m therefore, I can. So, therefore, my delta t by delta V at time t is equal to 0 is
equal to mu multiplied by R m divided by A times delta p ok. Plus the other term is going
to be constant is equal to 0 because V essentially is 0.
Therefore and I call this as 1 over q 0 ok. So, therefore, I can recast this equation as dt by
dV is equal to one over q. That is equal to K c times V. So, I have defined a constant K c
which basically is your mu multiplied by c alpha that is the numerator and divided by A
square into delta p and because, delta p is a is constant that basically is kind of included in
the in this constant K c.
Therefore, dt by dV is equal to 1 over q is equal to K c times p plus 1 over q 0 ok. I can

integrate this equation further and then if I put the limits that at time t is equal to at time t
is equal to 0, but you know. So, I have the coordinates t 0 0 and t V ok; that means, you
know at time t is equal to 0 I do not have any filtrate collected and you know at time t is
equal to 0 the total volume with the filter is essentially V ok. Therefore, this expression
essentially becomes t over V is equal to K c by 2 times V plus 1 over q 0 right. So, what
essentially done is your dt goes as K c into V into dV plus 1 over q 0 into d V right.
So, therefore, if I integrate this what I get is your 0 to t dt is equal to K c into 0 to V V into
dV plus 1 over q 0 into 0 to V dV right. So, that is becomes your t is equal to K c into V
square by 2 plus 1 over q 0 into V ok. Therefore, I can divide everything by, this becomes
t over V and 1 we essentially goes and this also goes. So, therefore, you basically end up
with an expression which is t by V is equal to K c by 2 into V plus 1 over q 0 ok. So,
therefore, we are going to developed working equation, which tells us you know how the
dependence of you know p and V during the process of constant pressure filtration.
577
𝑑𝑡 𝜇𝑅𝑚 1
( ) = =
𝑑𝑉 0 𝐴Δ𝑃 𝑞0
𝑑𝑡 1 1 𝜇𝑐𝛼
= = 𝐾𝑐 𝑉 + ; 𝑤ℎ𝑒𝑟𝑒 𝐾𝑐 = 2
𝑑𝑉 𝑞 𝑞0 𝐴 Δ𝑝
𝑡 𝐾𝑐 1
= 𝑉+
𝑉 2 𝑞0
One of the nice things about this expression is you know, I can actually if I perform
experiments under constant pressure; conditions I can actually estimate what is K c and
what is q 0. And you can if you go back and look at what is you know K c and q0 depend
on, you know the K c depends on what is alpha which is the specific cake resistance. And
1 over q 0 basically depends on you know is related to R m which is the filter medium
resistance. Therefore, what I can do is, I have a framework to essentially calculate what is
the filter medium resistance under specific cake resistance?
And you can also exploit the constant pressure filtration, this is to say something about
what is the type of the cake; that is kind of formed during the course of filtration. So, what
I can do is, I can conduct a number of constant pressure experiments; that means, you
know I basically maintain various pressure drops that is you know I do experiments at
delta p 1, delta p 2, delta p 3 like that ok.
578
And then I calculate what is R m and what is alpha that is essentially basically doing I do
a constant pressure filtration I do t versus V versus where t by V versus you know V right.
And, then I get a straight line I look at what is the slope and the intercept and from the
slope and the intercepts so I can actually calculate these things. Therefore, for a set of delta
p conditions, I can calculate what is alpha and R m. And, if alpha is independent of pressure
drop; that means, no matter what I am what is the pressure drop there is maintained.
If alpha remains constant then I can say then you know the cake that I have there is form
is incompressible. However, that need not be the case; that means, you know if I were to
plot. You know alpha for delta p if it remains constant; that means, it is a incompressible
cake ok. However, that need not be the case and one of the empirical equation that is being
developed to kind of capture, how the how does the alpha which is the filter which is the
specific cake resistance base is a function of pressure drop is alpha goes us alpha 0 into
delta p to power s, which is an empirical equation which is obtained by doing a lot of
experiments.
𝛼 = 𝛼0 (Δ𝑝)𝑠
And, this alpha and s are empirical constant. And s is something called as a compressibility
coefficient ok, it tells you how compressible or how incompressible the cakes are if s is
equal to 0; that means, it is a incompressible cake. However, typically people have found
that you know the alpha takes a value from 0.2 to 0.8 depending upon what pressure drop
you have maintained. And more closer this value is to 1; that means, it is highly
compressible, more closer it is to 0 that mean it is you know less compressible.
And, for you know alpha is equal to 0 I can say that the cake is completely in
incompressible. Therefore, this particular framework gives you a way of calculating both
the filter cake resistance and the filter medium resistance, which in turn can be used you
know as a method for identifying the nature that filter cake that is essentially formed during
the course of filtration ok. maybe I will stop here. So, I will continue with talking about
the constant rate filtration and the continuous filtration in the in the next video.
Thanks.
579
Fluid and Particle mechanics
Lecture – 48
Filtration
So, this is going to be the last lecture on Filtration. What I will do; I will just do a quick
recap of what we did in the last lecture.
So, we were looking at developing an expression for what is the total pressure drop across
a filter cake in the case of cake filtration. And what we did is, we ended up with this
particular expression. We ended up with you know showing that the total pressure drop
delta p, which is summation of the pressure drop because of you know the presence of the
cake and because of the presence of the porous filter medium is equal to you know the
contribution that comes because, if there is a that there is a term which is what is called as
a specific cake resistance and R m is what is called as a filter medium resistance.
And they are kind of defined in this particular way right. That is what we are done in the
in yesterdays in the last video lecture.
580
And then, what we did is and we took that general expression and then we considered a
case of a constant pressure filtration. Wherein, the objective is to relate the time of filtration
and the volume of the filtrate that is collected during the course of filtration. So, we ended
up with an expression which is t by V is equal to K c by 2 times p plus 1 over q 0 that is
what we are done in the in the last video lecture.
As I mentioned so one of the nice things about doing this constant pressure filtration
experiment is that, you know you can actually generate data with which I can say
581
something about the cake resistance. And, I had mentioned that you could do a series of
constant pressure filtration experiments. So, what you are looking at is data from for
different filtration experiments.
So, test I, test II, test III, test IV and test V they are being done at different pressure drops,
which are basically indicated here. So, 6.7 all they up to you know 49.1. And during each
of these experiments, you basically collect what is the filtrate volume that is collected and
the time ok. Therefore, I essentially have t versus you know t by V data.
And what I could do with this data is I could actually plot them. These are again further
five different trials. So, I have this is for trial I you know and then trial you know V right.
And so, if you plot t by V versus V I get a straight line you know with particular slope and
an intercept and from this slope and the intercept data I can actually calculate what is R m,
which is the filter medium resistance and what is alpha which is the specific cake
resistance.
582
Now, for this particular example, if you look at how this R m and alpha vary as a function
of you know for as a function of pressure drop. It turns out there you know both of them
are not constant ok. Which is which the fact that R m also changes is little what surprising,
but you know you could argue that look maybe the pressure drop that is you know applied
across the during the filtration process is so huge that you know, there is there is some
modification to the filter medium.
That means, you know there could be some pores that are clogged because of the clogged
by the particles, because of the higher pressure drop that I have applied. That is why you
know you see some variation of R m you know with respect to the pressure drop that I
have applied. And, similarly alpha which is the specific cake resistance also changes ok.
583
So, the point that I want to make is that ok in general alpha yeah can increase with delta p
because, you know because, most of the cakes that people deal with they are to some extent
they are compressible ok. And whenever, you have a compressible cake, you will have
alpha varying with delta p for highly compressible cake alpha increases very rapidly with
delta p, but you could have cases where you know there could be a or the variation could
be may not be that steep. And, you could also have cases you know where you know you
have a constant alpha as well ok.
It depends on you know, if you essentially; if you work with conditions where you know
my delta p that I have applied is smaller plus if the slurry that I am working with is consists
of rigid particles you know which do not kind of yield when you apply a pressure a
particular pressure drop you know I could assume that the cake is incompressible.
So, most commonly what people do is, people use experimental data and they fit in you
know power law kind of a you know expression of this sort where alpha goes as alpha 0
into delta p s to the power of s. And, depending upon you know what is the value of s ok,
if it is closer to 0 its it is if it is 0 it is you know incompressible and if it becomes closer to
1 you know it becomes highly compressible ok. And, for the example that we looked at it
turns out to know for this case you know the exponent s is about 0.26, which basically tells
me that you know the cake that is being dealt with is you know slightly compressible right.
584
So, now that we have talked about you know the constant pressure filtration, we will move
on to looking at constant rate filtration um. In the case of constant rate filtration; what is
done is the filtrate flows at a constant rate ok. Therefore, you know your u ok, which is the
superficial velocity is constant during the course of filtration. So, as I had mentioned that
during the course of filtration, the cake develops because of the formation of cake there is
going to be more and more resistance because of which ideally I should expect the flow
rate to fall ok. That the filtrate should start coming down at lower and lower flow rates.
However, to maintain a particular constant you know superficial velocity or the flow rate
of the filtrate, you carry out experiment by progressively increasing the pressure
difference; that means, you know what you do is you because, you want to maintain a
constant filtration rate you go on you start with certain delta p and then you progressively
go on increasing delta p. So, that you know your filtrate flow rate remains constant during
the course of filtration ok.
So, and because you know your u is dV by divide by A and that is constant for the case of
constant rate filtration u becomes V by At. Now, the expression that you see here is the
expression that we developed for the delta p c right. That is a pressure drop across the filter
cake. And now, if you know the how delta p c and alpha are related and if I have some
way of estimating what is delta p, you know then I can actually use this expression to relate
the overall pressure drop to the time right.
585
𝑑𝑉
𝑉
𝑢 = 𝑑𝑡 =
𝐴 𝐴𝑡
Δ𝑝𝑐 𝜇𝑢𝑚𝑐 𝜇𝑐 𝑉 2
= = ( )
𝛼 𝐴 𝑡 𝐴
So, in the case of constant pressure filtration, we were interested in looking at an

expression which will involve the time of filtration and the volume of the filter that I am
collecting. In the case of constant rate filtration, what I am after is to obtain the relationship
that relates the delta p versus the time that is what we are after.
And, it can be done in a fairly simple way. So, what we are going to do is, I have you know
this expression for delta p c that is delta p c by alpha is mu u mc divided by A and for u I
am going to replace that u with. So, what I am going to do is, I am going to write mu
instead of u I am going to write V by A times t that is for mu. And I have mc divided by
A is what I have what I am going to do is I am going to multiply this by t and divide by t
ok.
Therefore, what I essentially have is that that is mu and I have a. So, we also know that
mc is V times c therefore, what I am going to do is this right let me just ok. I have mu here.
So, that is and u and mc is V V times c right and divide by A. And what I am going to do
586
is, I am going to multiply this by t and divided by t so, therefore, this becomes mu into V
square by A square.
So, I do not have to do this multiplied by divided by t therefore, it becomes mu by V square

divided by A square into c divided by t right. Therefore, this becomes mu times c divided
by t into V by A whole square ok. That is what having right, but now if I so, we had said
that you know there are empirical relationships that relate alpha to delta p c right. So, if I.
So, we saw that you know your alpha is alpha 0 into delta p to the power of s.
Similarly, if you assume a relationship; similar relationship for alpha; so, I could write
alpha is equal to alpha 0 into delta p c to the power of s ok. I could do that. Therefore, all
I am doing here is I am retaining delta p c here right. And, then instead of alpha I am
writing alpha 0 into delta p c to the power of s right. And therefore, if I take alpha 0 to the
right hand side so, that I have alpha 0 multiplied by mu c t I can what I can do here is I can
multiply by t and divide by t here.
Therefore, I can I have t that comes in the numerator that is this t. And I have V by A t
whole squared right. And so, I because delta because delta p is equal to delta p m plus delta
p c I can instead of delta p c I can substitute it for delta p minus delta p m to the power of
1 minus s because, you know I have the power 1 here right and I have the power s here.
So, therefore, if I take it to the numerator this becomes 1 minus s therefore, delta p minus
delta p m to the power of 1 minus s is equal to some constant times t. And this constant K
r essentially is alpha 0 into mu c multiplied by V by A t whole squared. And we know that
V by a t is u therefore, your constant K r becomes alpha 0 into mu c u squared ok.
Δ𝑝𝑐 𝜇𝑢𝑚𝑐 𝜇𝑐 𝑉 2 Δ𝑝𝑐 𝜇𝑐 𝑉 2

= = ( ) → = ( )
𝛼 𝐴 𝑡 𝐴 𝛼0 (Δ𝑝𝑐 )𝑠 𝑡 𝐴
𝑉 2 𝑉 2
Δ𝑝𝑐1−𝑠 = 𝛼0 𝜇𝑐𝑡 ( ) → (Δ𝑝 − Δ𝑝𝑚 )1−𝑠 = 𝛼0 𝜇𝑐𝑡 ( )
𝐴𝑡 𝐴𝑡
(Δ𝑝 − Δ𝑝𝑚 )1−𝑠 = 𝐾𝑟 𝑡 ; 𝐾𝑟 = 𝛼0 𝜇𝑐𝑢2
So, this basically gives us the relationship between how this you know the how the delta
what is the delta p that I would have to maintain, you know to basically achieve during the
587
course of filtration that is you know as a function of time to basically maintain the constant
rate of filtration. That is the working equation for the constant rate filtration.
So, just to sum up what we have seen so far. So, the total pressure drop delta p is delta p c
by delta p m and it is it can related to the specific cake resistance on the filter medium
resistance. And for the case of constant pressure filtration, you have t by V is equal to K c
into V by 2 plus 1over q 0. And, for the constant rate filtration you have delta p minus
delta p m to the power of 1 over 1 to the power 1 minus s is Kr times t.
588
Now so, what I want to do, now is to look at continuous filtration. So, I would like to we
would like to develop. So, we would like to consider the case of a constant pressure
filtration, but; however, instead of looking at a batch filtration process that is what we
looked at so far, but you know, but, but a continuous filtration process.
Before we do that I just would like to kind of take an example of a rotary vacuum process.
So, and tell you little bit about what continuous filtration is all about. So, in the case of
continuous rotary filtration ok, which is in this case its a in this case the pressure drop that
is supposed to be maintained constant is done by applying a vacuum ok. So, in this case
what you essentially have is that, you have a horizontal drum that is the horizontal drum.
And so what you are looking at is if this is my drum, I am basically looking at it at a side
view right.
So, therefore, I basically see a circle that is what you know this circle that you are seeing
is all about. Now, this drum it actually has a slotted face; that means, you know there are
pores on the outer drum. And this drum rotates at a slow speed ok. They are typically point
one to two you know revolutions per minute and if you look at the diagram carefully.
So, what you do is, you this particular drum is immersed partially right in you know a
slurry trough. And there is a feed that basically helps us with feeding the slurry that needs
to be filtered into the slurry trough. And as you can clearly see that, there is a portion of
the drum that is immersed in the slurry trough. And over this slotted phase ok, you
589
essentially have a filter medium, you know it could be a canvas it could be, you know a
kind of a cloth for example. And that covers the face of the drum ok, the entire drum is
essentially covered with you know the filter medium.
And so, now, so, that is the outer drum. Outer drum and there is also an inner drum ok.
And the essentially the inner drum is actually is it is a solid it does not have pores; however,
it has some slots which are essentially kind of shown in this diagram ok. These slots are
essentially for carrying out the filtrate liquid into a rotary valve which is over here right.
And from there I can actually take out the filtrate liquid.
And between the two drums, there are these radial partitions thus these are the radial
partition that I was talking about ok. And they divide the space between the outer and the
inner drum into separate annular compartments ok. And each of this is connected through
an internal pipe that is these pipes into a one hole in the rotating plate to the essentially so
all these are connected to the rotary valve and from there the fluid is kind of taken out.
590
Now, what we are going to look at is. So, if you look at So, I am going to start with looking
at this region A that is that is you know we going to look at the panel A that is shown here.
And this particular region is about to enter the slurry um. That is present in the trough.
Now, as soon as this region enters the slurry, what is done is a vacuum is applied you know
through the rotary valve ok.
So, you have the capability of applying vacuum whenever we want and you also have the
capability of releasing the vacuum whenever we do not want it. So, now, as soon as the
vacuum is applied a layer of solid builds up right, because of you because you are applying
a vacuum right and the liquid is sucked into you know these radial tubes ok. And whatever
solids that you have in the slurry, they basically are deposited on the filter medium.
And of course, the filtrate is collected. So, it basically it comes to the wall and is collected
in the collecting tank oh in this case. If you look at this particular operation you know, this
is a the rotary filter that we are talking about. It basically has two collectors so one is for
the washed liquid which you are going to talk a little bit later and one is for the filtrate.
Basically filtrate gets collected in one you know container and the washed liquid that is
typically used for washing the filter cake is collected in a separate drum.
591
Now, so as soon as the panel A leaves the slurry, it enters you know the washing and the
drying zone. So, this is basically where the washing is done. So, you essentially have a
spray that is basically sprayed on to the cake that is formed. And that ensures the removal
of the residual filtrate liquid. And vacuum is applied to the panel from a separate system.
So, essentially once you have used a washed liquid of course, I would like to drain out
whatever washed liquid has also entrained within the filter cake.
So, again I would have to suck all the washed liquid. So, for that you have a separate
arrangement to do that. And once the washed liquid is drawn through the filter into the
separate collecting tank that is you know that is your washed liquid that is a tank for
collecting the washed liquid.
592
Now so, after the cake of solid on the face of the panel have been sucked so; that means,
so the objective of this is to make sure that the cake is as dry as possible. So, once that is
done it basically leaves the drying zone. And once the it leaves a drying zone ultimately,
you remove the cake by using a knife which is what is called as a doctor’s blade that is
doctor blade. Which is essentially used for scraping the cake that is formed on the filter
medium and then ultimately the cake is dislodged.
Now, if you look at the operation for at anytime it is cyclic ok, but that what I mean to say
is that. So, some of the panels in each part there you in any given time. So, they will always
they will always be. So, what I mean to say is that the operation of any given panel. So,
what I mean by any given panel is you know so each compartment right that is the each
panel ok. So, for the operation of any given panel, you know it happens in a cyclic way,
but since some panels in each part of the cycle at all times, the operation of the filter as a
whole is a continuous process ok.
That means, you know there are some panels you know in which the filtration is going on,
there is some panels you know where there is washing going on, there is some panels you
know where there is a drying going on ok. And, some panels where there is a the scraping
of the you know of the filter cake is going on ok. In that sense you know. So, the whole
operation occurs in a continuous way.
593
So, therefore, what we can do is, you can actually modify. So, if you look at the continuous
filtration process, the feed the filtrate, the cake they all move at a steady and constant
during constant rate during the entire course of filtration. So, if you consider the particular
element of the filter surface; that means, you know if I take different you know these
panels, the conditions are not steady, but they are transient right.
Because, you know some are undergoing you know the cake formation process some are
undergoing the washing process you know some are going the cake removal process. So,
therefore so for example, if you take an element of the filter cloth ok, from the time it
enters the point of slurry it undergoes the cake formation the washing, drying discharge
and each step involves progressive and continuous change right.
594
So, what we are going to do is, to develop the working equation for the constant pressure
filtration. For the continuous filtration process, what we are going to do is. We are going
to start with whatever expression that we developed for the constant pressure filtration that
does t by V is equal to K c divided by 2 times V plus 1 over q 0. We are going to modify
that to essentially suit the need of the continuous filtration process that happens in a
vacuum rotary filter right.
So, the way we do this is so now, this expression that you see that is a quadratic equation
in V right. So, therefore, it has two roots. So, if you have you know ax squared plus bx
plus c right, if that is your quadratic equation, we know that the roots are minus b plus or
minus square root of b square minus four ac divided by 2 a right. Now, if you if I get there
two roots one of them is positive that is what this root is you can get that just by you know
by this particular formula.
595
Now, so once I have this. So, I would like to because I know what is K c and you know 1
over q 0, I would like to substitute for these in this expression and then I would like to
reduce this expression from this into this. So, the way to do that would be what I can do
is, its just a simple algebraic manipulation. So, what I can do is instead of 1 over q 0, I am
going to replace that with.
So, I am going to write it as V is 1over q 0 is mu square R m square divided by you know

A square into delta p square that is for 1 over q 0 square plus 2 excuse me two times K c I
can write it as mu c alpha divided by A square into delta p. And of course, I have the t
there that is to the power of 1 over half minus again 1 over q 0 is mu R m divided by A
into delta p divided by instead of K c, I have mu c alpha divided by A square delta p right.
Now, what I can do is um; I can I can do some simple manipulations or what he can do is,
I can multiply and divide this by delta p. So, therefore, what I am going to do is I have 2
there which is which was there earlier as well ok. I am going to write delta p here and then
I am going to make this as delta p square right. So, then I have just multiplied and divided
by delta p, this is the second term in the in the parentheses um. Then what I can also do is,
I can multiply and divide this also by mu ok.
So, what I am going to do is again I am going to rub this off ok. I am going to multiply
this by mu divided by mu oops sorry about that ok. Divided by mu and I have 1 over 2
right. So, what I can do is. So, I have mu square there right I have mu square and R m
596
square divided by A squared delta p squared plus, I have mu mu here that basically makes
it mu square right. I have A and delta p square divided by A square delta p square. And I
am going to write two times ok. Delta p that is this I have c times alpha into t divided by
mu whole to the power 1 over 2 minus mu R m divided by A delta p whole divided by mu
c alpha divided by A square into delta p right. So, I can do that. Now, what I am going to
do is rub this all off ok.
So, what I am going to do is, I am going to take mu divided by A delta p is constant. I am

going to take it out of the. So, therefore this; so therefore, this becomes R m square plus
and this is also out this is also out ok. So, that that becomes two times delta p c alpha t
divided by mu to the power of 1 and a half minus. Again I have taken mu A delta p outside
that becomes R m divided by mu c alpha divided by A square into delta p.
So, I am going to cancel mu here I am going to cancel the delta p as well and one of the A
also gets cancelled. Therefore now, so what I can do? So, therefore, essentially what I have
is V is equal to A times, I am going to write the second term first delta p into c times alpha
t divided by oops um; divided by mu plus R m square to the power of half minus R m
divided by c times alpha right. I can take A to the other sides that becomes divided by A.
So, I am going to get rid of; I am going to get rid of A here right.
So, now what I can do is I can divide by t you know on both the sides. So, therefore, V
divided by A times t becomes 2 times delta p c alpha t divided by because, I am taking it
597
inside you know there is a to the power of one and a half there. So, therefore, it becomes
mu into t square plus R m by t square R m square by t square to the power of half minus
R m divided by t whole divided by c times alpha right. And one t gets cancelled here so I
have this ok.
1/2
1 1
(
2 + 2𝐾𝑐 𝑡) −𝑞
𝑞 0 𝜇𝑐𝛼 𝜇𝑅𝑚 1
𝑉= 0 ; 𝐾𝑐 = 2 ; =
𝐾𝑐 𝐴 Δ𝑝 𝐴Δ𝑝 𝑞0
1/2
𝑎Δ𝑝𝑐𝛼 𝑅𝑚 2 𝑅𝑚
𝑉 [ + ( ) ] −
𝜇𝑡 𝑡 𝑡
=
𝐴𝑡 𝑐𝛼
So, essentially that is what u is an end up with right. So, that is this expression. So, it is
basically a simple mathematical manipulation will essentially lead you to this expression.
Therefore, what we have done is we have kind of converted this into V by A t is equal to
2 times delta p c delta p into c times alpha divided by mu t right. That is this is mu here
right; mu t plus Rm by t whole squared to the power of one and a half minus R m by t
divided by c times alpha.
Now, So, in this expression this V by A is the rate of filtrate collection right. That is a
because this is the volumetric flow rate right volumetric flow rate divided by A will give
you what is the rate at which the filtrate is being collected. And in the case of batch
filtration, A was the filtration area; however, in this case this A is going to be the area of
the drum that is submerged in this slurry.
598
So, now what we are going to do is, we are going to start with this and then do some simple
manipulations. So, now, that you have this expression. So, I would like to kind of modify
this expression to bring in the; the experimental variables that are typically used in the
course of you know continuous filtration. That means, I would like to express this in terms
of measurable quantities such as the rate of solid production ok.
So, this m dot m dot c is the rate at which the. So, essentially we talked about mc being V
times c right and if I want the rate of production of the solid which is m dot c essentially
599
becomes you know V times c divided by t right. That is what I have here. So, I would like
to express this in terms of the rate of solid production. And if you look at the filtration
process, so we worry about what is called as a cycle time tc and also the speed at which
the drum is rotating that is n and of course, the total filter area right.
So; that means, I have the entire drum that is available for you know you know I have a
drum area which is which is At. However, only a part of it is available for filtration that is
what is A right. Therefore, if the fraction of the drum submerged is f. I can actually relate
this as t by t c right. That means, that is a total cycle time and that is a time for which the
drum is in contact with the slurry and this essentially depends on the rotation speed right.
Depending upon what is the speed at which the drum is being rotated, if it is rotating very
fast ok, you know that is you know; you know the drum is in contact with the fluid for a
shorter period of time. However, it is rotating very slowly that is the drum is in contact
with the slurry for a longer period of time. So, therefore, so I have these expressions which
is the fraction of this you know area that is submerged, you know goes as A divided by At.
That is A is the area that is available for filtration.
And At is a total area. And again I can express the same thing in terms of t by t c. And of
course, the filtration time basically depends on you know is f divided by n, where n is the
speed at which the drum is rotating. So, therefore, what I can do is I can use all this and
then recast this x this equation into this form ok. Again it is a simple manipulation. So, if
you are interested you know if you know you can you can try and do it you know. So, all
you have to do is maybe I will just do that quickly ok.
600
So, I would like to develop this right that is what I am after. So, what I will do is, I will
start with this expression right. I have V divided by At now what I can do is I can take this
c to the to the other side therefore, it becomes V c is equal to 2 A 2 delta p c times alpha
divided by mu t plus R m divided by t whole square to the power of one and a half minus
R m divided my t right.
Now, I know that f is A divided by at therefore, a is f times At. Where At is the total area.
Therefore, I am going to replace; I am going to replace A with At times f right. Now,
therefore, I have V c by t and we know that V c by t is m dot therefore, I have m dot c
divided by At. I am going to take f to the right hand side therefore, and because I am going
to and of course, I have divided by alpha here, because I am going to take f inside the you
know the bracket here.
So, therefore, essentially I have 2 delta p c times alpha mu into t multiplied by f squared.
Sorry multiplied by yeah of course, f squared, plus R m into f divided by t whole squared
to the power of half note that you know I have f square here and of course, f square here
and to the power of one and a half it basically gives me f that I have taken to the other side
right minus R m into f by t right.
Now, let me just, now that is the divide of course, I have it everything is should be divided
by alpha. Sorry of course, I have t here everything is divided by alpha. Now, so I know
that because t is equal to f by n therefore, t is equal to f by n like therefore, my n is equal
601
to f by t right ok. Therefore, what I can do is I have m dot c divided by At is equal to 2
times delta p c alpha into mu into f by t into 1 over t. Sorry I have f here right that is f
square plus R m into f by t whole squared one and a half minus R m into f by t, whole
divided by alpha right.
Now, f by t is n I have that here and then another f which basically is this f. And this f by
t is n right. So, therefore, this becomes n times R m whole squared and this becomes n
times R m. So, essentially we have kind of recasted this expression essentially into this
form that is what we have done.
1
2Δ𝑝𝑐𝛼𝑓𝑛 2
𝑚𝑐 [ + (𝑛𝑅𝑚 )2 ] − 𝑛𝑅𝑚
𝜇
=
𝐴𝑇 𝛼
So now so, that is the equation, that we have recasted. Now if you assume that you know
you had working under cases you know, where the filter medium resistance is negligible
there I can essentially kick this term out and also this term out. So, the only thing that will
be remaining is m dot divided by At is equal to 2 times delta p times c into f times n divided
by you have mu here you have alpha here. So, therefore, that is that is what you will I think
of course, you know you had alpha here.
So, therefore, it would become alpha divided by alpha square because, you know if I take
it to the there is one and a half there is therefore, one gets canceled therefore, in the
602
numerator you should have alpha right. That is what so that expression is right ok. And of
course, you know I can say that if you are working with a compressible cake, I know that
delta p is equal to. So, alpha is equal to alpha 0 into delta p to the power of s right. So,
therefore, I can replace for alpha in terms of alpha 0 into delta ps therefore, alpha 0 remains
in the you know denominator and you know the numerator becomes delta p to the power
1 minus s.
𝑚𝑐 2Δ𝑝𝑐 𝑓𝑛 1/2
=( )
𝐴𝑇 𝛼𝜇
1/2
𝑚𝑐 2(Δ𝑝)1−𝑠 𝑐𝑓𝑛
=( )
𝐴𝑇 𝛼0 𝜇
Therefore, so you basically have a framework by which I can actually find out, what is the
rate at which the solids are being deposited onto the filter cake. And, you know for a case,
where both the cake resistance and the medium resistance becomes important. And, I can
recast this you know expression to really look at cases where the filter medium resistance
becomes negligible.
And of course, you know I can also kind of modify the expression to those to suit the need,
where I have a cake that is compressible; that means, I am going to I have taken the
empirical expression that was developed to essentially get an expression in terms of
expression that relates m dot to the coefficient that basic basically tells me something
about, how much compressible the cake is.
603
So, the last topic that I want to discuss is to look at something called as the principles of
centrifugal filtration um. So, I mean if you so the way the centrifugal filtration works is
you have a container right. Maybe you can think about it a cylindrical drum and it is
basically rotated right. And so, what is so what you are looking at here that is the outermost
that is a inner surface of the drum ok. And this inner surface of the drum of course, would
be it will be perforated that means there are pores ok.
And on top of it you essentially put a filter cloth. And because of the centrifugal action so
what will happen is, the slurry is kind of forced to come in contact with the filter medium
and the solids basically get deposited onto the filter medium and that at least the formation
of a filter cake. And of course, you would have a liquid with the particles you know in the
inner region of the you know the centrifugal filtration set up.
So, if you think about this as a axis of rotation um. So, this r 2 essentially refers to the
inside radius of the basket or the drum that basically rotates. And, r i refers to the radius
of the inner face of the cake that is you know, that the cake is formed until this distance.
And ri is the radius of the inner face of the cake. And r one is the radius of the inner surface
of the liquid right and of course, you know if once the filtration is complete, if all the you
know the all the slurry is kind of consumed. So, you really would not have this liquid.
What will be left in the end would just be the fill the filter cake.
604
And what we can do again in this case you know, I can whatever a basic theory that we
have developed for constant pressure filtration. You can also apply the same kind of you
know formalism, but; however, you would have to modify that to suit the need of
centrifugal filtration.
Of course there are some assumptions that are kind of important when you develop these
things. So, we are going to assume that the cake is incompressible, the resistance of the
filter medium is constant; that means, you know. So, your R m is not changing during the
course of filtration. And, you know we are going to assume that the cake is completely
filled with liquid such that you know the flow is laminar right. So, you have all these
assumption that that have gone in terms of developing these expressions.
605
So, again we will start with the same working equation the total pressure drop that is delta
p is you know delta p c plus delta p m right that is what we had we had mentioned and this
u which is the superficial velocity, with which the filtrate is flowing. So, we are going to
write that as q by A that you know where q is the volumetric flow rate of the liquid. So,
essentially I have replaced u by q times A. Therefore, I have q here. And I have taken A
to the in the parenthesis therefore, because I have here. So, I have a square here and of
course, you know I have divided by A that comes into the picture.
𝑚𝑐 𝛼 𝑅𝑚
Δ𝑝 = 𝜇𝑞 ( 2 + )
𝐴 𝐴
And so, we are going to assume that area A that is the area available for filtration, does not
change with radius ok. Of course, that is this is only true if you are really working with the
drum that is really-really large and whatever filter cake that is formed it is very very thin
ok. Only in such cases I can assume that you know the area A is constant during the course
of filtration; however, you can of course, modify your working equation to also suit the
need, where the area could also be changing as well.
606
So, now if I want to solve this further I should know what is delta p ok. And this delta p
which is the pressure difference that occurs during the course of centrifugal filtration, you
can actually obtain by bringing this concept of hydrostatic equilibrium under the action of
centrifugal force ok. What it essentially means is this right if you have a liquid column
right and if I want to get what is delta p the pressure difference across like say if I have a
liquid column.
So, if this is say the position 1 and position 2. We know that you know delta p is equal to
is you know rho g delta h right that is the you know that you kind of derive it from the that
basically is the hydrostatic equilibrium right If you have a vertical column of liquid.
Similarly if you have a fluid in a centrifuge and if you are rotating at a at a at a speed
omega right that is the rotation speed. And if rho is the density of the fluid and I can
actually similar to relating the pressure drop to you know, the difference across two
locations. If I have if I am at some radial positions r 1 and some other radial position r 2.
The pressure difference delta p essentially goes as rho times omega square into r 2 square
minus r 1 square divide by 2. You can actually derive this by doing a very simple force
you know force balance. Now, so, therefore, so I am going to you know instead of delta p
I am going to you know use this expression that comes from the hydrostatic equilibrium.
607
So, therefore, from this I can actually get what is q that is, what is the rate at which the
filtrate is coming out you know from the during the you know from the centrifugal
filtration process. Again a simple you know rearrangement leads you to this expression.
And of course, I have assumed that you know A is essentially a constant right that as I said
will only be true if you are working with drums which are sufficiently large in diameter or
cases where the cake that is developed is very very thin right.
608
However, if you have cases where, you know the there is a change in the area is too large
ok, if you cannot neglect the change in the area. All you have to do is you would have to
replace you know this A square that you had here right with AL bar times A a bar and the
A that you had here right with A 2 that is the inner area of the drum itself right. A 2 is the
inner you know area of the drum that is available for filtration right. And A a bar is what
is called as a arithmetic mean cake area and AL is what is called as a logarithmic mean
cake area ok.
𝜌𝜔2 (𝑟22 − 𝑟12 )

𝑞=
𝑚 𝛼 𝑅
2𝜇 (̅̅̅𝑐̅̅̅̅ + 𝐴𝑚 )
𝐴𝐿 𝐴𝑎 2
̅̅̅̅
𝐴 𝑎 = (𝑟𝑖 + 𝑟2 )𝜋𝑏
2𝜋𝑏(𝑟2 − 𝑟𝑖 )
̅̅̅
𝐴𝐿 = 𝑟
ln ( 𝑟2 )
𝑖
So, these areas better capture you know quantitatively what how do the q changes during
the course of a filtration process ok. So, that is you know how one would deal with
centrifugal filtration process. So, with that I would end the topic on you know, centrifugal
filtration and you will have some assignment based on this in this particular week yeah.
Thanks.
609
Lecture - 49
Tutorial - 07
Hello everyone, today I am going to solve the Tutorial of a Fluid and particle Mechanics
and the topic is filtration. So, first I am going to solve the first question; the first question
consists of first I am going to read the question laboratory filtration conduct at a constant
pressure drop on slurry of calcium carbonate in S2O gave the data shown in the table.
Table is even here volume say time and t y V, the filtrate area is given 0.044 metre square,
the mass of solid per unit of volume of filtrate is given 23.5 gram per litre, and temperature
is given. The viscosity of water is 0.886 centipoise. And pressure drop is given in the
question they have mentioned that the pressure drop is constant throughout the process,
and the pressure drop is 46194.8 Newton per metre square. So, first I am going to convert
all the data in the 1 units. So, here I am going to convert all the data in f foot pound system
units.
So, forth first I am going with area filtrate area; area equal to 0.044 metre square equal to
0.044 divided by 0.3048 square, it will be 0.474 foot square. Similar to that field
concentrator filtrate is given 23.5 gram per litre in fpf it will be 23.5 in to 28.31 divided
610
by 454 is equal to 1.47 pound per feet cube, similar to 4 viscosity. Viscosity of water equal
to 0.886 centipoises equal to 5.59 into 10 powers minus 4 pound per foot second.
And for the pressure drop also delta P equal to 46194.8 Newton per metre square, in foot
pound system it will be 31073 lb per foot second. So, in the question that data table is
given that with respect to volume, I am going to make the table also. Filtrate volume time
second and t by V 0.5 1, 1.5, 2, 2.5 and 3.
And time 17.3, 1.3, 72.0, 108.3, 152.1, 201.7 and t by V is given from 34.6 sorry 34.6,
41.3, 48, 54.15, 60.84 and 67.23. So, in the question they mentioned it is a constant
pressure drop problem. So, for that there is a equation it is, for constant filter process
equation is t by V equal to k c by 2 into V plus 1 by q naught. So, if we plot t versus V
versus this is a equation of a straight line where intercept is 1 by q naughty and k c by 2 is
the slope.
So, I am going to do a rough plot for this t by V versus volume 1, 2, 3, 4, 5, 10, 20, 30, 40,
50, 60, 70. So, if you plot this t versus V versus V it will give you a straight line. Similar
to that this is the unit of this is the unit in second per litre.
This is whole imaging litre, later after getting the intercept and slope I will covert in the
same foot pound in a system. So, from the graph you will get intercept equal to 28.21
second per litre and slope equal to 13.02. So, in your intercept is normally is equal to 1 by
611
q naught from that equation and equal to 28.21 in foot pound system it will be about 800
second per feet cube. Coming to the slope part slope equal to k c by 2 equal to 30.02 in
foot pound system it will be 10440 second per feet cube. And now I am going to apply the
same constant pressure formula.
So, the equation was t by V equal to k c by 2 into V plus 1 by q naught where k c is equal
to mu c alpha area square to pressure drop. Here mu is the viscosity and c is the
concentration. So, now, everything is known except this alpha; alpha is a specific k a k
resistance that we have to find out alpha equal to area square delta p k c by mu into c.
So, just put the slop part k c by 2 equal to 10440 into 2 equal to 5.95 into 10 power minus
4 into 1.47 into alpha divided by 0.474 area square into pressure drop 31073. So, you will
get a specific cake resistance equal to 1.66 into 10 power 11 foot per 8 pound this is the
first part.
In the second part I have to find cake resistance that is from the slope we got the slope
equal to sorry intercept intercept equal to 800 second per feet cube and equal to this one
mu R m cake resistance into area into pressure drop.
So, everything is known here just the value 800 equal to is 5.95 into 10 power minus 4 R
m divided by 0.474 into 31073. So, you will get R Rm equal 1.98 into 10 power 10 foot
inverse, this is the first part of first question.
612
Now, I am coming to the second question second question is the just the extension of the
first question. Here in the second question only the pressure drop is not constant pressure
drop is varying with respect to specific resistance. So, in the in this question there is table
is given, pressure, in this question table is given second question versus pressure drop filter
resistance and a specific filter resistance alpha, this table are given, so ok.
So, I am going to plot the graph Rm versus delta P, it will show how the cake resistance
vary with the pressure. So, if you plot this graph it will come to the straight line, from this
graph from the graph we can say that the filter medium resistance is linearly varying with
pressure. This is the and, but the total pressure raise; but the total pressure rise is only 50
percent of initial resistance. But in this case if you see the table pressure raise is almost 6
to 7 times from the initial value.
613
In the second part the from the table I am going to plot the log; I am going to plot log log
plot of alpha versus pressure drop. So, I am going just giving the rough plot here, log alpha
log delta P.
So, it will come similar to a straight line if you fit the point. So, from this equation there
the slope from this plot the slope, will give the value of S. For varying pressure drop the
cake is not incompressible I am sorry ok. Oh sorry this S is the compressibility factor of
the cake ok.
So, this equation will be suitable alpha equal to alpha naught into delta P raise to power S,
here S alpha here alpha not and S is empirical constant. Normally S for S equal to 0 cake
is in incompressible and for normally the value of S vary from 0.2 to 0.8 for S is being
positive the cake will be compressible.
So, from this plot you will get S equal to 0.25. So, from this we can conclude that the slope
of the above plot is S and the this is slightly compressive. The value of S is about 0.25 the
so, the cake is not that much is slightly compressible.
614
Now, I am going to solve the 3rd question in this tutorial. In the 3rd question l e filter has
an area of 0.5 metre square and operate at a constant pressure drop 500 kilo Pascal. The
following test result were obtained for slurry in water which gave raise into a filter gate
the table is given here, volume versus time. 0.1, 0.2, 0.3, 0.4, 0.5, 140 second meter cube
360, 660, 1040, 1500 this is also a constant pressure programme. So, we are going to apply
the same equation as in the first one.
So, for that we need the value of t versus V for plotting the graph. So, if you calculate the
t versus V, you will get this one is 1400 this one you will get 1800, 2200, 2600, 3000. So,
we are going to apply the same formula t versus V equal to k c by 2 into V plus 1 by q
naught.
Here k c is equal to mu c alpha divided by A square delta p and 1 by q naught equal to mu

R m by A into delta P. So, I am going to do a rough plot for this. So, this one is a volume
this is time versus time by volume. 3000 of volume 0.1 0.2 0.3 0.4 0.5 so, here if you plot
the graph you will also you will get the same straight line. So, from the graph the slope if
you calculate the slope you will get slope equal to 4000 second per meter 6 and intercept
equal to 1000 second per meter cube.
615
So, for the first I even I am flowing the question area is given to 0.5 meter square and in
the first case the constant pressure being dropped is, 500 kilo Pascal equal to 500 into 10
power 3 Pascal.
So, coming to the intercept part 1 by q naught equal to mu Rm A into delta P. So, from
this equation I am going to calculate mu R m equal to 1 by q naught was 1000 into area is
0.5 meter square into pressure of 500 into 10 powers 3. So, you will get mu Rm equal to
2.5 into 10 power 7. So, from the slope part k c equal to mu C alpha by A square delta P.
So, from this equation I am going to calculate mu C alpha mu C alpha equal to 4000 into
0.5 square 500 into 10 power 3. It will give the value of 1 into 10 power 9 Pascal second
per meter square. In the question they have asked to calculate the time need to collect 0.8
meter cube of filtrate at a constant pressure drop of 700 kilo Pascal. So, again I in this
question they have asked to calculate with the constant pressure drop of 700 kilo Pascal to
700 into 10 power 3 Pascal. So, we will apply the same formula.
616
t versus V equal to k c by 2 into V plus 1 by q naught equal to mu C alpha by 2 A square

delta P just I am putting the value of k c here, and value of q naught mu alpha R m into A
into delta P.
So, now everything is know just we need to calculate the time, to calculate the filtrate of
0.8 meter cube 1 into 10 power raise to 9 into 0.8 square 2 into 0.5 square into 700 into 10
power 3 plus 2.5 into 10 raise to 7 into 0.8 0.5 into 700 into 10 into 10 raise to 3. So, it
will about 1828.1 plus 571.42 is equal to 2400 second equal to 40 minutes, this is the time
required to calculate the about 0.8 meter cube of all the filtrate.
617
Now in the fourth question; in the fourth question there is 1 plot given plot is there 50,
100, 150, 5, 10, 15, 20, 25. This is filtrate volume ml and this is time versus volume in
second per ml. So, in this question the given data are pressure drop; pressure drop is given
10 power 5 Newton per meter square filtrate area area equal to 20 cm square, viscosity is
given 0.001 Pascal second. Mass of the cake 24.3 gram volume of the filtrate equal to 154
c m cube there was what is the filter medium resistance and filter cake resistance. So, this
is also a constant pressure drop questions. So, I am going to apply the same equation here
also.
So, from the graph, in intercept is about 3.2 second; second per ml in this I unit it will be
3.2 into 10 power 6 second per meter 6. And slope will be if you calculate it it will above
0.21 in a sign it will be 10 power 12 second per meter 6 from the slope 0.21 into 10 to the
power 12 second per meter 6.
618
So, from the intercept part 1 by q naught is equal to 3.12 for a (Refer Time: 29:38) it was
3.2 into 10 power 6 equal to mu Rm by A into delta P. Here every everything is know have
been known viscosity just I have to calculate the filter medium resistance Rm.
So, Rm is equal to 3.2 into 10 power 6 into area is given 20 in meter it will be 20 into 10
power by minus 4 into delta 10 power raise to 5 divided by mu will be 0.001. So, Rm will
come 6.4 into 10 raise to the power 11 meter inverse this is filter gradient resistance.
Coming to slope, slope was given k c by 2 equal to mu C alpha where A square delta P
slope we have got from the graph is 0.21 into 10 raise to the power 12 into 2 equal to
everything else is given 0.00 C is mass of the cake is given 24.3.
And volume of the filtrate is from that we can get the concentration in a sign at you have
to multiply it by 1000 into alpha divided by area square 20 into 10 raise to the power minus
4 square into pressure drop to 5.
619
So, from the calculation it will come alpha equal to 5.3 into 10 raise to 11 meter per kg.
Thank you.
620
Lecture - 50
Laminar and Turbulent Flows - 1
So, the last few classes, you have actually been seeing you know various applications right
where interaction between fluid and particles come into the picture. So, such situations
many of times you might have used certain relations, certain empirical correlations and so
on. So, most of them you might have actually dealt with complex geometries. What we are
going to do today is to look at flow situations which you know the geometry can be simple,
but the flow can still be complex.
So, when I say the flow being complex, I actually meant turbulent flow ok. So, earlier part
of the course, we have really looked at what is laminar flow in various geometries you
have been able to really solve the equations in an exact form and you have been able to
calculate what is the velocity profile what is the pressure drop and so on. So, we are going
to look at slightly different situation now, where the flow itself become complex even if
the geometry is very simple. So, that is where the turbulence comes in and we are going
to see you know how turbulence comes what can we do, when you have fluid flow is
turbulent what is the best that we can do so, that you can go ahead with you know designs.
So, that is the idea now hm.
So, let us say you let us take any simple case let us say you are you have flow through flow
between two parallel plates. So, that is where we actually started our class.
621
We have a let us say two plates and there is a fluid flow in it. And let us say you are
interested in finding out what is the velocity and to find out velocity you actually you are
going to do an experiment and you are going to put a probe a velocity probe to measure
velocity at a particular point let us say.
So, there is a fluid that is flowing and let us say you want to measure at some location,
what is the velocity ok. So, you have put a probe into that system and that the probe actually
gives you what is the value of velocity and you are plotting that velocity as a function of
time. So, you are continuously measuring the velocity and plotting it ok. So, we will
assume that the flow is steady ok. So, there is a constant flow rate through the system. So,
if you plot velocity at that point let me just denote it using u at some random location as a
function of time.
If the flow rate is very low ok so, when I say flow rate is very low that is really characterize
by low Reynolds number ok. So, if you define a Reynolds number and if Reynolds number
is sufficiently small, we will see how small it is a is a small. But let us say if Reynolds
number is very small and you have put the probe and you measure the velocity and plot it
as a function of time the flow is steady you are going to see that the curve is going to look
like that velocity is not changing as a function of time. So, you will get just a straight line.
Now, let us say you increase the flow rate when you increase the flow rate what happens
to Reynolds number? It is going to increase. So, let us say you keep increasing your
622
Reynolds number, but you continue to measure it. So, this is what happens if Reynolds
number is small ok. Now beyond some certain Reynolds number what you are going to
see is that if you measure the velocity, it will more or less be constant, but in between you
will see some fluctuation and then it will again come back to a constant and then again
there will be a burst and then again it is going to come back to its constant value. There
will be sudden bursts in the flow at various points.
We do not know a priori when it is going to happen, how it is going to happen, but that is
what you see you do not care you will further increase the Reynolds number let us say ok.
So, this is some intermediate Reynolds number. Now you go to really large Reynolds
number and large Reynolds number would really mean that your flow rate probably is very
high ok. If you do that where you will measure velocity as a function of time you would
see that it is just a random signal something of that sort. There is nothing like you are going
to you are not going to see anything constant about it ok.
Even though you can see that there is a mean which is constant, but it is really go in to be
a noise because the velocity is continuously changing in time. So, the left hand picture
when Reynolds number is very small that is when we call the fluid flow is laminar and the
right hand side when the Reynolds number is very large. When you see this noisy signal,
you will call the fluid flow is turbulent and in between when you see this it is actually the
transition from laminar to turbulent. So, this is what we have as a laminar flow, this is what
we have as turbulent flow and this is when we have transition between the two.
And remember this is not with respect to time this is with respect to Reynolds number that
I am talking about. Is that ok? So, let us just see what it looks like. You might have already
seen several instances. See you have heard about Reynolds experiment right.
623
So, what you have is I do not know I think this is actually a pipe where the fluid is flowing.
The fluid I think I do not know why one of one way or the other and the blue line that you
see is basically some ink that you have inserted ok. So, what would you expect? If the fluid
flow is laminar that ink is just going to be carried along with the fluid flow ok.
But if the fluid flow is very turbulent, then the fluid will see this fluctuating velocity sorry
the ink will see this fluctuating velocity so, the ink will start actually spreading to different
regions. So, that is what this experiment is; let us just watch it. So, I think initially. So, on
the right hand side corner you can see the Reynolds number. So, Reynolds number is 355,
now they are hiking of the Reynolds number.
So, the flow becomes sort of unstable, but it still smooth now this is basically the transition
where you start seeing some bursts, the flow rate is increasing you can clearly see that.
Now the burst become completely dispersed. So, that is basically the turbulence.
Display that again. So, it does not look like a pipe because the Reynolds number seems to
be very small.
Yeah.
What is that?
624
Diffusion will happen, but diffusion will happen over a very small time scale sorry very
large time scale because diffusion is a very slow process ok. So, if there was no fluid flow,
you would have seen that of course, the ink is basically you know getting diffused. But
you are not letting that to happen your fluid flow is sufficiently bigger than that ok. So,
laminar let us transition and there goes the turbulence. Now we can actually see it in a
quite a lot of situations.
Let us just check this out. So, so, this is a fluid flow that is coming from a pipe and then
you must have seen this right when you open it very lightly you basically have a very thin
you know thin stream that is coming and that will look very clean and neat if the water is
actually clean ok. But if you start increasing, then you will see that it basically becomes
hazy ok. So, that is nothing, but laminar to turbulent flow transition. So, he is just going to
put the finger I think, you can see that clean stream and therefore, you will be able to see
the finger if you look through that that is because there is no velocity fluctuation everything
is like a laminar; it is a basically laminar that is for flowing down. But if you increase the
flow rate further. I hope he shows that; maybe I chose the wrong video. So, anyway you
could try that I think this might show no this is also same.
There it is, this a nice one there you see ok.
625
So, on the right hand side you see the laminar flow and on the left hand side you see the
turbulent flow hm. So, even in a pipe in an industry this is exactly what you are going to
see ok. So, what we have done so, far let us say we calculated the pressure drop in a pipe
right we derived nice relations, but all those are useful when the flow is laminar, but not
when it is turbulent. So, there is a clear distinction, you can see that visually when now
mathematically also that is going to be a distinction and therefore, we need to know what
is the velocity profile in a pipe when the fluid flow is turbulent ok.
So, our objective now is going to derive an expression for that. So, that we can calculate
what is the pressure drop and therefore, come up with the right pump. So, that you can do
you know pump fluids yeah. So, for a pipe typically this Reynolds number where the
laminar to transition happens is approximately when Reynolds number is 2300, when
Reynolds number is defined based on the diameter which is D V some average velocity
rho divided by mu ok. So, you could define Reynolds number based on radius diameter
anything. So, when it is defined based on diameter and when it is approximately slightly
greater than 2000, you would actually start seeing the transition. So, just to give you what
are the typical regimes.
𝜌𝑉𝐷
𝑅𝑒𝐷 =
𝜇
626
It can it can start forming those bursts. So, before that you can almost certainly say that
the flow will be laminar.
So, it must be either because the fluid the this may not be a cylindrical pipe now the other
thing is that its very sensitive to external disturbances any small vibration is sufficient to
basically make it up. So, if you want to exactly fine 2100 or 2300 in a lab you need to
basically make sure that everything is vibration free and so, on otherwise it will pick up.
So, it must be one of those reasons. And it should be sufficiently far from your inlet and
so, on and what is also there at the outlet yeah.
So, just to characterize the regimes. So, 0 less than Re less than 1.
So, Reynolds number is what is going to characterize it in that regime if Reynolds number
is very small the flow is very viscous and one another name that you use is called creeping
flow. So, if Reynolds number is saw between let us say 1 and 100 the inertia of the fluid
is going to become a important, but still you are going to get a laminar flow. So, this is
basically strong inertia, but the flow will still be laminar; between 10 raise to 2 and 10
raise to 4 is based you have the transition to turbulence. In fact, you could have one more
in between let us say 10 raise to 3 because 10 raised to 3.
So, this is where you have again strong inertia, but something called laminar boundary
theory is useful and let us say if it is between 10 raise to 3 to 10 raise to 4. This is when
627
you have transition to turbulence and when it is 10 raise to 4 to 10 raise to 5 this is turbulent
and this is one could say I will explain moderately Reynolds number dependent ok.
So, the equations that we had used earlier the ones which he derived will are strictly for a
laminar flow and when the flow is laminar. So, so this is the regime that we really have
talked about the equations that we used hm.
But typically when you go to large scales you are going to deal with large pipes large flow
rates and the flow invariably turns turbulent and therefore, we ideally would like to go to
larger Reynolds numbers, but and therefore, we would really like to go somewhere there
ok. So, we need a now equations for that, just to mention that this let us say the one it is in
between ok. This is still laminar flow, but you need you know there is something called a
laminar boundary layer theory which we probably will just touch upon towards the end of
the class. So, just theories are very useful in that Reynolds number regime and this one
this the transition to turbulence, there are its very difficult to do either theory or
experiments in that regime and typically what you do is you just avoid design doing any
design in that regime ok.
So, when you do it you avoid you either stick to laminar flow or you stick to turbulent flow
where you have some knowledge of it in between if you come rather you basically change
your parameters since that you can do a better design with either of the flow ratios. So,
that is what it is. So, we want to go to that green highlighted one now any questions?
628
So, we talked about head loss you remember what is head loss?
Yeah, go ahead something where you actually had a pipe you were let us say you know
fluid is flowing from one side to another and you wanted to characterize what is the
pressure drop that is you know through a pipe for example. So, we probably would have
taken some pipe we said you have fluid that is flowing and then there is a pressure drop of
delta P and we would have calculated h f the head loss as delta P divided by rho g plus
some delta is z if you want you can turn your pages and check that the relation is correct
ok.
So, if delta is that; that means, if there is no difference in the elevation between two ends,
it is just related to the pressure drop itself otherwise is the fact that the you know the effect
of gravity will help ok. So, that is where the delta is that is coming in. So, hf is nothing,
but really a measure of the frictional losses ok. So, the moment you know hf that is what
you can prescribe and that is what you can actually use for your design. So, the idea is
typically to calculate what is hf the head loss.
Now, the way we did it is we basically defined something called a friction factor and we
read wrote down some relation between friction factor and head loss how did it relate hf
into f does this friction factor.
Something of that sort way and then we would have finally, related our friction factor to
Reynolds number that ok. So, the idea of doing this thing was that you specify the
Reynolds number that will tell you what is friction factor that will tell you what is the head
loss ok. So, this is really this way that our calculation should go ok, but though we derived
it the other way ok. Now this calculation that we have done is for laminar flow and
therefore, we would want to repeat this calculation for a turbulent flow. Now there is a
challenge. So, the velocity profile that we saw right it was very noisy it was highly
fluctuating.
So, what do we do with that? So, let us say we again have you know flow some flow in a
pipe for example, and let us say the flow is turbulent we; that means, the velocity the flow
is no more unidirectional you can really say that.
629
So, if you have it and the velocity is highly fluctuating; that means, and if I call let us say
this direction as x hm. So, typically we will write down the three components of velocity
u v and w ok. So, we have u v and w and in this case what is going to happen is that u will
be a function of x y z and t ok. Similarly v is also going to be a function of x y z and t w
is also going to be a function of x y z and t and even the pressure is going to be a function
of x y z and t you measure any of them you are going to see that fluctuation ok. So, that
may see tells that you the governing equations that we use right we cannot do all those
simplifications that we had used earlier.
So, we need to invent something better. So, can we invent something better is the question.
So, let us just write down let us say the continuity equation and the moment the
conservation equation, these are the two equations that we had and we just kept our
manipulating that. So, continuity equation in Cartesian coordinates was just dou u by dou
x plus dou v by dou y plus dou w by dou z is equal to 0 and we had the momentum
conservation let us just write down the x momentum equation that is the x momentum
equation right when do we say something is a solution to this equation? When that
expression satisfies this equation for example, we had the parabolic profile if you substitute
that profile into this equation with the right boundary conditions you will find that it
basically satisfies ok.
𝜕𝑢 𝜕𝑣 𝜕𝑤
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
630
𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑝
𝜌[ +𝑢 +𝑣 +𝑤 ]=− + 𝜇∇2 𝑢
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥
Now, what is the kind of solution that we are looking for in a turbulent flow? We are
looking for a equation which basically is going to do like this we are going to look at a
fluctuating kind of a expression an expression which is fluctuating. And as of now we have
no idea or we do not know any equations that can actually predict like that or that can be
represented with a completely fluctuating field ok. So, therefore, it is basically useless to
look for a solution, which is going to predict exactly those fluctuations because at the
moment we even do not know, but we cannot give up right we need to do design. So, what
is the best that we can do? So, the this person Reynolds right who did this Reynolds number
experiment that you are familiar with he is a something interesting about that statement no
ok.
So, see he came up with this idea called Reynolds averaging, ok. What he said is this let
us say you have a flow which was that right this one on the right hand side where the
velocity field was continuously fluctuating. We are saying that we do not know any
function that can exactly prescribe those fluctuations, but why should we care about let us
only worry about a mean value because we are only interested in the mean value. So, let
us write down everything in terms of a mean velocity because that is what is going to
interest us ok. So; that means, we need to define a mean velocity from an exact velocity.
631
So, we have u as a function of time which is highly fluctuating we said all right.
We will define a mean velocity and worry about a mean velocity how do we define a mean
velocity? u let us say mean is equal to I will say I will integrate it from some 0 to t u dt
and then I will divide it with the time interval right. So, basically I have taken a time t or
sufficiently long interval that I will keep on adding the velocities and then I add up all of
them and divide with my time period I am going to get my average velocity right. So, that
is the u bar that I will have. So, therefore, one says that ok. So, therefore, my total velocity
at any point is going to be my mean velocity plus a fluctuating velocity correct. So, u is
the one that satisfies the equations what we are interested in calculating is u bar and the
difference between u bar and u is what you call the fluctuating velocity that is all right.
1 𝑇
𝑢̅ = ∫ 𝑢 𝑑𝑡
𝑇 𝑜
𝑢 = 𝑢̅ + 𝑢′
Similarly, you can define a v bar w bar p bar everything I mean pressure mean velocity
and so on what would be u prime bar? So, what is u prime bar? Mean of fluctuating
component what do you think that would be? It should be 0. So, let us prove that is equal
to by definition is 1 by T integral 0 to T u prime dt that is equal to 1 by T integral 0 to T u
minus u bar d t that is equal to 1 by T integral 0 to T u dt minus 1 by T integral 0 to T. So,
u bar is a constant. So, I can write it down outside u bar dt that is equal to u bar minus u
bar is equal to 0.
1 𝑇 1 𝑇 1 𝑇 𝑢̅ 𝑇
𝑢̅′ = ∫ 𝑢′ 𝑑𝑡 = ∫ (𝑢 − 𝑢̅) 𝑑𝑡 = ∫ 𝑢 𝑑𝑡 − ∫ 𝑑𝑡 = 𝑢̅ − 𝑢̅ = 0
𝑇 𝑜 𝑇 𝑜 𝑇 𝑜 𝑇 𝑜
So, as expected it comes out to be 0 how about u prime square bar ? So, I am taking the
square of the fluctuating velocity and me is you know averaging it will that be 0 or
nonzero? It will be nonzero and this distinction actually becomes crucial ok. So, let us see
where it leads to any doubts by the way till so far.
̅̅̅̅
𝑢′2 ≠ 0
632
Lecture - 51
Let us start with continuity equation ok. Let us apply the continuity equation for a turbulent
flow we will just take to Cartesian coordinates.
So, I know that my velocity field should satisfy this equation so; that means, my uvw
should satisfy this equation. I am going to now split that velocity field into a mean and a
fluctuating component. So, I am going to substitute u is equal to u mean plus u prime v
equal to v mean plus v prime w is equal to w mean plus w prime so, that I can know
whether I can deal with mean values.
𝜕𝑢 𝜕𝑣 𝜕𝑤
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑢 = 𝑢̅ + 𝑢′ ; 𝑣 = 𝑣̅ + 𝑣 ′ ; 𝑤 = 𝑤
̅ + 𝑤′
Del by del x of u plus u prime plus del by del y of v plus v prime plus del by del z of w
plus w prime is equal to 0. And split that dou u by dou x plus dou v by dou y plus dou w
by dou z plus dou u prime by dou x plus dou v prime by dou y plus dou w prime by dou z
633
is equal to sorry u bar v bar w bar u bar v bar w bar. I am actually splitting it into mean
and fluctuation. See what I know is that the exact value satisfy the governing equation and
I am substituting that exact value as mean plus fluctuation and just separated it out doubts.
𝜕(𝑢̅ + 𝑢′ ) 𝜕(𝑣̅ + 𝑣 ′ ) 𝜕(𝑤

̅ + 𝑤 ′)
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝑢̅ 𝜕𝑣̅ 𝜕𝑤
̅ 𝜕𝑢′ 𝜕𝑣′ 𝜕𝑤′
+ + + + + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
Yes.
That is incompressible the fluid flow can still be incompressible. What is that? You can
have a density fluctuation we are assuming that the flow is incompressible at the moment.
In principle, you can have a density fluctuation also though for most of the fluids x; I mean
if you deal with gas you will have to worry about otherwise you will not have to worry
about hm. So, a fluctuating velocity does not mean it is a incompressible sorry it is
compressible a fluctuating velocity does not mean it is compressible ok.
You can have a compressible you can have an incompressible turbulent flow and that is
what we are looking at. So, maybe I should write it down.
Yeah, one second yeah.
So, we did when we dealt with continuity equation right we showed that. So, what does it
come out to be d rho now del rho by del t plus divergence of rho u equal to 0 that is the
full equation. So, we could split this as del rho by del t plus u dot grad rho plus rho into
divergence of u equal to 0 so; that means, d rho by dt plus rho into divergence of u is equal
to 0 and we said if you are actually following a fluid element and if the fluid element
actually has a constant density, it will be an incompressible flow.
634
Because if d rho by dt equal to 0, then continuity equation is just divergence of u equal to

0 ok.
So, this is an incompressible fluid incompressible flow. Did we discuss this incompressible
flow versus incompressible fluid ok? So, look. So, the what I have written on the right
hand side is the continuity equation right. So, del rho by del t plus del dot rho u equal to 0.
Now I am expanding the second term which is divergence of the rho u and I am assuming
that rho is not a constant, then what happens is that del rho by del t plus u dot grad rho plus
rho into divergence of u is equal to 0. So, that is just the product rule of differentiation
except that the differentiation is now a divergence ok.
And the if I combine the first two terms, I get a d rho by dt. What is d rho by dt? The
substantial derivative of rho plus rho into divergence of u is equal to 0. What does
substantial derivative indicate?
Following a fluid particle ok. So, if I am following a fluid particle that is what d rho by dt
says that and if I am seeing a change in the density of that fluid particle, then that is what
is d that is what d rho by dt will capture, but if I am following a fluid particle and if its
density does not change, then d rho by dt is going to be 0. And so, for a incompressible
flow. So, for an incompressible flow d rho by dt is 0 and therefore, the continuity equation
is just divergence of u is equal to 0. But if you want to talk about an incompressible fluid,
635
then you would simply say rho is a constant ok. But either way you are going to get
divergence of u is equal to 0.
𝐷𝜌
𝐼𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑜𝑤: = ∇𝑢
⃗ =0
𝐷𝑡
𝐼𝑛𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑏𝑙𝑒 𝑓𝑙𝑢𝑖𝑑: 𝜌 = 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
Anyway that is not important if you didnt really follow the point is the divergence of u is
equal to 0 is for a incompressible fluid and incompressible flow ok. So, we are dealing
with continuity, we have written down in the Cartesian coordinates and we have split it
into mean values and the fluctuations ok. Now, what we are going to do is we are going to
integrate this equation ok.
This entire equation we will integrate from 0 to t dt, I am allowed to do that ok where t is
some time that I have taken so, that I can average out during that time and therefore, all
the fluctuations therefore, you know whatever comes out I can put to 0. So, that is the idea.
So, if I do that and then I can do a 1 by T of course, so the first term dou u bar by dou x is
that function of time? No. So, I can just do the integration ok. So, it basically gives you
dou u bar by dou x that is actually a constant as far as a integration is concerned.
Next term is also a constant next term is also a constant plus u prime u prime is definitely
a function of time, but I interchange my integration and differentiation I will first do the
636
integration and then do the differentiation. So, I will write this as rather del by del x of
integral 0 to T u prime dt plus del by del y of integral 0 to T v prime dt plus.
Plus del by del z of integral 0 to T w prime dt is equal to 0 and you do know that this is 0
that is 0 that is 0 and therefore you basically get back that dou u bar by dou x plus dou v
bar by dou y plus dou w bar by dou z is equal to 0.
𝑇
1 ̅ 𝜕𝑢′ 𝜕𝑣 ′ 𝜕𝑤 ′
𝜕𝑢̅ 𝜕𝑣̅ 𝜕𝑤
∫( + + + + + )=0
𝑇 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
0
𝑇 𝑇 𝑇
𝜕𝑢̅ 𝜕𝑣̅ 𝜕𝑤
̅ 𝜕 ′
𝜕 𝜕
( + + + (∫ 𝑢 𝑑𝑡) + (∫ 𝑣 𝑑𝑡) + (∫ 𝑤 ′ 𝑑𝑡)) = 0
′
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 0 𝜕𝑦 0 𝜕𝑧 0
𝜕𝑢̅ 𝜕𝑣̅ 𝜕𝑤
̅
+ + =0
𝜕𝑥 𝜕𝑦 𝜕𝑧
In other words the mean velocity satisfies continuity equation that is what you have proven
ok. So, if you if you want to only calculate a mean value of velocity, you can still use the
same continuity equation without any change that is clear right. So, let us worry about
what is momentum conservation equation going to do ok. So, we will just write down
again x momentum and we will just deal with that.
637
𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑝
𝜌[ +𝑢 +𝑣 +𝑤 ]=− + 𝜇∇2 𝑢
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥
So, that is my x component of momentum conservation and I am going to substitute u

equal to u bar plus u prime plus now in some of the next terms. So, if you substitute u bar
plus u prime into del by del x of u bar plus u prime plus the next term is v bar plus v prime
times del by del y of u bar plus u prime plus w bar plus w prime times del by del z of u bar
plus u prime is equal to minus del by del x of p bar plus p prime plus del square of u bar
plus u prime that is what if I simply substitute u is equal to u prime and what happens if
we expand ?
Rho first term dou u bar by dou t plus dou u prime by dou t plus there are four terms that
are going to come up now right.
Because its u bar plus u prime times del by del x of u bar plus u times. So, u bar del u bar
by del x plus u bar del u prime by del x plus u prime del u bar by del x plus u prime del u
prime by del x let us first the four terms. And then there will be four more terms from the
next expression v bar dou u bar by dou y plus v bar dou u prime by dou y plus v prime dou
v bar by dou dou u bar by dou y plus v prime dou u prime by dou y plus.
So, should I draw things. So, this is that this is that next is oh sorry I did a mistake. So,
that is that this is that custom w bar dou u bar by dou y dou z plus w prime dou u bar by u
prime by dou z plus w w prime dou u prime by dou z.
638
I am just expanding ok; did I miss something in the equation this equation? Some important
fluid property viscosity.
𝜕(𝑢̅ + 𝑢′ ) ′
𝜕(𝑢̅ + 𝑢′ ) ′
𝜕(𝑢̅ + 𝑢′ ) ′
𝜕(𝑢̅ + 𝑢′ )
𝜌[ + (𝑢̅ + 𝑢 ) + (𝑣̅ + 𝑣 ) + (𝑤
̅ +𝑤 ) ]
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕(𝑝̅ + 𝑝′ )
=− + 𝜇∇2 (𝑢̅ + 𝑢′ )
𝜕𝑥
𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑢̅ ′ 𝜕𝑢′ 𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑢̅ 𝜕𝑢′
𝜌 [( + ) + (𝑢̅ + 𝑢̅ + 𝑢′ 𝑢 ) + (𝑣̅ + 𝑣̅ + 𝑣′ + 𝑣′ )
𝜕𝑡 𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑦 𝜕𝑦
𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑝̅ 𝜕𝑝′
+ (𝑤
̅ +𝑤
̅ + 𝑤′ + 𝑤′ )] = − − + 𝜇∇2 𝑢̅ + 𝜇∇2 𝑢′
𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑧 𝜕𝑥 𝜕𝑥
Is this simple expansion ok? What is the next step we should do? Integrate ok. Integrate
this expression from 0 to T and throw away all unnecessary terms that we have. So, we
this entire expression integrated from 0 to T dt. So, that is what we are doing. So, all the
terms that have just u prime just one fluctuating variable will go away when we do this
integration right.
But there if there are any products that are going to stay there this is the way that happened
for the continuity equation right. For continuity equation we had seen that we could
interchange the order of differentiation and integration. And therefore, all the terms that
639
were there as just primes left you know it was it was gone, but then if there was a u prime
square or a u prime v prime all of them are going to stay back.
So, therefore, I can throw away all the necessary terms that we have written down now
which would include that that this and that. So, the moment I do the integration all of them
we will go away and therefore, what I have left out is rho times dou u bar by dou t plus u
bar dou u bar by dou x plus v bar dou u bar by dou y plus w bar dou u bar by dou z is equal
to minus dou p bar by dou x plus mu del square u bar minus integral 0 to T u prime dou u
prime by dou x multiplied by rho plus rho integral 0 to T v prime dou u prime by dou y
plus rho integral 0 to T w prime dou u prime by dou z that is what we end up with.
𝑇 𝑇
𝜕𝑢̅ 𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑢̅ 𝜕𝑢′
𝜌 [( ) + (𝑢̅ + ∫ 𝑢′ 𝑑𝑡 ) + (𝑣̅ + ∫ 𝑣′ 𝑑𝑡)
𝜕𝑡 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦
0 0
𝑇
𝜕𝑢̅ 𝜕𝑢′ 𝜕𝑝̅
+ (𝑤
̅ + ∫ 𝑤′ 𝑑𝑡)] = − + 𝜇∇2 𝑢̅
𝜕𝑧 𝜕𝑧 𝜕𝑥
0
𝜕𝑢̅ 𝜕𝑢̅ 𝜕𝑢̅ 𝜕𝑢̅

𝜌[ + 𝑢̅ + 𝑣̅ +𝑤
̅ ]
𝜕𝑡 𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑇 𝑇 𝑇
𝜕𝑝̅ 2 ′
𝜕𝑢′ ′
𝜕𝑢′ ′
𝜕𝑢′
=− + 𝜇∇ 𝑢̅ − 𝜌 {∫ 𝑢 𝑑𝑡 + ∫ 𝑣 𝑑𝑡 + ∫ 𝑤 𝑑𝑡}
𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧
0 0 0
So, some of the terms stayed back right all the terms which had two primed variables which
came which were the products of the fluctuations they will not go away during the
integration they will stay ok. So, what you are seeing is that the left hand side is its early
like what we had for our you know the exact solution all the terms that had these double
primes, I have moved to right hand side for convenience and written down and you see
that if you compare with the actual the full expression that we had this is what we used to
have but now this has come out to be extra which are basically some integrals which we
of course, do not know.
640
Or another way of saying it is that, the mean velocity does not satisfy the x momentum
equation ok. Then if you want to write down the x momentum equation for the mean
velocity along with the terms that you originally had, there are going to be additional
contributions those additional contributions are in terms of fluctuating velocities or rather
products of fluctuating velocities. But those are the products of fluctuating velocities that
we are really looking for we do not know what it is. So, we actually try to get that down
get remove that, but we could not remove it they will not go away they will remain there
and that is basically the main issue with turbulent flow.
So, the whole problem comes out is can we get rid of them or can we use some other fact
to remove that and therefore, able to solve the equation is that clear? So, this will happen
for any you know we y momentum is z momentum all the equations that is clear ok. So,
basically you are now convinced that its not a simple problem to solve ok. So, therefore,
we actually need to do some alternate ways of dealing with these equations which we will
see in the next class any questions?
Yeah.
there is a dts.
641
Lecture - 52
So, we were looking at the governing equations and we realized that there is no point in
really looking for an exact solution and then we figured out that you know we would
actually look start looking at a mean quantity like mean pressure, mean velocity and so on.
But then we also found that the mean velocity the mean quantities actually do not satisfy
the governing equations that is where we ended up ok. So, that was the point and therefore
the fact that we realized was that we need to go to another framework, we cannot just
proceed the way we have been doing it for laminar flow ok.
We can do that, but before we do that I need you to introduce you to this term called
Turbulent Stress and to do that we need to look at the equation where we stopped last time.
We will see what is turbulent stress in that equation and then we will see how to attack
turbulent flow problems is that clear.
So, can you all look at the equation that we left last time for the mean flow in let us say x
direction. So, we use just u for the x component of velocity right is this right this is what
we had yeah and we realized that those terms that I have written on the right hand side
642
which actually contained the primed variable. So, the primed variables are the fluctuating
parts of the velocity field u prime v prime w prime are respectively the fluctuating
velocities in x y and z direction and they gave rise to certain quantities right. Those
quantities are extra terms that have come up in the mean flow governing equation and we
said that is actually a complication.
And therefore, we cannot really solve for the mean quantity we need to know what these
fluctuations are. So, there is a nicer way to write these fluctuations and that is what we are
going to do first, let us say we want to say we are looking at incompressible fluid we are
going to look at rho being constant. So, rho is a constant let us see what is del by del x of
u prime u prime plus u prime v prime plus u prime w prime sorry, we do not want that plus
del by del y of u prime v prime plus del by del z of u prime w prime is ok. I am just trying
to look at a quantity which I have looked at as products of the fluctuating components and
I am taking some derivatives I will I want to connect that expression to the first expression
that I have written.
So, this is and I am going to simply apply the product rule. So, that will give me u prime
dou u prime by dou x plus again u prime dou u prime by dou x plus u prime dou v prime
by dou y plus v prime dou u prime by dou y plus u prime dou w prime by dou z plus w
prime dou u prime by dou z just expanded the products. And, if you look at terms this and
that and that; that is nothing but u prime into dou u prime by dou x plus dou v prime by
dou y plus dou w prime by dou z plus, the other quantity that we have is u prime dou u
prime by dou x plus v prime dou u prime by dou y plus w prime dou u prime by dou z.
𝜕 ′ ′ 𝜕 ′ ′ 𝜕
(𝑢 𝑢 ) + (𝑢 𝑣 ) + (𝑢′ 𝑤 ′ )
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝜕𝑢′ 𝜕𝑣′ 𝜕𝑤′ 𝜕𝑢′ 𝜕𝑢′ 𝜕𝑢′
= 𝑢′ ( + + ) + 𝑢′ + 𝑣′ + 𝑤′
𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦 𝜕𝑧
So, I have written all these things wrong this is correct now, so that is what we must have
derived in the last class ok. So, the why I want to do this is because what is this that 0 and
how do we know that 0 we are talking about the fluctuating components. But we know
that the mean quantities and fluctuating quantities separately satisfy the continuity
equation we proved it in the last class ok. So therefore, this quantity is 0 and therefore what
I have on the left hand side is same as what I have on the right hand side.
643
But, if you look at these quantities on the right hand side this is what I actually have inside
the integral in all these three terms is that right or in other words what my intention is to
write down this quantity in the square bracket with this this particular one this particular
quantity I want to substitute and use there that is the idea, should I say anything should I
explain again yes no should I say no ok.
So, then I am going to substitute the quantity that I have circled dou u bar by dou t plus u
bar dot grad u equal to minus dou p by dou x plus mu the laplacian ok. The laplacian of u
I am going to write it in terms of components right, different components del square u by
del x square plus mu del square u by del y square plus mu del square u by del z square
minus ok. The quantity that is inside the integral del by del x of rho u prime square bar
minus del by del y of rho u prime v prime bar minus del by del z of rho u prime w prime
bar what have I done.
So, look at the quantities that are inside the square bracket ok. So, the quantities that are
inside the square bracket I am replacing it with this general expression that I have circled
ok. So, this as that is because we found that the quantities are same the one which I have
circled and the quantity that is inside the square bracket ok. But each of those quantities
are integrated from 0 to t and divided by t, so by definition they are all average quantities.
So, I have put basically average of each of them.
So, the first is del by del x of u prime square so this is del by del x of rho u prime square
bar it is basically the mean quantity. So, not mean quantity it is the mean of the fluctuations
or mean of the square of the fluctuation. So of course, you know that that is not zero we
talked about it last time, so that’s why you get all this nonzero quantities on the right hand
side ok.
Is equal to I am going to combine now, let us say that and that is minus dou p bar by dou
x plus del by del x of mu dou u by dou x minus rho u prime square averaged, I am going
to combine that and that to get plus del by del y of mu dou u by dou y minus rho u prime
v prime averaged I am going to combine that and that to get plus del by del z of mu dou u
by dou z minus rho u prime w prime averaged that is ok. So, yeah so I was just going to
ask you what is that quantity mu dou u by dou x.
644
𝜕𝑢̅ ̅
𝜌[ +𝑢
⃗ . ∇𝑢]
𝜕𝑡
−𝜕𝑝̅ 𝜕 2 𝑢̅ 𝜕 2 𝑢̅ 𝜕 2 𝑢̅ 𝜕 ̅̅̅̅̅̅ 𝜕
= +𝜇 2+𝜇 2+𝜇 2 − (𝜌𝑢′2 ) − ̅̅̅̅̅̅̅
(𝜌𝑢 ′𝑣 ′)
𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑧 𝜕𝑥 𝜕𝑦
𝜕
− ̅̅̅̅̅̅̅̅
(𝜌𝑢 ′𝑤 ′)
𝜕𝑧
−𝜕𝑝̅ 𝜕 𝜕𝑢̅ ̅̅̅̅̅̅ 𝜕 𝜕𝑢̅ 𝜕 𝜕𝑢̅
= + (𝜇 − 𝜌𝑢′2 ) + ( − ̅̅̅̅̅̅̅
𝜌𝑢′ 𝑣 ′ ) + (𝜇 − ̅̅̅̅̅̅̅̅
𝜌𝑢′ 𝑤 ′ )
𝜕𝑥 𝜕𝑥 𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑧 𝜕𝑧
That is tau right so that is basically we defined a constitutive relation. So, originally we
had the tau itself in the governing equation then we base r we used Newton’s law of
viscosity and replace that tau with mu dou u by dou y or mu dou u by dou x whichever ok,
so that is how these terms that you have gotten ok. So, those are stresses that is really
associated with the viscous mechanism right, those are the viscous stresses ok. Now what
we are saying is that effectively if you look at the equation for the mean flow it is not
viscous stresses alone, but there is something else also that is appearing along with those
stresses right because each of them contain.
So, the first stress actually contain something extra here, the second one also contain
something extra, third one also contain something extra, so that extra stress is known as
turbulent stress ok. So, it is the stress that is coming because the flow field itself is
fluctuating.
645
So, that is what gives rise to what you call as the turbulent stress ok. So, every place where
you actually had this stress you could say that ok, there are two stresses one is coming
from viscosity another is coming from the fact that the fluid field is fluctuating is that clear
ok. So, remember now we originally when you derived the Cauchy’s equation right, you
actually had this stress you did not know what to do then. So, you said we will make use
of Newton’s law of viscosity and therefore you converted that stress into velocity and
proceeded. So, you get everything in terms of velocity and therefore you were able to solve
the equations.
Now, you have actually ended up with a similar situation, where there are quantities which
are these primed variables which you do not know and therefore what we need is
something like Newton’s law of viscosity ok. In other words we need an expression for
the turbulent stress ok, so that is the way people have been thinking about these terms and
people try to derive various terms, none of them are as good as what Newton’s law does
ok.
So, we still do not know: what is the best way to do it there are various approaches. So for
example when you look at people talking about turbulent flow and how they solve it they
will also tell what kind of a turbulent model that they have used ok. So, that is what
generally or really the field of turbulence modeling. So, it is really an unsolved problem
we do not know what is the best equation that we want to do ok, we want to replace these
fluctuating quantities with some mean values. But we really do not know what is the best
thing to do but there for you there are certain models that you would use when you really
want to solve a flow problem.
So, if you think you know you can contribute maybe this is one of the area that you should
think about, you know where what is the best way to replace the those fluctuating
quantities or is this split type of mean versus fluctuation itself is the right way to do it.
Anyway so that is that is what I wanted to say about turbulence stresses. Now, realizing
that we cannot take this route we are basically going to use a bit of non-dimensional
analysis some insights from the experiments and then try to come up with you know
expressions and quantities that we can use in our usual design routine.
646
Lecture - 53
Turbulent Stress and Turbulent Shear Layer
So, the first thing I want to talk about is Turbulent Shear Layer. So, we are going to give
up our governing equations completely and we are just going to look at how does actually
turbulent flow look like ok; if it is near a wall and whether that information give us some
lead towards solving flow in a pipe ; that is a stuff.
So, what I have is a wall; a plain wall and there is a fluid that is flowing on top of it. I am
not saying what is on the other side; I am assuming that the other side actually goes to
infinity. If you look at the velocity profile; it is going to look something like that and it
will go to wherever it wants to you know where whatever the profile that said far away or
from at infinity. We are interested in a region where the velocity profile is affected by the
plate.
So, you can imagine that the fluid when it is near the region very close to the plate; should
experience that friction you know by the plate a lot. But when you if you go infinitely far
from that plate then you know the fluid should not see whatever effect that you see should
be independent of the plate. So, we are going to restrict to that region where the fluid is
647
seeing the plate and I am going to give let us say some length which we do not know what
it is as some delta ok.
So, delta is the length in which the fluid flow is really affected by the plate ok. And I am
going to now split this into three parts and that is been found useful; the first layer I will
just divide it and then talk about. The first layer a layer very close to the wall; what do you
think the turbulent stresses would be in that region? It would be almost negligible because
the viscous action will be very strong because the solid is introducing it is giving a lot of
friction to the fluid flow. So, that is the region where the viscous stresses will be much
larger than the turbulent stresses ok.
So, this region is where tau let us say viscous is much larger than the tau turbulent stresses
and we call that layer wall layer or viscous layer ok. Now, this is really the outside fluid;
so, that is let us call it a free stream; a free stream is basically that fluid which is not
affected by the solid wall. The region between the free stream and a little bit inside that is
called outer turbulent layer and in between, you call something as overlapped layer.
So, why I am writing down it has three different layers is because the laws that you know
govern the fluid; so, let me talk about this first. So, in the wall layer we found that tau
viscous is much larger than tau turbulent ok. How about outer turbulent layer? It will be
exactly the opposite tau turbulent will be much larger than tau viscous.
So, in the outer turbulent layer you will have tau turbulent will be much larger than tau
viscous; in overlap layer tau turbulent will be similar to tau viscous. So, that is actually the
difference between these three layers ok. We are saying that if you are looking at turbulent
flow, we need to worry about viscous stresses and turbulent stresses very close to the wall
viscous action will be so strong; turbulent stresses will be small compared to viscous
stresses.
So, you have a wall layer, but if you go sufficiently far from the wall then you will have
turbulent stresses being large. So, turbulent stresses will be larger than the viscous layer
viscous stresses and in between there may be a region where the viscous stresses and
turbulent stresses will be of a similar you know nature.
And why I am talking about these things is because now you know that in the governing
equation which actually contains the viscous stresses and turbulent stresses; depending
648
upon which one is important or which one is more you will get different solutions. So, in
the wall layer you know you are going to get some profile, in the outer turbulent layer you
are going to get some other profile, in between you are going to get something else and so
on; is that clear? Yeah.
No, it will not be viscous free the; it will definitely have viscous stresses as well less
turbulent stresses.
Student: They are almost same right?
They are almost same. So, they are equally important; both of them will be there. So, in
other way of saying it is that; so, that was the governing equation we entered away with
right.
So, let me write the left hand side also right that was the governing equation. If I am solving
for the viscous wall layer or viscous layer what this in this equation what all terms we can
I put to 0? If I am doing that I can actually put this to 0 that to 0 and that to 0 and then tried
solving it because I know that turbulent stresses will be small and therefore, you are going
to get some profile some velocity profile right.
Now, this must be undo it; undo it ok. Now if I am solving for the outer turbulent layer
that is very close to the free stream, then I know that the turbulent stresses are much bigger.
So, then I will put this and that to 0 and that then I am going to get a different solution ok;
649
in the overlap layer I cannot do either of them I need to solve for the full equation. So you
now you realize that you are going to get three different solutions ok. So, that is why we
are actually dividing it into three different parts and that is convenient because we anyway
do not know how to solve the full thing that is why hm.
Doubts?
So, we are primarily going to depend upon experimental results and dimensional analysis.
So, the first thing that we are going to analyze is the wall layer and this is suggested by
Prandtl in 1930 ok.
He said let us try to do write down some expression for the velocity profile, but we will
just do a non dimensional analysis. And he said the velocity profile is just going to be a
function of how far it is from the wall that makes sense because it has to depend upon
where it is ok. And instead of talking about how you are driving the flow, we will talk
about a wall shear stress. So, you remember we talked about wall shear stress, we
connected the wall shear stress to pressure drop and so on.
So, instead of talking about what is driving the flow; we will really talk about what is the
force that is acting on the wall ok. We will use that as the quantity because that is a useful
quantity to talk about and that is what is really responsible for the flow. So, we will talk
about that driving force in terms of tau w and fluid properties like viscosity and density; I
650
cannot neglect viscosity because as we are talking about the viscous layer now or the wall
layer ok.
So, this is what actually he experimentally found also. So, he says that they should
therefore, exist some relation connecting these quantities. So, if you write down some
expression like this; now you know how to proceed right, what is the next thing to do? We
need to do the non dimensional analysis. Can you do the non dimensional analysis and
find out what happens, what is the x what is; what are the two non how many non
dimensional numbers would you expect?
How many variables are there?
Student: 5.
5, how many fundamental dimensions are there?
So, how many non dimensional numbers? Yeah, let us get them.
Student: Sir.
Yeah.
Yes.
Viscous stresses will be; so you are talking about.
Free stream right? Yeah, yeah.
That is right, it will be very close to what is happening in the outer turbulent layer. So,
that; so that we are we are keeping it free stream because that is a region where which is
completely unaffected by walls. So, for example, if it was a uniform flow that is coming
ok, we would say that free stream is basically uniform it will not see the wall, but in the
outer turbulent layer you will still see the wall because that is the region where things are
changing from the overlap to the completely uniform region. So, there will be some region
where we could expect that the turbulent stresses are much larger, but still it is affected by
the wall yeah.
It is basically going to be the sum of both tau w ok. So, it is basically the stress this is
basically the force that you need to apply to maintain the flow though in this. So, by this
651
analysis already tells you that when you talk about just tau w, we will be able to talk about
only a viscous contribution to that because tau turbulent will be much smaller there. So,
dimensions of u; L by T, y is length, tau w is Newton’s per meter square. So, m L T square
L square; so m viscosity is gram per centimeter second m by L cubed.
𝑢 = 𝑓(𝑦, 𝜏𝑤 , 𝜇, 𝜌)
𝑢 𝑦𝜌 𝑦√𝜌
=𝑓 = 𝑓( )
√𝜏𝑤 /𝜌 𝜌 𝜇 √𝜏 𝑤
𝜇√
( 𝜏𝑤 )
I just used epsilons method you could do whichever you like, you could end up with some
other combinations of these two quantities that I have written based on what method you
have done. Yeah. Yes.
Yeah, but you do not know what is then other quantity right; let us say the u prime v prime.
So, for example, what is the correlation between u prime and v prime ok. So, there is no
real way to see that though there might be this convention issue again what is this stress
ok? So, do you want to write your stresses minus mu dou u by dou y which case you would
see that they are adding up. So, there must be much something more to that.
652
So, it is going to take away; so, in terms of momentum transfer this is going to add up to
your momentum transfer mechanism. So, if you think about how we interpreted viscosity;
we said it is the fluctuating motion of the molecules that was resulting in viscosity.
Now, we are saying that the fluid itself is fluctuating ok. So, therefore in addition to viscous
mechanism there is this other mechanism that also will transfer momentum across ok; so,
that is really the meaning of turbulent stresses. So, if you have wall and you are applying
a force that forces; not only transferred because of the molecular motion, but also you
could think about it as the fluid itself is basically fluctuating and therefore, things get
transferred much faster ok. And you will start seeing more and more uniform things than
you know things having a gradient and we will see that as an example.
You got some form like that; I want to write this because that is going to be useful ok.
Few more terms or definitions; root of tau w by rho which is really a measure of shear
stress on the wall is known as friction velocity and you typically represented it as u star
what a bit the star alright.
And therefore, my relation is u by u star is equal to f of y; rho, u star by mu. So, if I replace
u root of tau w by rho it is just a u star. And u by u star is typically represented as a u plus
and this is represented as a function y plus, where y plus is y divided by what will be mu
by rho u star ok. So, this is just and therefore, u plus is equal to a function of y plus; it turns
653
out that it is a basically the right or the; so u plus and y plus are really non dimensional
numbers ok; that you would want to use to represent the velocity profile in the wall layer.
𝜏𝑤
𝑢∗ = √ → 𝐹𝑟𝑖𝑐𝑡𝑖𝑜𝑛 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦
𝜌
𝑢 𝑦𝜌𝑢∗
𝑢+ = = 𝑓 ( ) = 𝑓(𝑦 + )
𝑢∗ 𝜇
Just you could have written it in terms of the non dimensional numbers that you have
gotten I am just following some textbook. I have not done anything I have just written
down u is proportional to some quantities and non dimensionalized. And when I got a
quantity called the root of tau w by rho, I called it friction velocity; is this actually does it
have the units of velocity? You could verify that put stress divided by density and square
root of it ok. Should I explain anything again? Should I go through again? Just going to
say a few things ok.
So, what have we been doing this let us revise; we started with looking at the governing
equations; we wrote down a expression for the mean flow. We realized that the mean flow
governing equations now not only contains viscous stresses, but it also contains turbulent
stresses. And therefore, we need some sort of a constitutive relation for turbulent stresses
which we do not know ok.
And therefore, we are leaving it at that point and then trying to do something simpler and
to do that we started by looking at turbulent flow very near to the wall ok. And then we
are saying that look I can take that region where the wall is affecting the turbulent flow
and I may be I will be dividing it into three different regions, depending upon what is the
magnitude of viscous stresses versus the magnitude of turbulent stresses; that helps me to
identify three regions.
One region in which tau viscous is much larger, in the other region tau turbulent is much
larger and in between they are of comparable magnitude and depending upon which one
is more or less, you are going to get different solutions ok. So, now that is; so the first
region we call wall layer, then there is an overlap layer and then there is an outer turbulent
layer ok.
654
And then what I did is that now I will go with what Prandtl said in 1930 and he says that
look the velocity profile in the wall layer can be written down as a function of few
variables, which is distance from the wall and the thing that is driving which is tau w and
the fluid properties such as viscosity and density. And therefore, we wrote down that
equation and when we write down something like that the first thing that we should do is
to non dimensionalize and reduce the number of variables; we found out what is the non
dimensional form of that function.
And that came out to be just u plus is equal to f of y plus where we do not know what is f
at the moment ok. And u plus and y plus are just two non dimensional variables and during
this calculation, we also define something called a friction velocity which came out to be
root of tau w by rho; which is happened to have the dimensions of velocity; it is no velocity
at all, it is just a magnitude of tau w that is all that is all we have done.
So, having done the wall layer; the next thing that we want to do is the outer turbulent
layer.
And for that ; so these; all these calculations are essentially inspired by experiments ok, or
rather led by experiments when you change things and see what is happening and therefore,
you try to write down the expressions.
655
So, what Karman suggested is that let us say in your free stream ok; the outside the
turbulent layer the velocity is u; some u which could be a function of space. He said let us
write down U minus u; so that is the free stream velocity minus velocity in the turbulent
layer as a function of; of course, distance from the wall ; the wall shear stress which is
driving it. But this time the viscosity is not going to come into the picture; you will still
have density ok. And he says this is also going to be a function of delta which we defined
originally where delta refer to the total height of that entire layer ok.
So, this is now if you ask me why he wrote down that, one can only say that it has been
seen in experiments and therefore, you are writing down; I do not have any more
explanation to that. So, he suggested that these are the quantities that seem to be dependent
on the quant U minus u ok; the free stream velocity minus turbulent velocity is a function
of few other things. And therefore, you could again non dimensionalize and try to write
down a non dimensional expression for the outer turbulent layer; can you do that?
(𝑈 − 𝑢) = 𝑓(𝑦, 𝜏𝑤 , 𝜌, 𝛿)
So, U is the velocity of the fluid that is outside this entire layer that we are considering and
small u is the velocity in that layer. So, we are really finding out how; what is the difference
of that layer compared to the outside layer, so it is just that difference. And I do not think
it would make any difference if you were to write it in terms of u, but it is conveniently or
this away he suggested; so, and therefore, this is called a defect low. In fact, I mean you
know basically it is how much is that difference we will see that yeah.
Why we are taking? So, y is going to be the exact location ok; delta is going to be the total
length; the total length that is affected by the wall ok. Now, so as I said that I mean I have
no answer; why it should be a function? But this is what is see and therefore, we are putting
it into the expression.
No; no delta is the total height. So, delta is the total height of the fluid layer that is affected
by the wall. In fact, later you will learn that that is called a boundary layer thickness, while
y is going to be the exact position. So, you can expect that you should get something
proportional to y by delta really like you know radius of the pipe or whatever.
See, we will get something like this just u by u star; U minus u by u star is a function of y
by delta y by delta is the only other parameter that is going to come. So, this expression u
656
plus is equal to is a function of y plus is called as Wall law and this is called a Defect law
or ; yeah Defect law. So, so this is the region where Wall law is applicable, that is the
region where we have the Defect law applicable and therefore, this region which I am
shading is blue should be a place, where things change from this blue to the green ok.
𝑈−𝑢 𝑦
∗
= 𝑔( )
𝑢 𝛿
So, the expression that should be applicable in the overlap layer should be things, it should
be an expression which smoothly changes from the Wall law to the Defect law ok. So, that
is the way it was interpreted and therefore, people were actually able to derive an
expression for overlap layer. Some expression that will match with these two regimes ok.
So, I am talking about something in between where if I say that if my y is very small; then
it should be like my wall layer. If my y is very large, it should be my thing you know defect
layer; so what could be something in between ok. We do not derive it we are just going to
take the result and proceed I will write down.
So, this was actually done in 1937 by Millikan and he said; if you write u by u star is equal
to 1 over K lon y u star divided by nu plus B; this is the so called Overlap law. This Overlap
law will smoothly merge into the Wall law and the Defect law ok.
657
𝑢 1 𝑦𝑢∗
= ln ( )+𝐵
𝑢∗ 𝐾 𝜈
Now, nu what is nu?.
Mu by rho; this is also called a logarithmic law and therefore, sometimes this is referred
as logarithmic overlap layer, there is a log function connecting the velocity with the
distance. It is found out that experimentally K is approximately equal to 0.41; B is
approximately equal to 5.0 hm. So, in some sense what we have done is; we just looked at
a turbulent shear layer and we wrote down three different expressions, each applicable in
a layer ok. And we will try to see how these expressions can be utilized further to analyze
turbulent flow in a pipe and we will see that in the next class.
658
Lecture - 54
Turbulent Flow near a wall and in a pipe
In the last class we started analysing Turbulent Flow, we looked at that turbulent velocity
profile near a wall right and we had talked about three different layers of flow field.
So, we are looking at. So, that is the wall we have and then for convenience or rather we
found that the structure of the turbulent flow near the wall can be divided into three parts
the first layer is what we called as the wall layer and in the wall layer the viscous stresses
are much larger finished, the layer which is very close to the wall where we called as the
wall layer where the viscous stresses are much larger than the turbulent stresses. And we
had from dimensional analysis we said u plus is a function of y plus where we had defined
u plus and y plus as the two non dimensional relevant variables as a velocity variable and
a distance variable.
And then we had the overlap layer where we had a logarithmic law that we wrote down
which was u by u star is equal to 1 over k ln y u star by nu plus B and then we had an outer
turbulent layer where we wrote down a defect law. So, that is what essentially we did in
the last class. So, what I am going to do now. So, I am just going to draw a diagram
659
showing the velocity profile that is there in these three layers and I am going to do that
using u plus and y plus. So, u plus and y plus are the velocity and the distance variables
that I am going to use now throughout and I will keep my x axis as log and y y axis as a
linear scale ok.
So, the first part that I am going to draw is the velocity profile in the wall layer and that is
where I have u plus equal to a function of y plus we really do not know at the moment
what is f, but it is been now seen experimentally that u plus is approximately equal to y
plus is a good enough approximation ok. In other words in the wall layer the velocity
profile is simply u plus is equal to y plus ok. So, if you use that; so you are basically going
to get something like that its a log linear plot, so it would not be straight line ok. So, that
is what the wall layer would look like.
Then we have the overlap layer in the overlap layer the law is already logarithmic. So, in
this plot its going to look like a straight line. So, I will have; so it will be a smooth
connection I will have a logarithmic layer and then you will have the outer layer which we
do not know something else. So, that is my wall layer, this is my overlap layer and that is
my outer turbulent layer and here we already know the law. So, this the law that is
governed by is that here in the outer layer that is what we have plotted ok.
Now, the point is that this outside layer the outside layer depends upon what is really your;
so this could be multiples. Meaning that one could be a pipe flow, one could be a flow on
a plate and so on. So, whatever be the main flow that you have that is seem to be affecting
only the outer layer the overlap layer and the wall layer are independent of the flow
configuration that you really working on ok.
So, the wall layer which is dominated by your viscous stresses and the overlap layer is
really independent the profiles are independent of what is the outer flow field the outer
flow field is the only when that is really dependent upon the exact configuration that you
are looking at. So, that is why I have plotted a wall layer I have plotted this one line for
the wall layer and for the overlap layer because they are like universal profiles and then
things basically start changing all right.
And this wall layer in this coordinate y plus is approximately up to 5 and the overlap layer
is found to be approximately up to 30 ok. So, that is these are all experimental
measurements this is now way to really calculate that if you write it down in terms of y
660
plus up to y plus of 5 you will see that you know u plus is equal to y plus is a good
approximation. And from y plus being 5 to 30 you would see that u plus goes as
approximately logarithmic of 5 plus and beyond that these neither of these two laws are
there will you will start seeing different different profiles.
Why u plus is equal to y plus is between y plus being 0 to 5. So, in that region
approximately u plus is y plus. So, see only thing that we derived was u plus is a function
of y plus we did not know now I am saying that experiment show that u plus is equal to y
plus. We did not show u plus is equal to y plus, we calculated u plus as a function of y
plus. Any other questions?
I do not think it is really necessary it could be anyway, but I think these profiles are
probably stable. So, typically I think you would see like this you can see that its. So, for
example, it was laminar it would have been parabolic right. So, its some signatures of like
that the curvature should be like that the other curvature I think the flow the flow is never
stable that is why you do not see it.
So, that is actually the important point even though we introduced overlap layer as
something which is to connect these two regimes, it turns out that overlap layer is very big
and it is a straight line its a log its a straight line than log linear scale ok. So, the logarithmic
law is a very robust law it turns out and it spans quite a length as you see its not a small
region. In fact, what we are going to do now onwards is to going to skip the wall layer and
the outer layer I am going to stick or use the result of just logarithmic layer ok.
So, now we are ready to look at turbulent flow in a pipe. What is our ultimate aim? Our
aim is to get the pressure drop which we would write it as using friction factor and we
want to relate friction factor with Reynolds number that is our aim. So, f was is equal to
64 by Re was the relation that we found for the laminar flow we want to now get a similar
relation for turbulent flow. So, that given Reynolds number we can calculate the pressure
drop.
661
So, what we are going to assume now is that; so again. So, the first approach that we had
right starting from the governing equation is going to be impractical. So, what we are going
to do is having seen that this is the particular structure of the you know turbulent flow and
wall layer being very small we are going to assume that overlap layer is majorly the you
know the velocity profile inside the pipe. So, in the pipe if you have the wall layer will be
very near to the wall right and then we have a large region of the overlap layer.
So, let us say that we will just use the overlap layer as a velocity profile and try to see what
is it going to look like and how does it match with experiments and what we are going to
see is that its going to be a robust approximation or the approximation that the entire
turbulent velocity profile is given by that logarithmic law is a good approximation, but we
have to do it and see that its a good approximation ok. So, I am assuming that the entire
layer is now just going to be given by the overlap layer law or the logarithmic law or in
other words what will I write the velocity profile will be now given by u by u star is equal
to 1 over k ln y u star by nu right. So, y u star by nu plus b.
So, this is the logarithmic law that we wrote down. So, now, I am going to talk about a
pipe whose radius is R and let us say we are using going to use a cylindrical coordinate
system in which r is the radial coordinate. So, y I will replace with R minus small r because
y is the distance from the plate. So, in this coordinate system R minus r will be the distance
from the plate. So, this is the approximation that we are doing. R minus no we had defined
662
y to be the distance from the plate. So, now, the distance from the plate would be just
capital R minus small r.
𝑢 1 (𝑅 − 𝑟)𝑢∗
= ln +𝐵
𝑢∗ 𝐾 𝜈
U star; look at your notes and see whether this is consistent whatever we wrote in the last
class should be consistent with this expression. So, you feel a little more comfortable with
that expression. Where u star was defined as? What is the name of what is the what is it
called?
Friction velocity which is defined as square root of ok. So, u star is the friction velocity
ok. So, if this is the case let us actually calculate the average velocity in the pipe ok. So,
right now we have just taken it as an approximation average velocity. How do we calculate
average velocity? Let us call it let us say v is equal to 1 by. So, its a circular pipe pi R
square integral 0 to R u into 2 pi r d r is that right. So, this is the velocity integrated over
the entire surface divided by the surface area. So, we know the expression for u. So,
substitute and integrate and see what the velocity average velocity is.
𝑅 𝑅
1 1 𝑢∗ (𝑅 − 𝑟)𝑢∗
𝑣= ∫ 𝑢. 2𝜋𝑟 𝑑𝑟 = ∫ ( ln ( ) + 𝐵𝑢∗ ) 2𝜋𝑟 𝑑𝑟
𝜋𝑅 2 0 𝜋𝑅 2 0 𝐾 𝜈
663
You can just go ahead and integrate it will take a couple of steps. So, that is what you need
to integrate now right, there are two different terms one containing log and then there was
a constant multiplied by R.
𝑣 1 2 𝑅𝑢∗ 3
= [ ln ( ) − + 2𝐵]
𝑢∗ 2 𝐾 𝜈 𝐾
Where you get something like that that is what I. So, and the typical value of K is 0.41 and
B is 5 and if you substitute you get V by u star is equal to 2.44 ln Ru star by nu plus 1.34
664
ok. So, that is the average velocity that you get for a turbulent flow in a pipe if you assume
that its given by the overlap layer completely. Now, we are directly going to now evaluate
the friction factor.
𝑣 𝑅𝑢∗
= 2.44 ln ( ) + 1.34
𝑢∗ 𝜈
So, there are two ways of doing it one is what we followed in our laminar flow calculation
where you know calculate the pressure drop connect it with the turbulent sorry the wall
shear and then do it or you can do it in a simpler fashion here. If you look at this quantity
v by u star is v divided by u star was defined as square root of tau w by rho, so; that means,
v by u star square is rho v square divided by tau w correct. And what is the connection
between tau w and the definition of friction factor? You can just check your notes and tell.
𝑣 𝑣 𝑣 2 𝜌𝑉 2
= → ( ∗) =
𝑢∗ 𝜏𝑤 𝑢 𝜏𝑤
√𝜌
In terms of wall shear stress, it will be there where we introduced friction factor for the
first time. Somebody says 8 tau w by rho v square check whether this is there now in the
notes. So, that means, tau w by rho v square is what or rather rho v square divided by tau
w is equal to 8 by f right. So, if I can actually directly therefore, write down a relation v
by u star is equal to square root of 8 by f agreed. And why am I writing down because I
665
already know the expression for average velocity, but the average velocity the relation
between average velocity in friction factor we now know. So, therefore, I can immediately
say that the expression that we really have this expression is nothing, but root of 8 by f is
equal to 2.44 lon not y r; R u star by nu plus 1.34.
8𝜏𝑤
𝑓=
𝜌𝑣 2
𝑣 8
∗
=√
𝑢 𝑓
8 𝑅𝑢∗
√ = 2.44 ln ( ) + 1.34
𝑓 𝜈
We really want to get it in terms of Reynolds number, so, we want to make that logarithm
in terms of Reynolds number, can you make that logarithm argument of that logarithm in
terms of a Reynolds number? The argument R u star by nu, so u star if I replace with square
root of tau w by rho and again write it as tau w rho v square where v is the average velocity
I can write that square root term in terms of f again. So, basically I get Reynolds number
times square root of f by 8 that argument. Check whether its right.
666
𝑅𝑢∗ 𝑅 𝜏𝑤 𝑅𝑣 𝜏𝑤 𝑅𝑣 𝑓
= √ = √ 2= √
𝜈 𝜈 𝜌 𝜈 𝜌𝑣 𝜈 8
In other words what we are saying is simply square root of 8 by f is equal to 2.44 logarithm
of Reynolds number based on the radius times square root of f by 8 plus 1.34. If you rewrite
1 by square root of f is equal to 1.99 log Reynolds number based on the diameter times
root f plus sorry minus 1.02, that is the relation that we wanted that is the relation between
Reynolds number and the friction factor.
1
= 1.99 ln(𝑅𝑒𝑑 √𝑓) − 1.02
√𝑓
So, let us just go through what we have done and then may be. So, we started looking at
the flow through the turbulent flow through the pipe and we said the velocity profile is
simply given by the overlap layer log which is the logarithmic law. We assume that that is
the case and then we went ahead and calculate an expression for the average velocity.
And the expression for the average velocity came out to be that average and expression for
the average velocity came out to be that ok. Our intension is to calculate friction factor, so
we wanted to calculate friction factor which we could have done by calculating the
pressure drop and calculating the wall shear and so on. Instead of that what we have done
is we knew that average velocity wall shear stress friction factor are all related.
667
So, we are essentially exploiting that relation which is nothing, but that and rewriting the
velocity profile in terms of friction factor and Reynolds number directly that is all we have
done. And therefore, we ended up with this expression which is 1 by root f is equal to one
approximately 2 ln Re root f minus 1.02. As suppose to what was in laminar flow? That I
am putting the suffix d to just show that the Reynolds number is calculated based on the
diameter and not on radius ok. So, here you can actually see given Reynolds number if
you want to calculate friction factor in laminar flow it was much easier, but how is it in
turbulent flow? There. So, I have changed it. So, here we had R v by nu which you can
write it as Reynolds number based on the diameter divide by 2.
This is Re d. So, there will be log of root 2 that will come which I think is adjusted in that
pre factor you can check.
So, I what I wanted to tell was given Reynolds number how would you calculate f using
this expression as a straight forward?, Its a non-linear expression which contains a square
root and a logarithm ok. So, what can you do, how do you solve how do you solve non-
linear algebraic equations? Go ahead you want to tell something. Graphically is that the
only method that you have learnt in your third semester, tell? Numerically what methods
have you learnt?
I do not know the probably the best is Newton Raphson ok. So, this is a nice expression
where you can apply all those methods and calculate its a very complicated equation, but
its an extremely useful equation therefore, typically those problems are illustrated using
this equation. So, that is where you will see the use, anyway so, this equation.
So, remember we assumed overlap the logarithmic law and calculated it, but instead of
that let us say you wanted to calculate sorry the it is compared with experiments and the
actual coefficients have been replaced a little bit 1 by root f is equal to 2.0 ln Re d root f
minus 0.8 ok. So, this seems to be fitting very well with the experimental data which means
that our original approximation of replacing the velocity profile just with the logarithmic
law was a good approximation because we have ended up with equation very similar to
what is found experimentally ok.
668
So, that is the only justification for replacing the or using the logarithmic law. So, you
should be really using this expression if you want to calculate a friction factor which is a
straight forward relation ok. So, if Reynolds number is not very large which is let us say
10 raise to 4, 10 raise to 5 then you could use this direct relation which is a empirical fate.
So, this is going to be useful because I want you to calculate what is delta P proportional
to the let us see which function we let us do velocity first velocity to the power of question
mark, what is that question mark for laminar flow and turbulent flow can you find out?
669
So, what I am saying is that there is a another equation which is valid over a small region
ok. So, that is a useful approximation. So, use this useful approximation and tell me delta
P is proportional to what power of V? For laminar flow and turbulent flow. So, for laminar
flow of course, you should not you should use your older expressions that we already knew
and turbulent flow you can use this nicer expression 0.316 times Re to the power of this.
You know question like for example, you double the flow rate you double the velocity
what happens you pressure drop you want to calculate what would you do that is what I
am asking.
So, the first thing is to write delta P in terms of friction factor that is our basic definition
substitute for f and see what is the power of V that you get. So, for laminar flow f is that
turbulent flow f is. So, its almost square dependent ok. So, if you change the if you double
the average velocity laminar flow the pressure drop will increase it will double, but for
turbulent flow it will go by 4 times almost.
If the flow rate is fixed what is delta P proportional to d to the power of question mark
where d is d for a for a laminar and for the turbulent flow ok. Flow rate is fixed through
the pipe what is going to be the dependence of pressure on diameter or radius. Are you
following what I am saying which is just trying to explore with look at those relations and
see what the areas dependencies are ok. So, you will see this that it goes d to the power of
670
minus 4 for laminar flow and close to minus 5 for turbulent flow ok, so; that means, that
pressure drop decreases as you increase the you know the size of the.
Choose the volumetric flow rate. So, see you want to term something fixed and now you
have the choice of selecting the pump as well as the size of the pipe what is that you would
want to do would you take a smaller pipe or would you smaller pipe or would you take a
bigger pipe? If you take a smaller pipe then it says that the pressure drop is going to be
larger and larger ok. So, then what you would want to do is you want to reduce your
electricity consumption you would want to go for a less pressure drop when you would
say that you will take a larger pipe, but what is the issue with taking larger pipe? The initial
cost becomes extremely high maintenance will become large ok.
So, therefore, there is a competition now, when you are asked to design something whether
you want to reduce your electricity usage or whether you want to reduce other cost ok. So,
that is the relevance of this relation and of course, laminar and turbulent whether that flow
rate also makes a difference ok. This I have written down a relation at the end U max that
is the maximum velocity divided by the average velocity. What is maximum velocity by
average velocity for a laminar flow?
So we did that or you did it in one of these quizzes. So, show that its actually this factor
for your turbulent flow just use the expression that you already have and try to manipulate
and you should try to get this relation. Try it we will talk more about this tomorrow.
671
Lecture - 55
Effect of rough Walls
So, we are looking at flow turbulent flow in the pipe that is what we just did in fact, right.
We calculated V average divided by or other V max. So, what else then you may write V
average; V average by V max is equal to 1 divided by 1 plus 1.3 square root of f ok.
So, if you have not gotten it, just go back and try and do it ok. I just wanted to know,
whether you have done your homework or not. So, what is that? Yeah, so V average by V
max is this for the turbulent flow. So, the difference is that so, we talked about this
yesterday, V average by V max is half for laminar flow and this is the factor that comes
out for turbulent flow. So, let us see what this means ok. Typically, let us say if I take a
Reynolds number of 10 raised to 4, then V by V max comes out to be 0.8.
If I take Reynolds number as 10 raise to 8, then V by V max comes out to be 0.9 ok, just
some number. So, given Reynolds number you know, how to calculate friction factor, we
derived the expressions for that. So, if you use that friction factor and substitute, this is
what you get ok.
672
So, what does it mean; as Reynolds number get larger and larger V by V, this average
velocity divided by the maximum velocity becomes closer and closer to 1 ok. When can
average velocity become very close to the maximum velocity? When the profile is very
flat ok, when the profile is parabolic, you had that factors 2. Now, this is that the average
velocity is very close to maximum velocity; that means, if you look at the velocity profile,
the velocity profile is going to be almost flat, something like that. So, that is what the
velocity profile going to look like.
So, that is the information you have really gotten. And in some sense now, this should start
making sense to you, because turbulence flow or turbulent flow is full of fluctuating
velocities. There are extra mechanisms for momentum transfer. So, the gradients are not
maintained that well ok. The gradients get actually smoothened out very fast ok, that is
why you are getting this flat profile for the turbulent flow, is that clear, doubts ok. So, what
we are going to look at maybe next 10 minutes is some effect of rough walls.
So, when I say roughness so, you let us say you take any you know any material let us say
even if it is P V C pipe or iron pipe the finishing that typically, looks very smooth, but
when do we call it smooth? When do we call it rough?
Here, when you look at it you may not see that it has got any roughness at all, it may be
like very small, but if you actually put it under a microscope you will at see that it is not
really flat it has got some you know troughs and some you know uneven area right that
673
you would definitely see. So, what happens is that even though it is such a small scale, it
starts affecting the flow field. It does not effect a laminar flow, but it can effect turbulent
flow. We will see why it effects turbulent flow.
So, rough walls does not affect laminar flow, affects turbulent flow ok; so, to see that we
need to go back and check our viscous sub layer or the wall layer. Remember, we said
there are if you look at the turbulent flow near a wall we divided it for a into 3 parts right.
There was a layer very close to the wall, where the viscous stresses are much larger than
the turbulent stresses right. So, that is what we call the wall layer and we do a velocity
profile and we said up to what extent the viscous sub layer or the wall layer is right, does
anybody remember the number up to what distance would you see the wall layer?
Check your notes; 5 what is 5 y plus ok, when y plus if 5 up to 5. So, from 0 to 5 you are
expected to see the viscous wall layer. So, let us say y plus is equal to 5. So, from 0 to 5 is
the viscous wall layer. So, let us say y plus maximum is 5 that you are supposed to get.
So, let us write it as y, how is y plus defined? What was it non-dimensionalised with respect
to? Is y max, is 5 or y max divided by let us say D.
Where, D is now the diameter of that pipe is simply 5 u star divided by mu right is equal
to 5 sorry, I made a mistake no this is y divided by there is a D here. So, I am just dividing
D on both sides, because I am trying to what I am going to calculate now is the thickness
of the wall layer compared to the diameter of the pipe. So, I am looking for a fraction
which is y max divided by the diameter of the pipe yeah. No. So, y so u plus is equal to y
plus.
So now, y plus is up to 5 is when you have the viscous layer it is just that we do it on a
logarithmic axis. So, the value is still 5 yeah, it is not log y plus. What is that?
Is it the other way? Let us look at y u star by mu. Now, is it correct? y u star by yeah. So
therefore, I should change this nu by u star is equal to 5 nu divided by no, in the u star
there ok. I want to bring in an average velocity, because I want to write it down in terms
of Reynolds number star, let us say we define a u average D and I will put a u average
upstairs also.
So, that I can write it as 5 times nu divided by D u average, which will become Reynolds
number times u average divided by u star is equal to 5 times 1 by Reynolds number times
674
u average by u star, we wrote down an expression for u average by u star yesterday. What
was that? Not u max ok, not the one that you calculated. Now, use u average by u star.
In terms of friction factor you would have it in your notes, root of f by 8 by f 8 by f root
of 8 by f y max by D ok. So, that so, this is what the thickness of the wall layer is going to
be compared to let us say the diameter of the pipe, let us say you take a pipe of 10
centimetre diameter.
+
𝑦𝑚 =5
𝑦𝑚 𝑢∗ 𝑦𝑚 5𝜈 5𝜈𝑢𝑎𝑣𝑔 𝜈 𝑢𝑎𝑣𝑔 1 8
=5→ = ∗ = ∗ = 5( )( ∗ ) = 5( )√
𝜈 𝐷 𝑢 𝐷 𝑢 𝑢𝑎𝑣𝑔 𝐷 𝐷𝑢𝑎𝑣𝑔 𝑢 𝑅𝑒𝐷 𝑓
And, let us assume that Reynolds number that I have is 10 raise to 5 ok. So, a factor
Reynolds number of 10 raise to 5 can give you a friction factor 0.018 and therefore, you
must be able to substitute and calculate what is y m ok. So, y m by D comes out to be
0.001. So, what am I trying to do? I am just trying to estimate how big the wall layer is
going to be ok, in terms of some real number. So, I am just giving some numbers and
trying to see how big it is going to be for a Reynolds number of 10 raise to 5 and I find
that that max the high height maximum compared to the diameter is going to be .001. So,
that would mean that if I have a 10 centimetre pipe, if D is 10 centimetre y m is going to
be 0.01 centimetre ok, what is that 0.1 millimetre ok.
675
So, if I take a pipe let us say 10 centimetre diameter ok, the thickness of the wall layer is
going to be very tiny ok, less than a millimetre, very close to the surface of the pipe ok.
So, that us the number that comes out to be so, it is really thin, one thing is that remember
we yesterday, when we did the calculation we threw away all that thing that was coming
from the wall layer and the overlap layer did the logarithmic layer did very well is 1 is 1
reason is this that the contribution that is going to come from is extremely small.
Second is that the thickness is so small that would so, if you go to that scale ok, if you start
zooming in into the thickness of that wall layer would you see the roughness or not you
would definitely see it right. In other words, if you look at the wall layer the wall layer is
not really on a flat plate. It is actually seeing all the crusts and troughs and it is actually
moving on a very uneven surface agreed ok, because the thickness itself is very small.
So, that and then if some there is a flow that is happening on a very rough surface that flow
is going to get affected by the roughness, that flow is not going to be smooth. Actually
another way of saying it is that that wall layer itself gets destructed, because of the
roughness, the roughness that you might not have imagined would play a role actually that
destroy your wall layer and will affect the entire flow field ok. Therefore, pipe roughness
effects turbulent flow that is the take away point, is that clear. Yeah. So, because it affects
the wall layer, the other layer is on this layer.
Because of the continuity yeah so, for example, you need to maintain some stress balance
that will get destructed. So, the entire thing will get destructed. In fact, the log layer will
remain as a log, but you know the arguments and the constant everything will get changed.
Therefore, the friction factor expression that you have gotten should be where re looked at
in talking the context of the rough pipe yeah.
So, for the laminar flow, it is basically going all the way there right. So, the layer is there
is nothing like a wall layer that is confined to the wall. There may be a small region where
the roughness could affect it, but basically the viscous action will immediately smoothen
it out, because it is you know spanning all the way up to the radius. So, that is why it does
not get affected.
676
Lecture - 56
Roughness in Turbulent Pipe Flow
So let us start. So, what did we learn yesterday? We were looking at Turbulent Flow in a
pipe right and we realized that something that you do not see, ok. So, visually when you
see it you do not perceive it really that the small amount of roughness the small unevenness
that be there on a pipe can affect your turbulent flow, or in other words the pipe roughness
affects turbulent flow that is peculiar to turbulent flow. And we saw why it is the case
right, we saw that if you look at the wall layer which is very close to the wall; the thickness
of the wall layer is so small at that scale it becomes comparable to the roughness itself.
So, in real numbers it could be less than a millimeter and the wall layer is as that small and
therefore, if you really look at the wall layer, wall layer is not on a flat surface, wall layer
is on a surface which is very uneven. And such an uneven such a complicated you know
geometry could break your viscous sub layer or the wall layer. So, that is what we were
restored, ok. So, let us just make it quantitative so, that you can do calculations.
So, roughness in turbulent pipe flow. So, what we need first is some way of characterizing
them. So, one way we need is we need some way of characterizing the roughness ok. So,
677
roughness how would you imagine the roughness of the surface to be, very uneven like for
example, if you know let us say that is a plate that you are looking at plate or surface.
If you start looking at it under a microscope you would see something you know very
uneven like that. So, you can imagine that thick line that I have drawn is the main thing
that you typically see, but if you start zooming in then you will see that it is very rough
some arbitrary of arbitrary nature.
So, you need some way of characterizing this roughness, and one thing that you could use
that is typically used is some length scale for this roughness, ok. So, let us say epsilon we
will define a length scale epsilon which is a measure of this roughness scale, roughness
length or roughness height rather. You can imagine epsilon is let us say some average
height or some average or root means square height some height, some typical height of
the roughness, ok.
So, that is a scale in which you have these variations let us call that epsilon. We can argue
that it is not one; its one quantity is not sufficient maybe you should talk about multiple
quantities to really characterize, but we of course, do not want to bring in so many things
because it makes the analysis complicated. So, we want to you know stick to what we can
use practically. So, epsilon let us say is the thickness that characterizes the roughness and
the quantity that we want to look at is epsilon plus.
So, epsilon plus would be what? Epsilon divided by what is the scale that we would use to
define epsilon plus how did we define y plus? Y. So, v divide we define y plus as y u star
by nu. So, we will define it as epsilon u star by nu where nu by u star is some length scale
that we have. Why we do that? Is because we said the thickness the highest or the thickness
of the wall layer in terms of this plus quantity is 5 approximately right, we said y plus is
approximately 5 and up to 5 you see the wall layer. So, we can say that if epsilon plus is
let us say less than 5 what is that mean? That means, that if the height of the roughness is
going to be smaller than the height of the wall layer or another words height of the wall
layer is going to be bigger than the roughness.
So, therefore, the wall layer will not see the roughness and therefore, if epsilon plus is
smaller than 5, then you can say that my pipe is actually smooth and I do not have to really
worry about it about the roughness, ok. So, we say that epsilon plus is less than 5 we call
678
it hydraulically smooth pipes, ok. In other words the thickness of the roughness is smaller
than the height of the wall layer. Is that, ok?
We are just talking everything in terms of these plus quantities rather than talking about
real dimensions, hydraulically smooth pipes sorry. If 5 less than epsilon plus less than
approximately 70, then we say it is a transition regime, the transition is in terms of
roughness, ok. Because if epsilon plus is greater than seventy then we call it a full rough
surface, ok.
𝜖𝑢∗
𝜖+ = < 5 → 𝐻𝑦𝑑𝑟𝑎𝑢𝑙𝑖𝑐𝑎𝑙𝑙𝑦 𝑠𝑚𝑜𝑜𝑡ℎ 𝑝𝑖𝑝𝑒𝑠
𝜈
5 < 𝜖 + < 70 → 𝑇𝑟𝑎𝑛𝑠𝑖𝑡𝑖𝑜𝑛 𝑟𝑒𝑔𝑖𝑚𝑒
𝜖 + > 70 → 𝐹𝑢𝑙𝑙𝑦 𝑟𝑜𝑢𝑔ℎ
So, epsilon plus is greater than 70 means the wall layer thickness is much smaller than the
roughness itself, ok. So, then it is so high, so you basically have a; some classification so,
you can say that on one limit you basically have very smooth pipes and on the other hand
you have very rough pipes and in between you know anything that could vary you basically
have a transition regime. And as usual just like in the flow, here also if things are in
between smooth and fully rough calculations are harder, but otherwise so, that is why this
classification, is that clear.
So, what is that we can do in the fully rough region; in the fully rough region so, let us say
in this region the overlap low basically gets modified I will just take it for granted I will
just write down y u plus is equal to 1 over k ln y by epsilon plus 8.5.
So, one thing that you can notice is that see when we talked about the wall layer we had
defined a y plus and we defined a y plus as y u star by nu there was no really length scale
that was characterizing it. So, we just defined it as y u star by nu, but now when you talk
about the roughness you have brought a new length scale into the system that new length
scale is epsilon.
So, it turns out that you can define u plus as in terms of a non dimensional quantity which
is y by epsilon, everything else the numbers like you know k 8.5 all determined from
experiments you cannot do much about it, but this is. So, the point is that it is still
679
logarithmic, it is a logarithmic law the numbers are going to be different and it is also
dependent upon the roughness height, is that clear. Yeah that epsilon is the height.
This is for a fully rough surface. More than 30 would be the what was that, . so y to 30 we
had the overlap layer. So, what is the question? How do you know that?
Ok yeah, that might be true. So in fact, what we are saying is that. So, epsilon your saying
is 17. Yeah. So, you basically see that the wall layer is so thin that parts of your overlap
layer is occupied by the trusts itself of the roughness in some sense or maybe all the way
up to their. So, in other words you are going to see essentially the overlap layer with. Yeah
I think that is all it is; I am not sure whether you can really characterize that structure
because that graph that we drew was for strictly smooth pipes.
So, now, we are you know essentially borrowing some of those ideas to talk about
roughness. So, in principle some of those numbers might change that must be what is
happening in reality, but if you just stick the borrowing then we are really into the
overlapping, yeah. So, this is the expression for u plus when you have a, what a rough
pipe. So, if you know u plus you know out to go ahead and calculate the friction factor and
relationship between friction factor and Reynolds number the way you did it yesterday
should we try.
See finally, we do not care what exactly u plus is we need a relation between friction factor
and Reynolds number. So, it is I never have given you what is u plus let us go ahead and
calculate what is the relationship between friction factor and Reynolds number. What
would you calculate first?
Yeah. So, you can calculate the average velocity and then hopefully you will be able to
manipulate that expression to give your desired relation. You can go back and finish your
calculation not expected to get this relation friction factor. And what you find is that you
wanted to get the expression connecting Reynolds number and friction factor, but you do
not see that ok, Reynolds number will go away and only factor that would remain would
be epsilon by d, where epsilon is the thickness of the roughness and d is the diameter.
So, that is definitely a non dimensional number and it looks like at fully you know rough
surface is what we called the friction factor is independent of the Reynolds number. It
680
basically depends only on the thickness thick or basically depends only on the roughness,
yeah.
More the roughness; so, there is a limit to which it is affecting your you know it is
disturbing your wall flow, ok. So, see these things are not independent for example, that
Reynolds number would depend upon your average velocity, its coming from your flow
pattern itself. So, you do not really need both Reynolds number and epsilon by d cube
characterize your flow it looks like epsilon by d is the only quantity that will play a role
within these approach. Epsilon will be for the pipe, yeah. Yeah.
So, that is why it’s independent of the fluid flow now. Except, wait delta p will still be
dependent upon the fluid flow f is a non dimensional quantity and therefore, its
independent of fluid properties. So, you will get an f and when you connect your f to let
us say h f and then q a delta p all the properties density viscosity everything. So, that is a
another beauty of non-dimensional number there is right that you are getting everything
independently, yeah.
Yeah. So, I am saying that the logarithmic low is still so, it is experimental observation
and the fact that what we have seen for a smooth pipe can be borrowed. So, there are some
calculations can be done where you know you can take different layers and try to match
and so on, but for us it is an experimental observation.
Alright so, you try to win it. Yeah so, more importantly it is independent of form. So, we
have written down so many relations, it is sort of hard to you know keep everything in
mind. So, what people have done is they have plotted all of them, ok.
681
Lecture - 57
Pipes of non-circular cross section
And that leads to what we call the Moody’s chart. So, let us see what we have gotten what
we are going to do plot a graph of friction factor on y-axis and Reynolds number on x-axis
and we will do within a log coordinate system for laminar flow f is equal to 64 by R e. So,
if it plot it in a log coordinate system, how would it look like? it will be a straight line of
slope minus 1.
So, that is what is going to be laminar flow, some you know transition Reynolds number
somewhere around 2000, 2100 let us say, you would start seeing turbulent flow. We
calculated an expression for turbulent flow right. It was that some 1 by root f is equal to
some expression we have calculated put log that and you will see a curve like that ok. So,
this is for laminar flow that is for turbulent flow, but in a smooth pipe.
What happens is that if you start in incorporating the roughness factor, you will going to
get a curves like that ok. So, that is basically to go up in this direction this is increase in
the roughness. If this is you know each of the squares would be correspondent to particular
value of epsilon by d. So, such a diagram becomes very useful because then you do not
682
need to remember any correlation just use that given Reynolds number looked at your
friction factor and use them for your design. They all the calculation that we have done in
the last few classes is compiled into such a diagram.
If they have tried to come up you know relations we need everything together one such
useful expression, if you do not want to use the graphs 1 by root f is equal to minus 2.0 log
of epsilon by d by 3.7 plus 2.51 divide by Reynolds number d by root f, this expression is
also sufficient to be centre calculate friction factor.
1 𝜖 2.51
= −2.0 log ( + )
√𝑓 3.7𝑑 𝑅𝑒𝑑 √𝑓
Yeah, yes when I have written log it is base 10, if I have written ln that would be natural
logarithm. It is a general people have tried to fit that equation and reasonably reproduces
the flow. So, you would use either of them actually you know you could use any of these
things ok, but whenever you are going to do a design problem, you do not want to stick to
exact numbers.
You would want to you know say that you know whatever number you get, you want to
give some cushion to it, you will say you know I will be take 15 percentage or 20
percentage extra whatever is required. So, there things get adjusted. So, you could either
you will equation or the graph, the graph is definitely going to be better so that is more
discharge. Next, so, let us look at the flow you know that we had so far, we started with
laminar flow then we did turbulent flow in a pipe then we talked about roughness ok. So,
we are ready to do the next complication, what would be that in your opinion?
What is you know just imaginary life scenario where you want to design a pump oh yeah
or a piping system rather. The next thing that, yeah you said something right now, know?
Power yeah. So, power we can calculate because once you know pressure drop then you
can you may just multiply it with the flow rate and you get it.
So, you know we could encounter more scenarios for example, so far we assume the pipe
to be circular, need not be circular. You may have a square pipe or a triangular pipe
especially you know for example, heat exchanges and stuff like that you might have
different cross sections. What do we do then? Do we do the all the calculations that we
have done so far for a different cross section, it becomes extremely hard.
683
In fact, it you cannot even do that for many of the things ok. So, therefore, you define
something called a hydraulic diameter, have you heard hydraulic diameter? So that is
where it comes.
So, we want to look at pipes of non circular. So, if you define it, we need to go back and
look at how did we originally define our friction factor um. We analyzed laminar flow in
a pipe, we at some pipe which we must probably you would have drawn something like
that a circular pipe of radius R and let us say some length L with the pressure drop let us
say delta P then angle theta and some delta Z and if tau w is the wall shear stress. You
would have written down some forces balance then you note a check your notes and tell
me, what is the force value that we wrote for this configuration?
So, there is a pipe and then you have applied a pressure difference which makes it flow, it
is inclined and therefore, weight is playing a role and of course, the viscous friction is
acting against it. So, you need to apply you can write down on a pressure balance which
would mean that if delta P into pi R square is the force acting along the pipe plus the weight
which would be pi R square times L times row times g let us say sin theta should be balance
by the viscous friction that is acting on the walls which would be tau w times 2 pi R into
L, this is the equation that you might see in your notes.
Now, what we want to do is we want to look at the write down similar expression for a
non circular pipe let us say some non circular pipe with some side a everything else
684
remaining the same you can write. So, let us call this as an area of a and so, A is the area
and P is the perimeter ok, you can write down the same force balances delta P into A plus
this we must have simplified no L times sign theta would have given you a delta is z.
So, A times rho g times delta z is equal to tau w into p into L just generalizing it in terms
of A and P, one thing that you need to be careful is that when we talked about the circular
pipe ok, the wall shear stress was a constant quantity, it does not matter where you are
actually looking at or were you are calculating tau w were because nothing was changing
along the theta direction or z direction, but that is no more true when you talk about non
circular cross section depending up on where you are looking at tau w could change.
So, we can say tau w is some average tau w that we have calculated, yeah. So, that would
mean that the P plus that we let us say divide by rho g plus delta z is equal to tau w bar
and L divided by rho g A by P.
Δ𝑃𝜋𝑅 2 + 𝜋𝑅 2 𝐿𝜌𝑔 sin 𝜃 = 𝜏𝑤 (2𝜋𝑅)𝐿
Δ𝑃 ̅𝜏̅̅𝑤̅𝐿
+ Δ𝑍 =
𝜌𝑔 𝜌𝑔(𝐴/𝑃)
Δ𝑃 2𝜏̅̅̅̅𝐿
𝑤
+ Δ𝑍 =
𝜌𝑔 𝜌𝑔𝑅
685
So, what you see on the left hand side is what you originally had, now you end up with an
expression which contains tau w area and perimeter and so on and this quantity is what is
called hydraulic radius. May in other words R H is nothing, but cross sectional area divided
by wetted perimeter and that is actually for a circular pipe that will be equal to R by z right.
So, then define a hydraulic diameter D H is equal to 4 times cross sectional area divided
by wetted perimeter. So, that mean D H is equal to D for a circular pipe um. So, in case of
a circular pipe, we wrote h f the head loss is equal to friction factor times L by D v square
by 2 g. In case of a noncircular pipe, we will write similar relation h f is equal to f times L
by D hydraulic v average square will be different v square divide by 2 g, everything we
did for the circular pipe we just borrow it, but replace all the diameters with the hydraulic
diameter.
4 𝑋 𝐶𝑟𝑜𝑠𝑠 − 𝑠𝑒𝑐𝑡𝑖𝑜𝑛𝑎𝑙 𝑎𝑟𝑒𝑎

𝐻𝑦𝑑𝑟𝑎𝑢𝑙𝑖𝑐 𝑑𝑖𝑎𝑚𝑒𝑡𝑒𝑟, 𝐷𝐻 =
𝑊𝑒𝑡𝑡𝑒𝑑 𝑝𝑒𝑟𝑖𝑒𝑚𝑒𝑡𝑒𝑟
𝐿 𝑉2
ℎ𝑓 = 𝑓
𝐷𝐻 2𝑔
686
So, for a laminar flow, we had f is equal to 64 by R e. For the circular pipe here, we will
define f is equal to 64 divided by R e H or rather D H; D H where the Reynolds number is
now defined based on the hydraulic diameter. So, R e D H is D H times v times rho divided
by mu.
For turbulent flow, we will get some relation containing Reynolds number, friction factor
epsilon by d equal to 0. Here we will have you we will use the same relation, but we will
use Reynolds number based on D H, f, epsilon divided by D H is equal to 0.
64
𝑓= [𝐿𝑎𝑚𝑖𝑛𝑎𝑟 𝑓𝑙𝑜𝑤]
𝑅𝑒𝐷𝐻
𝜖
𝑓 (𝑅𝑒𝐷𝐻 , 𝑓, ) = 0 [𝑇𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡 𝑓𝑙𝑜𝑤]
𝐷𝐻
So, what I am saying is that everything they we have done so far, we will just borrow the
same frame work just replace our all diameter with hydraulic diameter, assume that that is
going to work. If you are wondered how good it is, this relation is going to give you things
with an error of 40 percentage cubes this relations seems to be a approximately 15
percentage . So, this depends on what kind of geometry and so on, this are definitely not
good relations, but what we definitely we do not want to do you no elaborate calculation
for any other geometry just use the same framework ok.
687
Let say you have a fluid that flowing between two pipes, have you already been you did
heat transfer shall until you heat transfer heat exchanger and so on. So, imagine you have
a fluid that is flowing between two pipes and you are interested in calculating the pressure
drop for the fluid which is between in the gap in the annulus, what would be the hydraulic
diameter.
It is going to be 4 times cross sectional area by wetted perimeter. So, now, you see right
where is cross sectional area coming from and where is wetted perimeter coming from.
Cross sectional area is coming because that is the one which is going to multiply with your
pressure drop; wetted perimeter is coming because that is where your shear stress is acting
ok. So, this really ratio what would this be, 4 into cross sectional area is pie into R o square
minus R i square divide by wetted perimeter would be 2 pie R o plus two pie R i right both
inside and outside both the cylinder both the surface is then generate you no it is not a
shear stress it simplifies and see what you will get. So, that is how you should do the
calculations.
4𝜋(𝑅𝑜2 − 𝑅𝑖2 )
𝐷𝐻 =
2𝜋𝑅𝑜 + 2𝜋𝑅𝑖
688
Lecture - 58
Minor Losses, Sudden Expansion and Contraction
So, the last class we looked at the effect of roughness on pipe flow and we also looked at
how to deal with non circular cross sections right. So, just to write down we started by
looking at laminar flow in pipes, then we realized that that is not always the case, if
Reynolds number is sufficiently large then the flow does not remain laminar.
The flow goes to turbulent and therefore, it was necessary to look at the relation between
friction factor and Reynolds number for turbulent flow in pipes, then we realized that if
you are looking at turbulent flow then the roughness of the pipe can be important. So, we
looked at roughness, we defined a roughness factor. We calculated what is the relationship
between friction factor and roughness.
And then we sort of generalized our procedure by looking at noncircular pipes and to apply
all the formalism that we derived for circular pipes. We actually defined a hydraulic
diameter and then we said that we would just replace use the hydraulic diameter to deal
with the non circular pipes. So, that has really been the course so far; so we defined a
hydraulic diameter.
689
So, is that all that we need if you want to talk about a pipe design in a practical case. So,
do we know everything now how to select a pump for a given application? Is there
something that you think is missing? Something is missing and that is what is the next
thing that we want to look at and that is called the minor losses.
So, the minor losses is pressure drop that can happen that we have not accounted for h and
the possible reasons could be what happens at the entrance or exit length. It could be due
to sudden expansion or contractions. You could have bends, elbows, fittings and so, you
could have valves you know open or partially open you could have a gradual expansions
or contractions.
So, when you think about a general pipe system where you are pumping fluid from one
place to another many of these things could come and all of them can contribute to
additional pressure drop. So, we also need to know what those losses are and the losses
that happen due to any of these things are called minor losses, though its called minor
losses it may not be really minor ok, many times that would be the major losses. So, its
important that we account for this. So, that we need to just look at some of these situations
and then see what is the way in which we can really attack it. So, the first thing that we
would look at entrance and exit length. Entrance length is the only thing that I will talk
about.
690
So what is entrance length? So, when we started our class you may remember that we
talked about pipe and a fluid that is coming in with a uniform velocity and as soon as the
fluid touches the wall, the fluid will get slowed down and then it will take a it will take
some length before it will reach the fully developed flow right and the length that is
required to reach the fully developed flow is called the entrance length right. So, and all
our analysis so far has been assuming that the fluid flow is fully developed.
So, the flow or the pressure loss pressure loss that would happen during the entrance loss
is something that we have not calculated yet and therefore, that should be accounted for
additionally whenever you have. So, the point is can that be significant ok. So, let us see
what are the typical entrance lengths that you will have ok.
There is no real theory, but there are correlations which one could use. So, if so let us say
this is entrance length let us say L E beyond which you will start seeing fully developed
profile. So, that is L E by R where R is the radius of the pipe is given by 0.49 plus 0.11
Reynolds number for laminar flow. So, if the fluid flow is laminar, then you can estimate
what is the approximate entrance length.
𝐿𝐸
= 0.49 + 0.11 ∗ 𝑅𝑒
𝑅
So, let us say the highest Reynolds number at which you will get laminar flow in a pipe
would be approximately 2000 2100. If Reynolds number is 2100 let us say what would be
691
L E some 2000 into 0.11 plus 0.49 something like 200 plus comes out to be 230 R ok. So
that means, is there anything? Ok. So, 230 R ok; that means, the entrance length is going
to be 200 times longer you know compared to the radius. So, its not a small number; so, if
you are dealing with 1 meter sized pipe the first 230 meter would be entrance length. Only
after that you are going to get a fully developed profile ok.
So, therefore, the calculations are the calculations that we have done so far is not true for
the first 230 R length and therefore, it has to be separately accounted for. Even in turbulent
flow its L E is found to be approximately 160 R to 200 R ok. So, that is also sufficiently
large. So, whether its laminar or turbulent there is enough length of the pipe in which the
flow is just in its developing state. So, the way typically that you would see is that we need
to talk about a head loss which is related to your pressure drop right that is the way we
defined our head loss.
We can say our delta P is now coming from the fully developed flow plus a delta P that is
additionally there due to the entrance effect. So, how did we write delta P in terms of tau
w the wall shear stress, what is the definition of what is the connection between a head
loss and wall shear stress? Is it all ok? So, therefore, its a good idea to define delta P
divided by 2 tau w is delta P FD divided by 2 tau w plus delta P En divided by 2 tau w that
is equal to L by R plus let us say some L star by R.
Δ𝑃 2𝜏𝑤 𝐿
ℎ𝑓 = =
𝜌𝑔 𝑅
Δ𝑃 = Δ𝑃𝐹𝐷 + Δ𝑃𝐸𝑛
Δ𝑃 Δ𝑃𝐹𝐷 Δ𝑃𝐸𝑛 𝐿 𝐿∗
= + = +
2𝜏𝑤 2𝜏𝑤 2𝜏𝑤 𝑅 𝑅
So, what I am just trying to rewrite is there the pressure difference that the pressure drop
ok. Can be really thought about as an extra length and that extra length is called L star and
what is what you will find in literature is expressions for L star which you can use and
calculate what is the extra pressure drop that can come from entrance length. Is that ok?
So, not L star is not same as L E ok. LE is actually the original entrance length, L star is
that that length which you would have contributed to the pressure drop that an extra
additional length L star by R is equal to 0.0709 Reynolds number plus 0.589 for laminar
692
flow and yeah is equal to 0.72 divided by f for turbulent flow. So, you can use those to
calculate what is the additional pressure loss. So, that is what typically happens with
entrance and we can talk about sudden expansion and contraction ok.
𝐿∗
= 0.0709 ∗ 𝑅𝑒 + 0.589 (𝐹𝑜𝑟 𝑙𝑎𝑚𝑖𝑛𝑎𝑟)
𝑅
𝐿∗ 0.72
= (𝐹𝑜𝑟 𝑡𝑢𝑟𝑏𝑢𝑙𝑒𝑛𝑡)
𝑅 𝑓
So, it is very common for example, you work on pumping fluid through some pipe and let
us say it needs to be divided or it needs to be connected to a larger pipe ok. So, it is very
usual that the pipe dimensions will immediately change. So, you will see that up to some
length you had something and then suddenly you will connect it to a bigger pipe to which
through which the fluid needs to go ok. So, that is where you have a sudden expansion.
So, what would happen is at the fluid that is coming like that? It does not stick to the wall.
In fact, the one which is coming at the center will go at the center the one which is next to
it will expand a little bit and go. The one which is next to it will expand more and go like
that ok. So, that is what the fluid profile is going to look like ok.
So, the fluid layer that is very close to the wall would expand and then go. So, here what
happens is that there will be fluid, but those fluids would be just recirculating like that.
693
There will be fluids which are just getting recirculated the fluid that is coming would
expand and occupy the pipe only after a certain length. So, that means, that you the fluid
that is occupying this region right, the region at the corners are just getting rotated, but
they are not moving anywhere. So, you are actually losing additional energy because of
that right. You are not getting anything from that you are having extra energy to just
generate those vortices there and those losses are actually huge ok.
So, therefore, sudden expansions will result in additional losses that we have not captured
yet from our friction factor calculations. So, we need a way of calculating friction factor
associated with this sudden expansion. So, similarly this similar thing could happen in case
of a sudden contraction. So, in if you have a pipe with a large cross section and is suddenly
changing to a small cross section the fluid that is coming would start actually so the one at
the center would go, the one would actually contract a little bit before its expanding
something like that.
So the fluid that is coming as soon as it see a contraction it will continue like that it will
contract a little bit before it expands ok. So, again you will have regions where you will
have this kind of circulations, I am not going to write ok. So, this is these are called flow
separated regions and we will see the and the regions why that happens later, but those
regions are again created ok. So, these losses are again something that we have not
considered. So, therefore, there can be expansion losses, there can be contraction losses
whenever the cross section changes, which needs to be accounted for in addition to the
non losses.
So, really you know whenever the so see originally for example, if you look at laminar
flow, where is the friction really coming from? The friction was coming from the walls
right that is what we accounted for. Now, here the additional friction is coming because
either the magnitude of velocity changes or the direction of velocity changes. Actually
both change here right the magnitude and the direction both of them change. So, if you
look at 2 consecutive fluid elements their velocities might be different both in direction
and in magnitude; that means, that they are going to experience extra friction ok. So, it is
really that extra friction that is resulting in additional losses, which was not the case when
we had a uniform pipe the there.
694
The only the only way in which the velocity is changing is when you are going
perpendicular to the flat plate or perpendicular to the pipe cross section. So, that is the only
way friction was exhibiting. If you are going along the pipe there was no difference in the
velocity of fluid elements and there was no friction, but now that is not the case. Here any
2 consecutive elements you are looking at there is going to be difference in the velocity
and if 2 things are moving with a relative velocity you would expect that there is going to
be friction acting between them ok. So, therefore, whenever there is a change in the
magnitude or direction of velocity you would expect that there is going to be additional
losses in addition to the circulations that we have already talked about.
So, there are like two different contributions really, one is because of the change in the
magnitude and direction of the velocity and second is because the flow is separated like
you know near the corners or near this corner, which I have drawn here. So, these are
various contributions that are going to give additional pressure drops and are going to be
something major also. So, again the one cannot really do any calculations and figure out
these things, so it is very difficult to do it. So, typically one goes with again correlations
and often you would write that you will talk about head loss let us say minor head loss h
m is yeah this is the pressure drop ok.
So, that was, so our original definition of h f the friction factor were we defined it as delta
P by rho g. I am not including delta is at because delta is at contribution is going to be
small for any of these changes. So, h m is delta P by rho g and we often define a loss
coefficient K m as h m divided by v square by 2 g.
So, v square by 2 g is the velocity head. So, we often represent our minor head loss as a
ratio of minor head loss to the velocity head and therefore, that at K m is going to be a non
dimensional quantity and the numbers that you will have often access to is K m. So, the
moment you have K you can multiply it with the v square by 2 g the velocity head and get
your friction loss yeah.
Δ𝑃
ℎ𝑚 =
𝜌𝑔
ℎ𝑚
𝐾𝑚 =
𝑉2
(2𝑔)
695
So, this is loss coefficient that is velocity head. K m again often independent of Reynolds
number roughness and so on. So, you will see that the dependencies are v; so you will
actually get it as some number like 0.1 or 10 or things like that ok. It is very it is a function
of the geometry. So, for example, you will find that K m could be dependent upon the you
know the radius of the smaller section and the radius of the bigger section and so on, but
you will not generally see it as written as functions of the flow field or Reynolds number.
And the other thing is that what is the v that I have written there velocity head, what is that
v?
Student: Average velocity.
Average velocity, but average velocity where? So, there is a convention involved and that
convention is that the average velocity in the smaller pipe ok. So, that is what typically
one should use. So, in the expansion it will be the inlet average velocity in the contraction
it will be the outlet average velocity. So, that v associated is that in the narrower channel.
So, we will look at expansion a little more detail maybe next class, but before that let us
look at other things bends, elbows, fittings bend why would you know.
Let us say a bend in the pipe or an elbow that you will collect. For example, you need to
take the pipe like this and then take a corner and go. Why would that cause additional
losses? Something here the fluid is going like that. It could be a sharp 90 degree bend or
696
anything. So, what would the fluid experience when it goes through something like this?
The additional force that it would experience would be centrifugal force ok.
So, in addition to the fluid that is going to go like that because the fluid experiences
centrifugal force it will start exhibiting additional flows ok. We have actually seen one
example when we did Taylor Couette flow where we had flow between 2 parallel plates
and one sorry 2 concentric cylinders and one of the cylinder was rotating and we saw that
when the Reynolds number was really small it was only like this, but as it was increasing
additional flow fields developed ok.
So, similarly what happens is that even though the flow field is the primary flow is along
the bend it will develop a secondary flow and that secondary flow will be actually you
know across the section. So, something like that you know some sort of a helical pattern.
So, the fluid will start exhibiting flow like that. So, that is something that we have not
accounted for yet ok; so that gives rise to additional losses friction losses yeah. Also
depending up on how strong this bend is you could have again flow circulation. So, maybe
I should draw another one. So, the primary flow is like that then it could develop a
secondary flow ok. So, that is secondary flow and then also you could have flow
separations like this ok.
Near the bend there could be certain regions which are not moving they would be separated
from the main flow. So, these all will additionally contribute to frictional losses and
therefore, you need to take care of. So, you when you design something you need to k.
Now, therefore, if you are actually giving a bend you should think about what is the loss.
So, again you will define a loss coefficient ok. You have to find out what is the loss
coefficient for a 90 degree bend or for a smooth bend and then add that to the total frictional
losses.
So, you might want to for example, think about I will just avoid having all the you know
bends at all. I will try to give as smooth as possible so, that you can reduce the frictional
losses, but many times that may not be a good idea because you need additional space you
are thinking about giving longer curvatures; that means, longer pipes. So, that long length
would also cause you know additional pressure drop.
So, you will have to estimate if the space is available do you need actually a long pipe ok.
With the minimum you know bend or do you need actually a short pipe with a 90 degree
697
bend what is best required; so you need to do a comparison. So, again the cost will the
operating cost will change the initial you know investment will change. So, you need to
know which one is suitable. So, that is why it becomes important to estimate these things
and small differences here and there over a month would make a lot of difference. So, that
is why people real worry about what are the numbers associated with, any questions.
That is about bends valves; valves you must already be you have submitted assignment no
i have not looked at it yet. So, you were; so dealt with different kinds of valves right what
all valves did you write about? Gate valve ok then?
Check valve, then maybe we should have had a surprise quiz on valves today yeah
therefore, for final exam there will be questions from assignments. So, yeah so, there
several like you know disc valve, butterfly valve, globe valve I think all of you must have
written different things. So, and when you have written it you might have probably found
that certain kinds of valves are useful for gases, certain kind of valves are useful for liquids,
some of them have a better control, some of them are like you know completely open and
completely closed or you want to fine control over your changing. Why do we need flow
valves by the way?
To control the flow rate we always. So, for example, you know you have you know for a
plant and there are fluids that are coming from outside ok. Let us say for a heat exchanger
698
you are taking water from the nearby river or let us say you have a petroleum which is
getting you know which is coming from some port somewhere which is getting injected.
So, there are lots of external influences and all of them are going to affect the fluid
properties that are coming in the temperature might be different, the composition might be
different. So, the equipment that is under operation must be you must be able to change
your flow rate. So, that the equipment works the best way ok. So, you would always want
to therefore, control your flow rate that is why you would use valves and depending upon
what fluid it is you may use different valves.
The valves may be completely open or it may be partially open and its important whether
it is completely open or partially open or partially because often a partially opened valve
will generate more pressure drop than a completely opened valve. So, therefore, one needs
to know how much it is open and then the pressure drop calculation should really depend
upon that. Yeah and the valves the pressure the again the frictional losses are coming
because of its complicated geometry ok.
So, you must have seen all sort of complicated geometries associated with for example, a
globe valve and so on or a butterfly valve ok. There will be the geometry is complicated
the velocity is changing quite dramatically in the valve the both magnitude and direction
there could again be regions of separation. So, all of them will contribute additional losses
and all of them will be calculated based on the loss coefficient that you get.
The loss coefficient just the way we defined for our sudden expansion or contraction; we
defined a loss coefficient. Similarly when you buy a valve from the manufacturer you will
get what is the loss coefficient or in other words you should ask them about it they would
have done an experiment and would have calculated.
So, you get those numbers and use it in your design that is the way you should go ahead
those numbers may not be so accurate and therefore, you should give always some margin,
say typically the error could be as large as 50 percentage in those numbers because it
depends all lot on you know the flow conditions. So, it might be very different the way the
manufacturer would have calculated the loss coefficient, the conditions may be very
different from your operating conditions. So, you should worry about those details, but
typically there are some guidelines would be available which you can take for granted and
do the first calculation; so valves and gradual expansions and contractions.
699
So, we talked about sudden expansion and contraction a pipe a small pipe suddenly getting
enlarged. So, you would think that you know and you realize that yeah of course, the major
losses are coming because the you know the direction the magnitude is suddenly changing
also there are regions of the separation and so on. So, why should we have a sudden
expansion or a sudden contraction?
We would rather go for something that is smoothly changing. So, that we can avoid the
you know sudden changes in the velocity and therefore, reduce the frictional losses. The
idea is ok except that the feasibility is very little ok. So, let us see if we talk about a sudden
expansion, sorry its not sudden expansion a gradual expansion; fluid is coming this side.
So, the because the its an expansion ok, the velocity downstream would be smaller or
bigger you have an increased area. So, the velocity is going to be smaller or in other words
the velocity head goes down and we have seen the energy equation now the energy
equation was what? There is a pressure head plus there is a contribution that will come
from the elevation, then there is a velocity head and these 3 basically balance or in or you
could have a friction right. These are the 4 contributions that would have come up in your
energy equation.
Now, we are talking about something, let us say we do not care about the elevation change
ok. So, you could have a pressure head and a velocity head. So, if the velocity comes down
that can get converted into a pressure that is what Bernoulli’s equation really says right
velocity decreases the pressure increases. So, what happens is that if you have an
expansion many much of the kinetic energy can actually get converted into a pressure form
ok. It becomes pressure head, but the difference here is that if you look at the fluid that is
coming in the velocity profile is going to be that right. The velocity at the center is large
the velocity near the wall is small. So, let us draw 2 streamlines maybe one is that and one
is very close ok.
So, let us call this a so difficult this is a and this is b and somewhere else this is this a
prime, this is b prime ok. So, a prime is a streamline, b plus b prime is a streamline. At
location a the velocity is going to be definitely larger than the velocity at location b agreed
because of the presence of the wall. At location a not a prime at location a which is also
actually part of that uniform pipe a is slightly away from the wall and therefore, a will have
a larger velocity then if I if I look at a fluid particle at location a, it will have a larger
700
velocity than velocity of a fluid particle that is located at b or in other words the velocity
head of particle a at location a is going to be larger than at b.
Now both a and b are the fluid particle velocities are going to slow down as it goes
downstream ok. By the time it reaches a prime, b prime part of the velocity head would
have been converted into pressure head agreed. Now, it so happens that the pressure itself
is not going to change much across the cross section. If you look at the pressure at location
a prime and pressure at location b prime the pressure remains almost constant.
And the only way that would have happened is that, because the fluid particle at location
b had a smaller velocity head it would have lost much more than the one which is at a
prime along the streamline, or you could say that for example, the fluid particle at location
b would have come to rest by the time it reached b prime, but the one at location a would
still be moving ok, because the velocity head is actually getting converted into pressure
head.
So, because a was coming with a larger velocity the particle at location a was coming with
a larger velocity it gets converted into pressure head, the one with at b also got converted
into pressure head, but the velocity at location a would be larger than the velocity at
location b even though both of them have been the velocity heads have been converted
into pressure head ok.
Actually taking that to the limit in which you would see that if b prime actually comes to
you know a stagnant situation if b prime actually comes to a halt a prime would still be
moving. So, what you would see is that as you proceed along this direction you will see
that b prime gets slower and slower while a prime will start moving faster and I mean
compared to b prime a prime will be moving faster ok.
So, the relative velocity between a prime and b prime keeps increasing. After some point
what happens is that the b prime will come to 0 velocity while a prime would still be
moving and beyond some point you would see that b prime would actually move in the
opposite direction compared to a prime. Now, whatever I said just now we will see how
the equations predict that may be when we do the boundary layer for the time being let us
just get this fact. That therefore, there can be a region where the fluid will actually start
flowing in the opposite direction even though the main flow is in a different direction. So,
701
along the center the fluid would be actually flowing from left to right, but near the wall the
flow could be in the opposite direction ok.
So, that is coming because of this differences in which how much velocity head is
converted getting converted into pressure head. So, the numbers let us say we will see
later. So, what happens is that because of that the fluid. So, even though outside this layer
the fluid is in one direction, inside the fluid could actually start going in the opposite
direction. So, that is what you really call as flow separation. So, whenever I do flow
separation there; there; there; there, you know here in all these cases this is what is
happening even though you have a major flow a primary flow there could be a flow
separation which would which would be in an opposite direction and can result in much
more pressure drop ok.
So, even though you would want to change your sudden expansion or a sudden contraction
there could still be regions where this flow separation can occur and it turns out that if you
want to really stop flow separation the angle of divergence that you need is less than 7
degree. So, if it is less than 7 degree you will be able to reduce the that separation, but
otherwise it is difficult to stop the separation; that means, if you want to get rid of a sudden
expansion you should think about an expansion which is having a 7 degree; that means,
that expansion would be over a very large length because the angle is really small where
you can stop separation ok.
So, you if you want to replace a sudden expansion with a gradual expansion your pipe
should be really from you know the expansion should be happening over a really long
length. Now, that becomes a problem because you want to get a pipe which is slowly
changing in its cross section over a large length that is going to be really a custom made
thing that you are asking for and you need to also plan how to you know space it out and
so on. So, other course will actually become important and material cost will also go up
ok. So, therefore, it is also not always a good idea to replace a sudden expansion with a
gradual expansion unless its you know really going to help us.
So, therefore, you would want to again compare whether you want sudden expansion or a
gradual expansion when you are designing something ok. So, it is all due to the separate
the fluid that get separated. So, these regions where the reverse flow happens is called flow
separation ok. Its separated from the main flow and that flow separation is a reason for
702
additional losses. So, when you add when you install something the thing that you would
look for is to not have any flow separation. Do not worry if you did not get the idea of why
it happens, we will learn about it in detail, we go to the boundary layer theory.
So, let us look at a situation fluid is coming in a larger pipe, then it got contracted and let
us say you have some valve and then the fluid is going and then its suddenly expands. Let
us say this is the particular section that you are interested in and you want to know what is
the pressure drop across this section, let us say this is very small so; some valve is there,
so something is coming here fluid and going out.
So, if you did not have that valve and if you did not have that expansion and contraction
you would have just written down your usual loss which we called as the friction loss h f
as f times L by D times v square by 2 g right. Where we relate now f with the Reynolds
number, so you would calculate it. So, remember this diameter is that diameter, this length
is the length of the small pipe, v is the average velocity in the small pipe.
Now, you have got an expansion, you have got a contraction and you have a valve. So,
you need to add 3 minor losses to this expression. So, your total loss h t let us say we can
write it as h f plus h m the friction loss that is coming from the fully developed part of the
flow plus additional minor losses and h f you will write it as f into L by D into v square by
2 g and minor losses could be many of them. So, you have a minor loss that is going to
come from the contraction which you will write sorry yeah contraction which you can
703
write as K c into v square by 2 g. Remember how we defined our loss coefficient, we
defined our loss coefficient as the minor loss divided by the velocity head.
So, the loss coefficient times the velocity head is going to give you the minor loss. So, K
c which is the contraction. So, I have the c subscript is for contraction K c into v square
plus 2 g plus there is a valve. So, let us say K v into v square by 2 g plus there is an
expansion K e into v square by 2 g. So, that is the total pressure drop or the total head loss
which is equal to delta P by rho g; so, you know the pressure drop. Can you if you like you
can write it as v square by 2 g into f L by D plus sum of all minor losses ok.
Δ𝑃 𝑓𝐿𝑉 2 𝑉2 𝑉2 𝑉 2 𝑉 2 𝑓𝐿
= ℎ𝑡 = ℎ𝑓 + ℎ𝑚 = + 𝐾𝑐 + 𝐾𝑣 + 𝐾𝑒 = ( + ∑ 𝐾𝑚 )
𝜌𝑔 2𝐷𝑔 2𝑔 2𝑔 2𝑔 2𝑔 𝐷
So, all we have said today is essentially how to attack this problem ok. If you are you have
got a pipe design, then the way to proceed would be look at the loss coefficients associated
with each element that you would put it in the pipe and add it your frictional losses that is
a take home message hm .
704
Lecture - 59
Friction Losses in Sudden Expansion
So, we learned about minor losses in the last class and we said minor losses would come
into the picture whenever you have additional fittings in a pipe. So, we said it could be
sudden expansions, contractions or gradual expansions, contractions, valves; you know
elbows, turns, bends any of them could give rise to additional frictional losses and that
could be huge and we need a way of a characterizing it. And we said we so, we are talking
about loss coefficient ok.
Now, typically look calculation of loss coefficient is a pretty difficult and you need to
depend on literature or published data to figure out what is the loss coefficient that is useful
for you, except for you know sudden expansion; they exist a nice calculation. So, I think
it is a good idea that we just look at it to get a feel of what it means, it also might give you
a little more insight into the definition of loss coefficient.
So, what we are going to do today is to first look at frictional losses in a sudden expansion.
So, what we have in mind is fluid entering through a small pipe and then exiting through
a large pipe. So, that is a fluid stream line at the center; others as we said would expand as
705
it go along ok. So, that is what the fluid stream lines are going to look like. Let us say this
fluid that is entering, let us say the fluid that is entering as a velocity v a and it is exiting
with a velocity let us say v b where v and v b are the average velocity.
And, let us say this cross section the cross sectional area is Sa and the leaving cross
sectional area is Sb. And, let us say the pressure with which is entering is P a and the
pressure with which it is not leaving is P b. And, we want to calculate what might be the
losses associated with it, what I am going to do is I am going to. So, dealing with
differential equations in this case, it is not going to be useful because, you cannot do any
simplification. So, here the flow is very complicated so, we can do an integral approach.
So, that is what we will do, let us say we will take a small section very close to the wall
and developed part. So, let us call it let us say section AA prime and B B prime. So, how
do I am basically looking at a control volume which is you know between A A prime and
B B prime and I am going to write down a force balance and an energy balance for that
region ok. So, first is the force balance so, you have a fluid entering at some pressure, fluid
is leaving at some other pressure.
So, there is definitely going to be a some forces that might come from pressure, you could
have shear stress the wall shear stress that is acting. So, that is something that you should
consider and then the fluid is changing its momentum because its velocity is changed right.
So, these are the three relevant things. What we are going to do is we are going to neglect
the contribution that is coming from wall shear stress ok that is an approximation. We
assume that wall or the viscous contribution really contribution from the wall is irrelevant
in the force balance.
And, I can find out what is the net force acting on this fluid that is only going to come
from the pressure forces. The pressure that is acting on the left hand side let us say is so,
we know the pressure in the pressure of the incoming stream is P a. We do not know the
pressure here which is very near the wall let me just call it P prime ok, then the force that
is acting along x direction is P a times S a plus P prime times S b minus S a ok. So, there
is a force acting along x direction minus the force at the section B B prime P b into S b
that is ok. And, just multiplying pressure with area to get my force is equal to change in
the momentum which is going to be let us say some mass flow rate m dot into v b minus
v a.
706
𝑃𝑎 𝑆𝑎 + 𝑃′ (𝑆𝑏 − 𝑆𝑎 ) − 𝑃𝑏 𝑆𝑏 = 𝑚̇ (𝑣𝑏 − 𝑣𝑎 )
So, I have said the rate of change of momentum which is I have written on the right hand
side is equal to the total forces acting in that control volume where, I have neglected the
force that will come from the wall region ok. And, I am doing that assuming that that
contribution is going to be small, other losses are going to be large and also I know that
that tau w the contribution that will come from the wall is really over a small section ok.
So, that is not going to be very large compared to other losses so, that is the assumption
that is going in that is.
So, m dot is the mass flow rate. Energy I will simply use a you know energy flow equation
which again you have seen I will apply it between section A A prime and B B prime. So,
I will say P a by rho g plus half v a square plus z a is equal to P b by rho g there is a g
know g we said yeah, P b by rho g plus half v b square plus g z b plus the other losses
which is what we want to calculate. Let us just applying between section A prime and B
B prime. They missed something? Yes yes yes.
𝑃𝑎 𝑣𝑎2 𝑃𝑏 𝑣𝑏2
+ + 𝑧𝑎 = + + 𝑧𝑏 + ℎ𝑓
So, you have the fluid that is coming, it is expanding and going so, that there is a force that
is going to come from that section that is neglected yeah should be P by rho. So, P by rho
plus half v square is it ok, this is correct now.
Wait it cannot be correct now because this is v square so, that is v square plus second
square. So, it should be g is z that sounds correct.
Oh ok. So, wait let me just check what should be then right. So, it should be P a by rho g
plus v square by 2 g plus is z is equal to P b by rho g plus g this, is this now? Yeah.
So, we are looking at the total force right exerted in the flow direction so, it is just force
sorry pressure times area. So, you have a this is coming from that part and this is coming
from this part, pressure on the wall so ok. So, one way to argue is that I am going to look
at a section which is very near to the wall, but why would pressure on the wall be 0?
No no no. So, that is you are blindly applying your Bernoulli’s equation, but no why should
still it be 0? It need not be 0 so, let me give you a counter example, let us say you take a
707
fluid it is a static fluid there right. What would be the pressure on any side of the wall? Ok,
so its there is nothing like pressure should be 0 on the wall, top meaning this region again
oh. So, in the static case you mean here so, even if the pressure here is 0 ok, you have a
hydrostatic pressure that will arise right which will act on all sides of the wall right and
that pressure will continue to increase. So, the pressure need not be 0, it is completely
independent of whether the fluid is moving not moving anything ok. So, the pressure could
be some quantity ok.
So, the forces that are acting on a solid surface need not be just viscous, it could be pressure
force also. I have not made any assumption yet, but the derivation that we are going to see
you would see that it would hold better for a turbulent flow than laminar flow which I will
argue later which one. So, I have a h f no now? So, so you can say that it is just a so, see
P a v everything is an average quantity here ok. So, it is just writing down what is the net
energy in is equal to net energy out plus extra force extra losses that is yeah.
Yes yes.
No h f is not so, there can be additional losses as we said in last time. So, the velocity
magnitude is changing, the velocity direction is changing both of them will contribute to
additional losses plus there will be re-circulations that will happen at the corner which will
also contribute. So, what h f will give you is you know the sum of all those things yeah ok.
So, the only thing is to just so, we will throw away the z a z b because we are not going to
worry about the change in elevation. So, let us simplify so, there comes the second
assumption ok.
708
Now, we assume the about something about the wall shear stress, the second assumption
is that we do not know what is P prime ok. And, I am going to claim that P prime is going
to be approximately equal to P a that is the inlet pressure. Now, if you ask me why do we
do that we really do not know, but if you do experiments you find that it is very close by.
And, the results that you are going to get from this calculation also seem to be validating
this assumption ok. So, that is the second assumption that is going in.
So therefore, I can simplify our pressure balance so, that is P a S a plus P prime into S b
minus S a minus P b S b is equal to m dot into v b minus v a. So, if I substitute P prime is
equal to P a P a S a plus P a S b minus P a S a minus P so, this is P a S b minus P b S b
correct is equal to m dot; m dot is the mass flow rate, m dot is either rho v a S a or is also
equal to rho v b S b mass flow rate ok, volumetric flow rate into density. So, I will
substitute rho v b S b into v b minus v a. P a minus P b into S b is equal to rho v b S b into
v b minus v a or P a minus P b is equal to rho v b into v b minus v a.
𝑃′ ~𝑃𝑎
𝑃𝑎 𝑆𝑎 + 𝑃′ (𝑆𝑏 − 𝑆𝑎 ) − 𝑃𝑏 𝑆𝑏 = 𝑚̇(𝑣𝑏 − 𝑣𝑎 )
𝑚̇ = 𝜌𝑣𝑎 𝑆𝑎 = 𝜌𝑣𝑏 𝑆𝑏
(𝑃𝑎 − 𝑃𝑏 ) = 𝜌𝑣𝑏 (𝑣𝑏 − 𝑣𝑎 )
709
I just simplified the force balance so now, I am going to look at our energy balance ok;
simplifying that P a by rho g plus v a square by 2 g is equal to P b by rho g plus v b square
by 2 g plus h f; just by substituting P a minus P b and simplifying and see what where what
do you get for h f.
I am going to write this as v a minus v b because, v a is going to be the larger velocity; v

a minus v b all squared divided by 2 g is equal to v a square by 2 g into 1 minus v b by v
710
a whole a sorry 1 minus v b by v a square. But, we do know that v a S a is v b S b. So
therefore, v a square by 2 g into 1 minus v b so, it is S a by S b square.
𝑣𝑎2 𝑆𝑎 2
ℎ𝑓 = [1 − ( )]
2𝑔 𝑆𝑏
So, what would be our loss coefficient therefore, K e will be equal to so, that c expansion
loss coefficient and seems to be working fine.
𝑆𝑎 2
𝐾𝑒 = [1 − ( )]
𝑆𝑏
So, the assumptions that we might have made should be ok, this equation is has a name
yeah. So, so this is one of the very rare ones ok. So, again more any of the other cases as I
said you may have a to just take the values from other places, but for expansion loss
coefficient this equation should work.
711
Lecture - 60
Tutorial - 08
Hi, everyone I will discuss about the question number 8. I will read the question, the system
consists of a 1200 meter of 5 centimetre cast iron pipe 245 degree and 490 degree flanged
long radius elbows, a fully open flanged globe valve, and a sharp exit into a reservoir. If
the elevation at 0.1 is 400 meter. what gage pressure is required at point 1 to deliver 0.005
meter cube per second of water at 20 degree Celsius into the reservoir ?
712
So, this is our system, elevation 500 meter. So, as given in the question here you can see,
that 245 degree flanged long radius elbows, those are here; one is 45 degree, one is a here.
And, for 90 degree long radius flanged those are one 90 degree, 90 degree, then 90 degree
and 90 degree.
So, we have to calculate gage pressure at the point 1. So, for water and 25 20 degree
Celsius, we can take the density of water as at 20 degree Celsius, at 20 degree Celsius
density of water. We can take it as 998 kilogram per meter cube and also the viscosity
equal to 0.001 kilogram per meter per second and for the cast iron, we can take the
roughness factor as epsilon equal to 0.26 millimetre.
So, that epsilon by d the roughness ratio which will be 0.0052; where d mean, in the
question it is given d like diameter is equal to 5 centimetre equal to which is equal to 5
into 10 raise minus 2 meter, ok and L equal to 1200 meter. So, if we know the flow rate,
we can calculate the velocity and also from that we can calculate the Reynolds number.
So, V velocity equal to flow rate divided by area.
In the question phi it is given which is equal to 0.005 divided by area equal to pi by 4 d
square into 0.005 whole square. So, velocity will be equal to 2.55 meter per second. And,
from that we can calculate the Reynolds number; Reynolds number equal to d V rho by
mu; so, it will be equal to 0.05 into 2.55 into density 998 divided by 0.001 which is equal
to and from the moody’s chart, we can get the friction factor.
713
So, that you have to calculate and you will get and there are minor losses and which can
be listed as 445 degree long radius elbow; 45 degree minor losses, 445 degree long radius
elbow, k equal to 0.2; similarly, for 90 degree, k equal to 0.3.
And, for open flanged globe valve, k equal to 8.5. Similarly, for submerged exit, k equal
to 1; 1 and also for the entrance, sharp entrance k equal to 0.5 and if you are writing the
energy equation between the 0.1 and the 2 which is the reservoir surface we can write their
relation as. Since we are considering 2 as the reservoir, the gage pressure will be 0.
So, here we can neglect this and also velocity equal to 0 and these are the head loss due to
friction and this term is due to some a bends elbows and all. So, this term we can replace
it as V square by 2 g into fL by d plus sigma k [noise].
714
So, P 1 by rho g plus V 1 square by 2 g plus Z 1 equal to 0 plus Z 2 plus V 1 square by 2

g I mean, this term is due to friction, due to minor losses. So, from this we have to calculate
the a gage pressure at 0.1, that is P 1; this one we have to calculate. For that P 1 by rho g
equal to Z 2 minus Z 1 plus V 1 square by 2 g fL by k plus sigma k minus 1.
So, when we are substituting all the values, we will get 500 minus 400 plus velocity we
got it as 2.55 2 into 9.81 into and from the Moody’s chart, we have calculated the friction
factor which is equal to point 0, which equal to 0.0315 into length equal to 1200 divided
by and this is d, diameter equal to 0.05 plus and we are taking all the summation of this
loss coefficients. Those are here we have 245 degree, 490 degree and one open flanged
globe valve and also one submerged exit and also we have one entrance, sharp entrance.
So, we are taking all the summation, which is equal to for the sharp entrance 0.5 plus for
the 245 degree we are taking 2 into 0.2 and for 490 degree 0.3 plus, this is for the exit and
this is no this is for the open flanged globe valve and this is for the submerged exit. After
doing the calculation you will get 353 meter and from that we have to calculate P 1.
715
So, P 1 equal to 353 into rho g; which is equal to 353 into 9.81 into 998 which is equal to
3.46 mega Pascal.
Coming to question number 9, I will read out the question; the pressure drop through a
particle bed can be used to determine the external surface area and the average particle
size. Data for a bed of crushed or particles show, delta P divided by L equal to 84 pound
per in square per feet. For air flow at superficial velocities 0.015 feet per second the
measured void fraction is 0.47 and the estimated sphericity is 0.7. Calculate the average
716
particle size and the surface area per unit mass, if the solid has a density of 4.1 gram per
centimetre cube. How sensitive is the answer to an error of 0.01 in sigma?
So, here what we have to like mainly, we have to calculate the size of the particle and then
surface area of the particle and also I mean, how sensitive it is if the error is 0.01 in sigma.
So, first thing what we have to do is, we know Kozney Carman equation. So, I will write
that equation, which is applicable for Reynolds number equal to I will write the Kozney
Carman equation, where Reynolds number less than 1.
So, delta P by L equal to 150 mu V into 1 minus epsilon the whole square divided by phi
D p the whole square epsilon cube. Where mu is the viscosity, V is the superficial velocity
and epsilon is the porosity of the bed, phi is the sphericity, D p is the diameter of the pellets
and delta P is the pressure drop, L is the length of the tube.
So, here we have to substitute all the values, if I am taking phi D p the whole square in the
left hand side we can rewrite this above expression as mu V into 1 minus epsilon in the
whole square divided by epsilon cube into delta P by L. So, 150 into mu of air is, air equal
to 1.81 into 10 raise to minus 5 kilogram per meter second.
So, if I am writing sign in SI units 1.81 into 10 raise to minus 5 and velocity is given in
the question; which is equal to I mean, in the question is given in FPS units we have to
converted into SI units. So and velocity will equal to 4.572 into 10 raise to minus 3 into
epsilon equal to 0.47, so 1 minus 0.47 in the whole square divided by 0.47 in the whole
cube into delta P by L which is equal to 84.
So, finally, you will get so, after doing the calculation you will get 1.89 no, 84 is in sorry
[noise], 84 is in FPS unit. So, here we are taking SI units; so, I have to write in SI unit
delta P by L is 1900130.709. So, after doing the calculation you will get this as 1.89 into
10 raise to minus 10. So, from this we have to calculate D p which is size of the particle.
717
So, in the question sphericity is given as 0.7 and we also got the value of phi D p the whole
square, which is equal to 1.89 into 10 raise to minus 10. So, after substituting the sphericity,
we should we will get the size of the particle.
So, sorry phi D p will be equal to mean, first you can take root of this and after substituting
this sphericity you will get D p equal to 5.99 micrometer. So, this is first part of the
question. And, initially we have assumed the Reynolds number is less than 1.
So, we have to check whether we are getting Reynolds number which is less than 1. So,
for that we are just substituting the all the values ah. So, after substituting all the values
even if I am writing in FPS system, 1 or you can write in SI unit, whatever it is you will
get the same. In question it is given both in SI unit and FPS unit so, you can use whatever
you want.
So, after substituting all this you will get Reynolds number equal to 10 raise to 1.8 into 10
raise to minus 3 which is less than 1. So, we can use the above equation, which we have
used initially, so, our assumption is correct. So, I will like so, that we can say that the size
of the particle which we got is correct.
And, the next thing is we have calculate the surface area of the particle. For that we have
the relations sphericity equal to 6 V p divided by from this we will get, we have to calculate
surface area of the particle. So, we have to calculate the surface of the particle so, we can
718
write the this expression as S p by V p equal to 6 divided by phi S into D p; which is equal
to 6 divided by 0.7 into 5.99 into 10 raise to minus 6.
In the question the density of the particle is given as 4.1 gram per centimetre cube. So,
here you will get the value as 1.43 into 10 raise to 6 meter square per meter cube. So, after
substituting the volume, you will get S p equal to 1.43 into 10 raise to 6 into 10 raise to
minus 6 into 4.1 gram per centimetre cube. So, which is equal to 5.9 meter square by per
gram.
So, the surface area of the particle is 5.9 meter square per gram, and the others part is that
if we are taking error of 0.01 in sigma how sensitive it is? That we have to find out; so, in
the question initially, it is given that sigma equal to 0.47 so, if you are taking an error of
0.01 this sigma will be point 0.46.
If a sigma equal to 0.46, what is the like how sensitive it is? That we have to calculate, for
that we have the relation from the Kozney Carman equation that D p square equal to 1
minus sigma whole square divided by sigma cube.
So, D p 2 divided by D p 1 the whole square equal to 1 minus sigma 2 divided by 1 minus
sigma 1 the whole square into sigma 1 divided by sigma 2 the whole cube. So, which is
equal to 1 minus initially, it was point sigma 1 equal to 0.47, sigma 2 equal to 0.46, 0.46
719
divided by 1 minus 0.47 the whole square into 0.47 divided by 0.46 the whole cube and
you will get is equal to 1.086.
So, D p by this is so, you have to take the square root of this. D p 2 by D p 1 equal to 1.042
implies that D p 2 equal to 1.046 of D p 1. Which implies that there is 5 percentage
increment in the size, or we can say, error is 5 percentage. I mean, error is about 5
percentage for an error of 0.01 in sigma value. This is the final answer, ok.
Thank you.
720
Lecture - 61
Momentum and Kinetic Energy Correction Factor
So, I want to introduce you to the two terms that we did not talk about, but you might
actually come across these terms. And the when you are reading let us say books again
you are going to see this; one is momentum correction factor and another is a kinetic
energy correction factor ok. These two terms should really appear in any of the force
balances or energy balances that you write, but I have not been really taking that into
reconsideration. So, let me just introduce them to you.
So, the first one is momentum correction factor. So, this will come whenever we talk about
an integral force balance or integral energy balance. The reason is that let us say we go
back to our situation let us say the pipe flow and you have fluid that is flowing fully
developed.
And now you know that the profile is parabolic ok, then we define an average velocity as
integral; let us say if the velocity profile is given by u; u dA divided by integral dA; that
is what you would have done to calculate your average velocity right. Now, let us say we
721
are worried about; so let me write that as 1 by A; something that you have integrated
integral u dA.
∫ 𝑢 𝑑𝐴 1
𝑣̅ = = ∫ 𝑢 𝑑𝐴
∫ 𝑑𝐴 𝐴
Let us say we want to calculate average momentum flux or not; no point in telling average
actually momentum flux. We can calculate momentum flux either as an integral or we
could just use our averaged velocity itself ok. If you do an average velocity what would
you do? So, you if you want to calculate momentum flux, you would basically take the
volumetric flow rate times the; sorry not volumetric flow rate, mass flow rate times the
velocity right.
That is going to be the momentum flux. So, you would say that if I have you know a fluid
of density rho is going with the velocity with an average velocity V bar; so A into V bar
is going to be my volumetric flow rate, times rho will be my density, times V bar is going
to be my momentum in right. So, you would simply write the momentum flux in as rho A
V bar square that is a momentum flux in.
𝑀𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑓𝑙𝑢𝑥 𝑏𝑎𝑠𝑒𝑑 𝑜𝑛 𝑎𝑣𝑒𝑟𝑎𝑔𝑒𝑑 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 = 𝜌𝐴𝑣̅ 2
However, if you take let us say a small section. So, I will take a small section here and try
to write this as an integral. So, this is momentum flux based on average velocity, but if I
do momentum flux as an integral; I will say its rho times dA, times some velocity. So, u
into dA is going to be a differential volumetric flow right; in a differential element if I
consider of cross sectional area A; u into dA is going to be my volumetric flow rate through
that section, times rho is going to be density, times u is going to be my velocity; my
momentum flux and integrated over all area is going to give me the total flux and that is
nothing, but integral rho u square dA right.
𝑀𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑓𝑙𝑢𝑥 𝑎𝑠 𝑎𝑛 𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑙 = ∫ 𝜌𝑢2 𝑑𝐴
So, what have I done? I have just calculated the momentum flux either based on an average
velocity or based on the actual velocity. And if I have done it using the actual velocity; I
would have written down it as an integral ok; the question is; are these two quantities same
or not? So, if we have a flat velocity profile meaning that, if u was a constant; there is no
722
question u v equal to u average and everything will be same ok, but now we are saying
that this the if the velocity profile is not flat ok; if it is not a constant these two quantities
in general will be different ok.
So, that is because the integral; so because we have defined V bar as an integral of u dA
we do know that; if I do V bar square that should really be 1 by A integral u dA square
that will not be equal to 1 by A square, integral u square dA right. So, they are not same
actually I should write this. How do I do this? V bar square is this and that is not equal to;
I do not know, this is u square this is not integral of u square dA.
So, you can only write it like that; V bar is V bar square is not equal to 1 by A integral u
square dA right; average of square of average is not equal to integral of square. So,
therefore whenever you write down momentum flux you have to be careful. If you are
writing down your momentum flux based on your average velocity, you are actually not
doing the correct value you will be making a; you will be making a difference.
So, that difference is denoted by factor beta; beta is the ratio of integral rho u square dA
divided by rho A V bar square or that simply 1 by A integral; u by V bar whole square dA
ok; this is called the momentum correction factor.
∫ 𝜌𝑢2 𝑑𝐴 1 𝑢 2
𝛽= = ∫ ( ) 𝑑𝐴
𝜌𝐴𝑣̅ 2 𝐴 𝑣̅
Now, the question that I wanted to ask is should you have considered beta in our previous
derivation? And if we should have where should have come where should it have come
can you check hm. Yeah, so where we basically wrote down? Our force balance ok, I think
on the right hand side we said it is mass flow rate times velocity ok. So, that is really the
total momentum in and I said V is the average velocity, but writing down that expression
we have made an assumption that you know V square is actually giving you the total
momentum in.
But that is not really right there should have been a factor that should have come which
that factor is beta. So, in our derivations we have assumed beta to be 1 ok; so, therefore,
that calculation is not exactly right. For what kind of flow would you expect beta to be 1
or closer to 1; laminar or turbulent because you know that the profile is much flatter ok.
723
So, the calculations are expected to be better for a turbulent flow rather than a laminar
flow.
We have made a similar error now in the energy equation also; can you guess what that
would be? Because we have estimated kinetic energy based on average velocity should we
do that now because we know kinetic energy cannot be just average velocity square.
In fact, let us try writing down what would be the Kinetic energy, you know flow of kinetic
energy ok; if you do it based on average velocity you would write it as flow rate which is
rho into A into V bar into V bar square by 2. So, mass flow rate into V bar square by 2 is
the kinetic energy; in if you do it based on an average velocity.
But if you do it over a differential element, so you have a small area through which
something is going with a velocity u; area dA, you would say that is rho times dA, times
u, times u square by 2 and then integrated ok. So, the quantity that you have on the right
hand side is going to be the total kinetic energy in and that is the way you should have
calculated.
But you would simply write it as rho A V bar is the; if you do it based on the average
velocity you would have done it on the left hand side way ok. So, therefore, you know that
these two quantities are not same and therefore, you define a kinetic energy correction
factor; alpha as integral rho u cube divided by 2 dA divided by rho A V bar cube divided
724
by 2. And; that means, you have 1 by A integral u by V bar cube dA ok; so that is the
kinetic energy correction factor.
𝜌𝑢3
∫ 2 𝑑𝐴 1 𝑢 3
𝛼= = ∫ ( ) 𝑑𝐴
𝜌𝐴𝑣̅ 3 𝐴 𝑣̅
2
Should we have used the kinetic energy correction factor also in our previous derivation;
yes no? Yes; where; so let us write down our actual equations that we should have used.
Yeah, we should have had a factor of beta b they should have been a factor of beta a, that
should have given us our original or actual momentum flux. And here I should have had
an alpha a and I should have had an alpha b ok. And we have really said alpha a equal to
1, alpha b equal to 1, beta a equal to 1, beta b equal to 1.
So, the flow the calculations are better when these things are closer to 1; the ones which
we have derived. Any questions? In fact, I have not tried you could probably try if you
introduce these factors would you be able to simplify the equations this way and get
something extra or does it become impossible to simplify further. The idea is clear now?
The idea of a kinetic energy and momentum correction factor?
Contraction, you study by yourself ok; there is nothing really you can do theoretically, but
there are some arguments that you can see in the literature.
725
So, if you have done your assignment correctly; you might already have come across these
words and so on ok.
726
Lecture - 62
Pressure Drop in Pipes Connected in Series
(Refer Time: 03:21)
Pipes in series; let us say you have a system in which you have three pipes, of three
different diameters connected together ok. Of course, L 1 is much larger just for
convenience that we have drawn it this way and you have connected three of them together.
Let us say Q 1, Q 2, Q 3 are the volumetric flow rates through each pipe.
And delta P 1, delta P 2, delta P 3 are the pressure drops across each pipe ok. So, we have
been dealing with a single pipes so far; now I have connected three of them together. What
do we know about Q 1, Q 2, Q 3? They are same. So, Q 1 is equal to Q 2 is equal to Q 3;
what do we know about delta P 1, delta P 2 delta P 3? There will be individual contributions
that will add up to the total pressure draw and therefore, we can say delta P total is equal
to delta P 1 plus delta P 2 plus delta P 3 hm. So, what does it like; Q is like?
Current and delta P is like your? Voltage and the friction losses would have been your
really a resistance ok.
727
Question is Q calculate h total, how would you proceed? So, you know let us say you have
this system ok; you know this is the flow rate that you want to get through the system,
what is the pump that you have to select? What will you do?.
Q is given; that means, Q is equal to Q 1 equal to Q 2 equal to Q 3; you know flow rate
through each pipe; that means, you know v 1, v 2, v 3 velocity through each pipe. You
know f 1, f 2, f 3 friction factor in each pipe you know h 1, h 2, h 3 head loss in each pipe
and h total will be simply h 1 plus h 2 plus h 3 right. So, if you have a complex system like
that you could do it for individual pipes and then add up and then find out the total pressure
loss ok.
Of course, I have neglected the expansion and the contraction losses if you had that you
should have added that also; simple. Given h total calculate Q ok; you have this pipes in
series you also have got a pump; you want to know how much you can pump, how will
you do? Can you tell? So, I will take Q as a variable and then I do not know Q I need to
solve for Q. So, so I assume Q calculate v 1, v 2, v 3 ; that means, I can calculate f 1, f 2,
f 3 for some assumed value of Q.
And therefore, I can calculate h 1, h 2, h 3 and calculate and h total as h 1 plus h 2 plus h
3; check whether h total is as given; if not go back and correct your Q and you are right
that is the only way to do it at least that is the one way to do it ok. So, it has to be an
iterative procedure ok; the equations involved are non-linear you do know you will not be
728
able to really do anything especially because if you do not know v 1, v 2, v 3; you will not
know what Reynolds number you are going to operate.
So, you might be probably using laminar you know the 64 by Re as the correlation or you
have to use anything that will come from Moody’s chart or any of the other things. So, a
priori you would not even know whether you should worry about your roughness factor or
not because of that complication it becomes hard. So, one way to do it would be to start
with a Q ok, calculate and then find out if a total matches; if not go back and check it and
you know how to do this loops and so on. You know how to solve a set of non-linear
equations right, how would you solve a set of non-linear equations numerically; unless
you want to use a mat lab routine.
Of course, you can use the MATLAB routine and do it much simpler; so, either way ok.
So, it is basically solving a set of non-linear equations for a given system that is what it
turns out to be hm. We will see two more examples tomorrow and then, then we will get
into boundary layer theory ok. Any questions?
I do not know. So, if it was all laminar; maybe you can write down an expression and try
to simplify and get an expression for Q, but a priori; I do not think there is any other way.
729
Lecture - 63
Pressure Drop in Pipes Which Connected in Parallel
So, we so, what how does one proceed in terms of calculating the core connecting Pressure
Drop versus volumetric flow rate, when the fluid is flowing through pipes which are
connected in series. Now, let us look at the other configuration that comes or that one often
encounters is when pipes are connected in parallel.
So, let us take an example. So, let us say you have a fluid that is entering the pipe and let
us its divided into 3 parts ok. So, the pipe is divided into 3 parts ok. So, let us call the first
1 pipe 1, pipe 2 and pipe 3, all of them let us say have a length L and let the diameter be
different.
So, let us say this is D 1, this is D 2 and then this is D 3 and fluid or flow rate Q is coming.
So, it is through the flow there is a fluid that is entering that junction the flow rate is Q and
then it is passing in through 3 tubes before joining again and coming out. So, this is the
configuration that you have and your interest is to connect the relation between Q and
pressure drop. So, let us say Q 1 Q 2 Q 3 are the flow rates through each of the pipes and
let us say h 1 h 2 h 3 are the head losses through each of the pipes. So, again you can take
730
the same analogy as we saw in case of pipes connected in series, that the volumetric flow
rate it is like your current, the pressure losses, the head losses it is like your voltage.
So, you can see that the current should be divided into 3 while the voltage across the
section should remain same. In other words in this particular configuration h 1 should be
equal to h 2 is equal to h 3, in other words the pressure drop across the inlet and the outlet
will be same whether you take pass 1, 2 or 3. However, the flow rate the total Q should be
divided into Q 1 plus Q 2 plus Q 3, just the fact that the other volumetric flow rate will be
divided into 3 parts and going in. So, these are the two rules that will dictate the flow
configuration in this system.
Now, we could ask the question given h can you determine Q ? So, let us say you have got
a particular pump so, you know that it has got a certain head loss associated with it. And,
then you were interested in finding out what is the maximum flow rate that you can pump
through the system.
Now, you could write down the relations you can say that h 1 is f 1 into let us say L 1 by
D 1 into V 1 square by 2 g; h 2 is equal to f 2 L 2 by D 2 V 2 square by 2 g, h 3 equal to f
3 L 3 by D 3 V 3 square by 2 g. Now; so, this is how you would calculate h 1 h 2 and h 3,
but we do not know it what is h we know h 1 h 2 h 3, but we do not know V 1 V 2 V 3.
731
But that is simple because, actually if you look at it since h 1 is given, we can we need to
just calculate V 1, but to calculate V 1 you need to know what is f 1. And f 1 of course,
depends upon the Reynolds number and Reynolds number again depends on V 1. So, there
is no simple way to calculate it, except that you have to assume value for V 1, calculate a
Reynolds number or let us say calculate Reynolds number calculate f 1. And, then see
whether that f 1 is consistent with V 1 or in other words you’ve to calculate a V 1 and f 1
that is consistent with each other that can give you a measure a pressure head loss that is
given by h 1; because h 1 is prescribed in the problem.
Note I have I am specifying h so, h is equal to h 1 equal to h 2 equal to h 3 so; that means,
we know h 1, we know h 2, we know h 3; we do not know V 1 V 2 or V 3 are unknowns.
And our idea is to calculate V 1 V 2 V 3, but we need f 1 f 2 f 3 which are all functions of
Reynolds number which are in turn functions of V 1 V 2 V 3 so, this are really a non-linear
equation.
So, you assume a value of V 1, calculate the Reynolds number, calculate f 1 and see
whether that f 1 is actually for the for the Reynolds number that you have calculated. And,
this procedure can be repeated or the another way of doing it is to start by looking at a
value of f 1 you could also do this assume f 1.
So, since you know h 1 calculate V 1, then calculate Reynolds number and then calculate
f. And see whether it matches with the f 1 you have assumed if not correct it and so on.
And, this might be probably a better way of doing things either way you can do either you
start by assuming f 1 and calculate V 1 or start by assuming V 1 and calculate f 1 and
continue that procedures till you till you get a prescription.
And, you have to do the same thing for h 2 or rather V 2 and you can do the same thing
for V 3 and then you can calculate V 1 V 2 V 3. So, this is the procedure to evaluate ah
you know the flow rate through a parallel connection if h is given; you could also ask the
reverse question.
732
Given Q can you estimate h? So, what are the things that we know? h is given by h 1 plus
h 2 sorry h is equal to h 1 equal to h 2 equal to h 3, Q is equal to Q 1 plus Q 2 plus Q 3.
We know h 1 is f 1 into L 1 by D 1 into V 1 square by 2 g, h 2 is f 2 into L 2 by D 2 into
V 2 square by 2 g, h 3 equal to f 3 into L 3 by D 3 V 3 square by 2 g. The second rewrite
as f 1 into L 1 by D 1 into V 1 that is Q 1 divided by A 1 square. So, because if I divide
flow rate with area I am going to get velocity.
This is f 2 L 2 by D 2 into Q 2 by A 2 whole square, this is f 3 L 3 by D 3 Q 3 by A 3

whole square. Then I can write so, this expression tells me that I can write Q 1 is equal to
h 1 divided by L 1 by D 1 and then you have 1 by A 1 square to the power 1 by 2. So, I
am just calculating Q 1 from the expression. So, I have got h 1 and f 1 sorry into f 1 into
L 1 by D 1 into 1 by A 1 square that is going to be my Q 1.
1/2
ℎ𝑖
𝑄𝑖 = [ ] ; 𝑖 = 1,2,3
𝐿𝑖 1 1
𝑓𝑖 𝐷 ∗ 2 2𝑔
𝑖 𝐴𝑖
733
Similarly, I am going to get an expression for Q 2 is going to be simply h 2 by f 2 into L 2

by D 2 into 1 by A 2 square to the power 1 by 2 and I will also get an expression for Q 3
in a similar fashion. Then if I substitute here I will write Q is equal to the expression for
Q 1 which is really h divided by f 1 into L 1 by D 1 into 1 by A 1 square to the power of 1
by 2 plus h 2 by f 2 into L 2 by D 2 into 1 by A 2 square to the power of 1 by 2 plus h 3
by f 3 into L 3 by D 3 into 1 by A 3 square to the power of 1 by 2.
But, we know that h 1 h 2 h 3 are equal therefore, I will write Q is equal to h to the power
of half into 1 divided by f 1 L 1 by D 1, 1 by A 1 square to the power of half plus 1 by f 2
L 2 by D 2 1 by A 2 square to the power of half plus 1 by f 3 L 3 by D 3 1 by A 3 square
to the power of half.
1/2 1/2 1/2

1 1 1 1
𝑄 = ℎ2 ( ) +( ) +( )
𝑓1 𝐿1 1 1 𝑓2 𝐿2 1 1 𝑓3 𝐿3 1 1
[ 𝐷2 𝐴12 2𝑔 𝐷2 𝐴22 2𝑔 𝐷3 𝐴23 2𝑔 ]
So, what I have done really is that I have now gotten a connection between Q and h ok.
So, our problem originally was that given Q can you estimate h? So, by manipulating the
expressions we have got an expression between Q and h, again we have a problem because
we do not know what is f 1, f 2 and f 3 are.
734
So, the only way one can do is that assume f 1, f 2, f 3 given. So, what is the question given
Q? Given Q calculate h from the above expression so, but we do not know whether f 1, f
2, f 3 are right, but we have calculated h. But, we know that h is same as h 1 same as f into
L 1 by D 1 into V 1 square by 2 g ok. So that means, you can now calculate V 1 because,
in this expression you know h 1 you have assumed f 1 so, you can calculate V 1.
Similarly, from h 2 h 3 calculate V 2 V 3 and if you know V 1 V 2 V 3 you also know Q

1 Q 2 Q 3. And, if you know Q 1, Q 2, Q 3 you know that Q 1 plus Q 2 plus Q 3 should
be equal to Q given. If not then go back and change f 1, f 2, f 3 there and repeat the
procedure. So, this is a so, this is an iterative procedure and this iterative procedure needs
to be done because there is no other way to solve these expressions. So, you can start by
assuming f 1 f 2 f 3, then you can calculate a Q from that sorry given the Q you can
calculate an h from that.
And, then you can calculate V 1 V 2 V 3 and calculate again see whether the V 1 V 2 V 3
that you get is balancing your constraint Q 1 plus Q 2 plus Q 3 equal to 0. So, let me tell
that again we start by assuming f 1, f 2, f 3 then we use the expression connecting Q on h
to calculate h. And, since we know h you can use the individual expressions like h 1 equal
to f 1 L 1 by D 1 V 1 square by 2 g, those expressions to calculate V 1 which means you
can calculate Q 1, Q 2, Q 3. And, then you can check the constraint whether Q 1 plus Q 2
plus Q 3 is given to be the total flow rate and this procedure can be repeated.
735
So, this is just one way of doing it, essentially what you have is a system of equations and
you need to solve them and it is your choice really whichever way you want to do it, this
is just one really algorithm for that ok. So now, given that so, we have seen how one should
think about when pipes are connected in series, how should one think about when pipes
are connected in parallel, you could also have junctions.
So, let us look at a case where you have you know junctions, you can have elevations. So,
far we basically said we are going to look at pipes that are all you know horizontally
aligned and therefore, there was no elevation change.
736
Lecture - 64
Pressure Drop in Pipes Which Connected at Junction
Let me give a new example; can you know which has got junctions and elevations. So,
you have let us say some reservoir from which a pipe is coming. Let us call this reservoir
1, this is another one, and then there is another one ok. So, this reservoir 1, this as reservoir
2, that is reservoir 3, and all of them are filled with some fluid ok. Let us say this is at an
elevation is z 1, this is at an elevation is z 2, and this is at an elevation is z 3.
Now, you, so the what is going to happen is that the fluid from the highest reservoir or the
highest two reservoir is going to flow into the third reservoir. You do not know what it is,
and let us say your interest is to calculate what is the flow rate through each of these pipes
ok. So, let us assume that Q 1, Q 2, Q 3 are the flow rates, but you do not know what is Q
1, Q 2, Q 3, you do not know what is the pressure drop except that you just know that it is
all you know at different elevations. So, here we will have P atmospheric here we will
have P atmospheric here we will have P atmospheric.
So, the flow rate Q 1 is determined by the atmospheric pressure or the and then this
hydrostatic head. So, it is really is z 1 that determines what is the flow rate through Q 1,
737
and that would generate a pressure difference between the top of the reservoir and this
point at the junction. Similarly, there will be a pressure drop across your second pipe, there
will be a pressure drop across your third pipe.
So, you can write down general equations. You can say that your h 1 which is the pressure
drop across the first pipe is given by delta p 1 by rho g that is a pressure drop plus you
know delta is z 1, so that is essentially so that is the definition of our head loss ok. So, so
far we have not been taking you know z into consideration, because we never worried
about elevation changes, and that is equal to f into l 1 by D 1 into v 1 square by 2 g ok.
Where v 1 we do not know, but we know it in terms of Q 1. So, let us write it as f 1 into L
1 by D 1 into Q 1 by A 1 whole square and then there is A 1 by 2 g ok.
Similarly, h 2 is equal to delta p 2 by rho g plus delta z 2 that is equal to f 2 into L 2 by D

2 into v 2 square by 2 g or that is equal to f 2 into L 2 by D 2 into Q 2 by A 2 whole square
into 1 by 2 g. Similarly, h 3 equal to delta p 3 by rho g plus delta is z 3 equal to f 3 into L
3 by D 3 into v 3 square by 2 g that is f 3 into L 3 by D 3 into Q 3 by A 3 whole square
into 1 by 2 g ok.
Δ𝑃𝑖 𝑓𝑖 𝐿𝑖 𝑣𝑖2 𝑓𝑖 𝐿𝑖 𝑄𝑖 2 1
ℎ𝑖 = + Δ𝑧𝑖 = = ( ) ; 𝑖 = 1,2,3
𝜌𝑔 𝐷𝑖 2𝑔 𝐷𝑖 𝐴𝑖 2𝑔
Now, what we know is that h 1, h 2, h 3 should be different, because the z 1 is z 2, z 3 are

different, but delta p 1, delta p 2, delta p 3 should be same. Because on one side, it is all
exposed to atmosphere and delta p 1 is what the pressure difference between the top of this
reservoir and this junction. Similarly, delta p 2 will be the pressure difference between the
top of the reservoir, and this junction in which three pipes join and then so it is delta p 3.
So, we would know delta p 1 is equal to delta p 2 is equal to delta p 3, so that is one thing
that we know and therefore, what you can do is you can start again you need an iterative
procedure let us say assume delta p. So, let us call it equal to delta p assume delta p. So, if
you know delta p, then what you can do is that let us take the first equation which is that.
So, let us write it down if you assume delta p, then delta p by rho g plus delta is z 1 is equal
to f 1 into L 1 by D 1 into Q 1 by A 1 whole square into 1 by 2 g. So, you assume f 1,
calculate Q 1, check Reynolds number, check Re gives assumed f 1 that is this f 1 if not
continue till you get a f 1 Q 1 consistency ok.
738
So, then you can do that and repeat, the same for Q 2 and Q 3. And once you get Q 1, Q
2, Q 3, then you know that Q 1 plus Q 2 plus Q 3 should be equal to 0. If not, then you
have to go and change your assume delta p, and the process to be continued. So, what am
I trying to say I am trying to say that you can for example, assume delta p, you know the
equation for each pipe, so you can solve for what is velocity in each pipe. But when you
try to do that you would not know what is the friction factor, so you have to assume a value
friction factor find out the velocity in each pipe ok, and you have to do that in an iterative
manner, so that you get an f 1 v 1 or f 1 and Q 1 combination that satisfies the assumed
delta p.
And then you can do the same thing for second pipe, you can do the same thing for third
pipe that means you have gotten the flow rate for first second and third pipe. But you also
know that you are actually looking at a junction and there cannot be a net flow, so Q 1 plus
Q 2 plus Q 3 should be equal to 0. So, you can check that condition and if that condition
is violated, that means, this delta p that you have assumed the one which you started with
is actually the wrong one, and you need to change that value. And then continue this
procedure still till you find that Q 1 plus Q 2 plus Q 3 equal to 0.
So, this for example one way of solving the set of equations. As I mentioned earlier you
can come up with any procedure, so that this condition is satisfied. So, the point in note is
that as far as the junction is concerned, there is no net flow through that or whatever is
739
coming into that junction the flow it should be such that that fluid flow should go out or in
incoming flow should be equal to outgoing flow. As long as you impose that condition and
your solution satisfies that that is going to be your solution.
So, in general whenever you have a system it is you know you let us say you have a pipe,
it might be complicated one, this is something like this cames. So, there is a fluid that is
going to go like that, there is a bifurcation, and then it goes, it goes that way, it goes this
way and so a complicated system. What you can do is that given this that, if I take any
loop let us say a loop that you start from there, and finish that loop. And if I am coming
back at that point then you can insist that there cannot be a pressure drop across a closed
loop, because I am starting from the same point and ending at the same point.
So, you can has insist that there cannot be a net pressure drop across a closed loop. Second
thing that you can say is that there cannot be net flow through any junction. So, these are
the two constraints that you should impose. And if you are trying to impose that constraint
and then you are going to get all values for velocity, you are going to get Reynolds number
for each pipe and then you can insist that all you know friction, all velocities, all velocities
that you calculate or flow rates that you calculates should satisfy friction factor correlations
Moody’s chart.
So, as long as you write down you know set of correlations that satisfy these three
constraints, then you basically will have a system that needs to be solved ok. And most of
740
the times you would end up the solution being in an iterative form because you will have
more than one unknowns, and the equations may not be straightforward to do it. And
remember when I have done all these calculations, I have assumed all minor losses to be
0, but the minor losses should be accounted for because we know that some it can be it can
play a major role.
So, then it will really depend upon the system that you are looking at. So, so for example,
when you write down the total head loss, you should write down the total head loss due to
the length of the pipe as well as that due to the minor losses like bends and valves and so
on in the system.
So, it becomes a very system specific configuration. But what we have done is really
looked at what are the constraints that you should impose ok. So, the constraints would be
one is coming from the mass conservation which is about how a flow gets divided or added
up. The second is about coming from the force balance, which is about your pressure loss
or how does the pressure drop whether they get added or equally divided and so on.
So, these are the two rules that you should remember. I just realized that while I wrote the
first problem, I have not taken a one by 2 g into consideration. So, here when I replace v
1 with Q 1 by A 1 there is a factor of 1 by 2 g is a factor of 1 by 2 g is a factor of 1 by 2 g
which would appear in expressions for Q 1, Q 2 and in Q 3. And when you have substituted
741
that would again come, so that is 1 by 2 g, 1 by 2 g, 1 by 2 g, 1 by 2 g, 1 by 2 g and 1 by
2 g, it is just coming as a factor ok.
742
Lecture - 65
Boundary Layer
So, we have almost done with the Fluid Mechanics part ok, where we learned how to
calculate essentially pressure drop and the whole idea was to get the right pump that is
suitable. Now, but as we saw in the beginning of the course ok, this fluid flow is not just
confined to analysis of pipe flow.
There are several situations I mean in almost every place that you look at fluid flow
becomes important ok. So, there are many more things that needs to be done and as
chemical engineer one of the context in which again fluid flow will become important is
in the context of boundary layers, so that is what we are going to start looking at from
today onwards.
So, boundary layer is nothing but you know the region of the fluid flow that is affected by
a nearby solid boundary ok. So, for example, if you have a flat plate, let us say a solid plate
and if there is a fluid that is on top of it that is flowing, there will be a region which is
affected by the solid plate and then there will be a free stream that is outside. So, when I
said free stream that is a region which has which does not feel the effect of the solid plate,
743
but there would be a region which is close to the plate where its experiences or in other
words the this presence of a solid boundary will exert some drag, it will retard that fluid
motion ok, so that region which is retarded by this by the solid plate that is what you call
as the boundary layer.
Now, this boundary layer is very important for example, in heat and mass transfer. So,
either you might have learned or you might be learning that you know you have situations
where you know either heat is going to get conducted from the solid plate to a nearby fluid
just like in a heat exchanger and that process ok. So, one of the ways of happening heat
transfer is conduction, you could have as well as convection which means it is determined
by the fluid flow right.
So, you would want to know so here we have a typical situation where the fluid is very
near to a solid, let us say solid is at a different temperature from the fluid, you want to
know how much is the heat transfer happening and you would then really want to know
what is the velocity profile, because it is not just due to conduction, but it is going to be
convection as well ok. Similarly, you can have mass transfer also.
For example, that is a solid ok and let us say it is dissolving into the fluid or let us say
some crystallization is happening. So, there is going to be a mass transfer between the solid
region and the fluid region. Again the rate of mass transfer is going to depend upon how
the fluid flow is behaving there.
If the fluid flow is extremely slow, then diffusion is the only thing that will happen, but in
general you would want to you know increase the speed of or rate of these processes, you
would probably be you know pumping fluids its sufficiently large velocity. And therefore,
there will be a boundary layer formula formation, and the mechanics in the boundary layer
is actually going to determine how fast these you know transfer processes are going to take
place ok, so that is one of the reasons why boundary layer is so much relevant to chemical
engineers.
So, there is so this concept of boundary layer ok. It happened I mean the concept came up
somewhere early 1900s and this was done by a person Prandtl. I think we heard this name
sometime before or you would have definitely heard Prandtl number by now. So, see
people were initially thinking about or let us talk about like this. So, Reynolds number
right, Reynolds number is defined as the ratio of two forces. What are the two forces?
744
Inertial by viscous forces. And people thought if we keep increasing the Reynolds number,
then inertial forces are going to be much larger than viscous forces right. Or we would
think that the effect of viscosity becomes much smaller; much smaller, we can actually
neglect all the viscous effects, and we can only talk about just inertial effects.
Or in other words, if you talk about Bernoulli’s equation right, this Bernoulli’s equation
contain viscous effects or you neglect viscous effects when you derive Bernoulli’s
equation, you neglect it right. So, you would think that if I am at a very high Reynolds
number, I should be able to solve my fluid flow by just using Bernoulli’s equation and I
can completely forget about the viscosity of the fluid ok. So, that was the initial thought,
and then some of the experiments of what they did is they looked at flow past body. So,
let us say we will talk about flow past a cylinder ok.
So, if you have a cylinder and then fluid is approaching and then it basically goes that way
ok. So, we now know that the solid is going to you know exert or let us talk about this say
the fluid is going to exert a force on the solid object ok. And that forces can have actually
two components ok. A component that is parallel to the flow direction, and a component
that is perpendicular to the flow direction.
So, the component that is parallel to the flow direction is what we call as drag ok; the
component that is perpendicular to the flow direction is typically called a lift force and you
know why the lift force is coming because that is typically the case when you talk about
aeroplanes and it is that lift force that helps you to fly the aeroplane ok, so that is where
the origin is.
So, you typically you can talk about a drag force and a lift force. Now, people thought, ok,
we will calculate these forces and when they try to come up with you know try to compare
experiments with Bernoulli’s equation. They found that yes they are able to predict lift
forces very accurately, but not drag forces. Now, we associate drag so much with viscous
forces ok, but we also say that if you go keep increasing your Reynolds number, viscous
if it should become lesser and lesser important and therefore, we should be able to predict
that drag force just from Bernoulli’s equation.
But when we did experiments, the experiments never matched with you know the
predictions from Bernoulli’s equation ok. If you neglect the specifics the calculations were
never correct ok. So, then actually Prandtl came up with this idea of boundary layer. What
745
he said is that however large Reynolds number is there is going to be always a thin layer
of fluid very near to the boundary where you cannot neglect viscous forces.
So, what he said is that even in this case, there is going to be a very thin layer near the
solid region, where you cannot you know neglect the viscous forces. And therefore, you
need to account for that particular you know mechanics of whatever is happening in that
layer to predict what is the viscous forces or really the drag force ok and that is way
Bernoulli’s equation was not able to give you a good result for the drag force. So, that is
actually the origin of boundary layer ok.
So, the layer could be very thin. It was just like our roughness ok. You would think that
roughness does not play any row, but roughness actually plays a major role. Similarly,
here the boundary layer might be so thin, but it could actually be very important in
determining what are the forces that are going to act on the particle ok, so that is where the
origin of boundary layer has come from.
And now I mean at various places, you see this concept ok. You would think that the
majority of the forces or the main mechanism that is going to be acted is something, but
very near to something else there could be a thin thing which you might generally tend to
neglect could actually be a very important thing.
So, it depends upon situations and so on. So, if people call all such things as you know
boundary layer, boundary layer concept and so on. So, this is why this is how you know
the origin of boundary layer happened and then of course, then you know in heat transfer
and mass transfer everywhere you start seeing the importance of it, so that is the history.
So, let us look at. So, there are two things that we would be interested in this course ok.
One is to see what is going to be the thickness of boundary layer ok, because that is a
region where the fluid flow is seeing the solid plate ok and then we you want to of course
it not in this course, but you may want to talk about heat transfer across that layer and so
on. So, you want to calculate what is the thickness of the boundary layer; second is that
we will also use this concept to explain flow separation.
So, you remember in one of the classes I was talking about expanding sections and so on
and I said the flow could get separated ok. So, the analysis of boundary layer will actually
tell you where does the flow separation happen you, why does it happen and so that is the
746
another thing that we are going to look for. And it is important to know whether flow
separates or not because that will also help you to determine how you know whether the
transfer process of heat and mass transfer are going to be affected or not. So, these are the
two aspects that we will look at.
So, we will look at a very simple situation ok. We will only worry about a flat plate nothing
else that is a flat plate and we will say there is a fluid that is coming at a constant velocity.
So, the velocity profile is a constant ok and it is approaching the solid plate. And as soon
as it approaches the solid plate what would happen the fluid element that is very close to
so the last one when it approaches there at this region, it is immediately going to be
retarded by the solid plate ok.
While everywhere else it is still going to be moving with so let us call this a velocity U
infinity ok. So, the fluid is moving with a U infinity velocity doubt ok. And then as soon
as it touch the edge of that plate that fluid element is going to get slowed down and then
the next fluid element and it keeps going right.
So, if I look at if I draw the velocity, so up to here the velocity profile was flat or vertical,
immediately as soon as it touches, there will be a small region let us say I should draw
this, there will be a small region where velocity would the middle be you know affected
so much and immediately after that it will regain its U infinity, so that is what is going to
be the velocity profile ok, so that small region is where it is affected by the solid plate and
747
everywhere else the velocity is U infinity ok. But now you as it you know goes downstream
acid proceeds further, even more and more, layers are going to feel the effect of the solid
plate ok, because this already is accelerating the drag.
So, after a while if you look at further downstream, you will see that it is something like
that ok. So, that much is the region which gets affected by the solid plate and that will keep
going. In other words, this, the thickness of this layer would keep expanding ok. So, now,
it is even larger ok. So, the thickness of this thing keeps expanding. So, this is where we
are going to define our boundary layer thickness. We want to define boundary layer
thickness as the region which is affected by the solid plate ok. Now, in principle all the
fluid up to infinity will also see it ok.
So, if you look at the solid fluid here ok, let us say this is the solid plate, and the fluid layer
it gets slowed down, and it really reaches you know when it is really far, but we do not
want to talk about that. So, we need to you know give some criteria of telling what is the
boundary layer thickness and we can say that let us say whenever the velocity reaches 99
percentage of the free stream velocity, we will call this as a boundary layer thickness or
you could call it 99.99 percent whatever you like ok. So, we will say that whenever the
velocity is 99 percentage of U infinity, we will call that as a boundary layer thickness.
So, if you mark that boundary layer thickness, so here you are starting from the edge and
then you are starting from the edge and then you know it keeps increasing ok. So, that is
what the boundary layer is going to be, so that is just an imaginary surface that you have
drawn to identify the boundary layer part of the fluid flow. So, your interest is to find out
what is that right. So, typically we denote it with a with delta as the thickness, and it turns
out this we will derive delta of x. So, this is my x-direction delta of x ok, delta at any place
in x divided by x is approximately equal to 5 divided by square root of Re x.
𝛿(𝑥) 5
=
𝑥 √𝑅𝑒𝑥
So, when I said Re x the Reynolds number is defined based on the distance from the
leading edge ok. So, typically we will define Reynolds number based on a given length
scale, it could be the radius of the pipe or you know the size of the particle, here there is
no other length scale other than really the distance from the plate itself.
748
It does it decrease with x. So, let us see delta of x basically is 5 x divided by square root
of x U infinity rho divided by mu, so that is actually going as x to the power of half. So, it
in; it is increasing with. Why it increases? So, as we said so as soon as the fluid touches
the leading edge ok, only that fluid particle that is touching the leading edge will
experience the effect of the solid plate. Now, what is the problem with the guy who is just
behind you?
So, the only the fluid particle that is touching the leading edge will get retarded,
everywhere else it is going to be still coming with that velocity of U infinity right. But as
it enters a little bit more, two of the layers will start experiencing it because this is. So,
what is the solid plate doing, the solid plate is retarding the fluid motion right. So, it
basically has to get transferred right across various layers. So, here there is a fluid that is
coming with a strong velocity.
So, only the fluid particle that is going to touch is going to be experiencing that. But then
the next layer we will see, but a little downstream, and the next layer we will see it a further
downstream. So, therefore, the boundary layer will keep growing. Does that answer your
question? Yes.
Yes, you are right. So, the well also at the moment in this picture, I have assumed that the
fluid outside the boundary layer is actually going all the way to infinity ok, so that is the
only reason I have assumed that ok. But you are right that yes the other region the velocity
will increase and I will point out that a little later also today. Any other question?
So this is so for a laminar flow if so the flow field in the boundary layer could either be
laminar or it could be turbulent ok. If the flow in the laminar in the boundary layer is
laminar, then the thickness of the boundary layer is given by this expression delta of x by
x is approximately equal to 1 by square root of Re ok, x, where Reynolds number is defined
based on the distance from the leading edge ok. And if you simplify that, you will find that
the boundary layer thickness actually increases with x in a square root form ok.
749
Now, what happens is that, so let us say that shows your growth of boundary layer ok. If
the Reynolds number can get sufficiently large ok, then even though the flow would be
laminar at the near the leading edge of the plate after a certain point, it will actually go to
a turbulent ok. It can basically turn to a turbulent flow. So, you would see that if you were
to actually track the boundary layer thickness you will see that it nicely you know grow
like a root x and then suddenly there will be some point in which is it will start you know
increasing suddenly ok, it will start increasing suddenly and then again it will slow down.
So, what happens is that up to here, the flow would be laminar. And beyond that point the
flow would turn turbulent ok. Now, why that happens is not a part of this course, I, we
cannot say. But what is going to happen is that the layer in the boundary layer ok, you
already know that we have already talked about that, that will constituted by three different
parts. There will be viscous sub layer very near to the wall, then there can be an overlap
layer and then there will be turbulent boundary layer and then there can be a free stream
ok.
So, from here what happens is that, so remember, so this region the region which is laminar
basically will become smaller and smaller and will become a very thin layer which is what
we called as the viscous sub layer or wall layer. And then there if there will be another
region which is our overlap layer.
750
And then we have our outer turbulent layer and then outside is where we have our velocity
as U infinity or the free stream velocity. So, this region the delta by x is approximately
going as 1 by square root of Re x. This region is where delta of x goes as Re x to the power
of 1.5 much faster and much downstream delta by x goes as 1 by Re x to the power of 1
by 7, which is much weaker ok. So, this is really the onset of transition. So, this is the
structure of the transition from laminar to turbulent layer, turbulent flow in a boundary
layer. Do you want to clarify something? Ask.
Re x for 1.5, yes. So, in the laminar, it is going to be root ok, it is x to the power of half or
I have written this delta by x is 1 by root of Re x. In the after immediately after the onset
ok, it is going to be much faster, it is x to the power of 1.5 and then much further down, it
will be delta by x is 1 by Re x to the power 1 by 7. So, delta will go as x by x to the power
of 1 by 7. So, it is x minus x to the power of 6 by 7. Now, 1 minus 1 minus 1 by 7, so that
is how the growth is going to be.
What we are going to do now, let me then now answer his question. So, we have seen this
before ok, correct. So, we earlier we have looked at you know somewhere there it is so
much further downstream ok. We never talked about the region, which are very close to
where the transition is happening.
So, it is really those three layers are formed and stabilized and so on. See from the left if
you see is laminar. So, it is not that at one point it turns turbulent ok. It is a top actually
that turns turbulent and then that affects the cascades down. So, it basically takes a while
for things to stabilize. So, it is just going to say that this boundary layer formation ok, that
is really what we talked also when we talked about the entrance length.
751
If you remember in the first class, I might have spoken that if you have a channel and if I
say that if fluid of constant velocity comes as soon as it touches the edge, it basically slows
down, and then the region which is affected by that solid plate will grow like that ok. So,
inside this layer, so the fluid is the velocity profile is going to be like this. So, the regions
which are very close to those plates, the velocity slowed down, but in between the fluid is
still coming with the velocity which is uniform ok, so that is really basically the boundary
layer growth ok.
And whenever that fluid flow reach the when the boundary layers actually merged beyond
a point, so this will happen; this will happen and it will merge and then after that we had
everywhere in some sense boundary layer only that is what we called as the fully developed
flow and that is what we analyzed and so on. And to answer your question, this is the
region where you know you would see that, so if this was coming with a velocity of U
infinity ok.
Here this region and this region has slowed down, so that means the this part of velocity
field should be having a velocity larger than U infinity. In fact, you can use that mass
conservation to calculate the velocity in that region ok. So, this is what you know really
boundary layer was in the context of what we had earlier. What we are going to do now is
try to derive an expression for boundary layer thick.
What is larger?
752
So, see something is coming with a uniform velocity ok, the entire mass that is coming
should go out, but part of it has retarded, that means, the other part should get accelerated
right, then only you can maintain the total flow rate. Otherwise, there is a mass loss right
or in. So, as long as the fluid that is coming in has to go out, part is slowed down means
part is sped up.
Because, the other parts have been slowed down.
Yes, which would be actually in terms of pressure rather than anything else. So, it is not
that we so viscous forces are only going to anyway slower down or you can think about it
as you know things so you so fluid was coming, but it is getting squeezed. So, it is actually
speeding up that is what really it is ok.
753
Lecture - 66
Boundary Layer - Momentum Integral Analysis - 1
So, this thing is called momentum integral estimate, we will do an integral analysis on the
boundary layer, get some you know insight into it, we will do a little bit of differential
analysis and get a little insight into it. So, let us first do the integral version, my solid plate,
my boundary layer x y delta the boundary layer thickness, the uniform velocity u infinity.
I am going to consider a streamline that is going to reach this point so, that is going to
reach this point let us say that is a streamline; so, rewrite a stream line.
So, what is a streamline? If I draw tangents to the streamline I will get velocity vector. Can
there be flow across a streamline? No, there will not be a flow across a streamline because
by definition the tangent is the direction in which velocity is so, there cannot be a flow
across the stream. So, let us say this height is h, the height on the right hand side is delta
and let us say b is the width of the plate in z direction ok. So, I am looking at x y plane is
z direction is the plane per line perpendicular to the screen that you are looking at. So, b
let us say is the width in that particular direction which let us imagine to be very large.
754
And, I want to write down a force balance to this control volume which is this that; so, that
is the control volume that I have. So, on the left hand side it is bounded by the uniform
velocity profile up to height of h. On the right hand side the height is delta the boundary
layer thickness, on top it is bounded by a streamline, on the bottom it is the plate; that is
the volume that I am going to look at and I am going to write down a force balance. The
control volume is clear? So, what do we write? For when the flux in minus out is equal to
total forces, the only force so, we are writing it down on the fluid ok.
So, the solid is going to exert some retarding forces, some drag force on the fluid that is
flowing right. So, let me call that force as D. So, what is D? Total drag force exerted by
the solid plate on the fluid in the control volume. And, that should be equal to the flux in
minus flux out yeah, you have some flux in what the flux in is where is the flux. So, there
is only two places where the fluid is coming and leaving right. The fluid is coming from
the left hand side through a height of h, the fluid is leaving through the height delta. Since,
I have chosen a streamline there is no fluid flow across the streamline.
So, there is no fluid flow through the top part ok, bottom part anyway it is a solid plate so,
there is no fluid flow through the plate. So, I have to only worry about the fluid that is
entering through h and the fluid that is leaving through delta ok. So, what will be the
momentum of the fluid that is entering through left hand side h? We will say that so, b into
h is the cross sectional area times U infinity is the flow rate times rho is going to be my
mass flow rate times U infinity is going to be my momentum flow rate.
So, that is going to be in the plus x direction minus what is exiting will be if I do the same
thing except that. So, here you know on the left hand side the flow is uniform, but on the
right hand side the flow is not right there is a profile. So, I cannot do just like that I need
to write down an integral.
So, if I take a small section I will say let us say b d y is my cross section times let us call
u as the velocity. So, that is my volumetric flow rate times row is my mass flow rate times
u is my momentum flux integrated from 0 to delta would be the flow rate or the momentum
flux out ok. So, that is going to be in minus out is equal to the total force be exerted or
experienced by the control volume fluid that is ok.
755
Let us write it neatly D is equal to rho b h U infinity square minus integral 0 to delta rho b
u square dy. Let us also write down a mass conservation principle fluid in is equal to fluid
out.
𝛿
2
𝐷 = 𝜌𝑏ℎ𝑈∞ − ∫ 𝜌𝑏𝑢2 𝑑𝑦
0
So, what is the fluid in? Rho b h U infinity should be equal to integral of rho b y u dy,
sorry rho b dy times u so, that is correct. So, remember this is my force balance, this is my
conservation of mass. Any doubts on the equations? Total drag force is node 0. Why is the
total drag force 0?
That is what we have called as D so, this oh this is D ok, D it is not 0. I will try to write it
better maybe yeah, I mean our interest is to calculate really that force and connect it with
the thickness ok. So, mass conservation if you simplify what do you get? h so, rho goes b
b goes h is equal to 1 by U infinity integral u d y ok, take that and substitute into your force
balance D is equal to rho b U infinity square, instead of h I will write 1 by U infinity
integral u dy minus integral 0 to delta rho b u square dy, get one of them.
So, that is equal to rho b so, this limit is also 0 to delta right. So, rho b integral 0 to delta
U infinity u dy minus rho b integral 0 to delta u square dy is equal to rho b U infinity
square into integral. No, we do it yeah that is fine I think U infinity square u by U infinity
756
minus u square by u infinity square d y 0 to delta, is this right ok. So, that is really the drag
force yeah, I think that is fine.
𝛿 𝛿
1
𝜌𝑏ℎ𝑈∞ = ∫ 𝜌𝑏𝑢 𝑑𝑦 → ℎ = ∫ 𝑢 𝑑𝑦
0 𝑈∞ 0
𝛿
2
𝑢 𝑢 2
𝐷 = 𝜌𝑏𝑈∞ ∫ [1 − ] 𝑑𝑦 = 𝜌𝑏𝑈∞ 𝜃
0 𝑈∞ 𝑈∞
Now, what we are going to do is that we are going to give this part a name and that is
momentum thickness theta ok, it is really a measure of drag force ok. So, if you really look
at it D divided by rho b U infinity square is equal to theta, because that integral is theta ok.
So, that is nothing, but just a measure of total drag exerted by the plate on the fluid. Now
so, this is how we have gotten an expression for the drag force on the fluid, but you also
know it can be evaluated from your shear stress ok.
In other words you know that D is nothing, but tau w if let us say tau w is the wall shear
stress exerted on a small element d x and b integrated. So, the drag force the total drag
force should be an integral of the wall shear stress right. The force exerted by the plate on
the fluid is through the viscous stresses so, that is nothing, but what you call as the shear
stress; so, that I have this integrated. So, why have I done the integration because, it need
not be a constant as we saw earlier.
757
So, if you look at you know the first, second, third and so on the fluid slope is going to be
different ok. So, the shear stress is going to be a function of x, shear stress will not be a
constant along x. So, I said if I take a small area of length dx and width b so, it is a b dx
and then on that tau w is acting. So, tau w dx is going to be the force on that small element
and if I integrate it over all my length up to x I will get the total force ok.
So, that means if I do a d D by d x I am going to get b times tau w or in other words shear
stress is nothing, but the derivative of the total drag force with respect to x. I could also
758
write d D by dx is equal to rho b U infinity square d theta by dx right where I had defined
it theta some integral. And therefore, if I use these two what I see rho b U infinity square
d theta by d x is really b times tau w or tau w is equal to rho U infinity square d theta by
dx ok.
𝑑𝐷 2
𝑑𝜃
= 𝜌𝑏𝑈∞
𝑑𝑥 𝑑𝑥
𝐷 = ∫ 𝜏𝑤 𝑏 𝑑𝑥
𝑑𝐷
= 𝑏𝜏𝑤
𝑑𝑥
2
𝑑𝜃
𝜏𝑤 = 𝜌𝑈∞
𝑑𝑥
So, it is really rewriting the wall shear stress in terms of some derivative, it is just for
manipulation alright. Now, so far we have not assumed anything about the flow profile
inside the boundary layer ok. So, we have done an integral estimate; so, this should be true
whether the flow is laminar or turbulent ok.
So, this should be true whether the flow is laminar or turbulent, yeah steady state. Well,
what is steady state? Yes, it is in terms of the incoming fluid, it is a steady thing yes.
759
Yes, but turbulent flow you have to be careful, you must be talking about an averaged
quantity.
Why should it be a straight line?
No in fact, the path of the particle will basically go a little up oh ok, I cannot think about
any other explanation that why would it go up, but the fact is the, oh yeah one thing is this
continuity would say that ok. So, if you look at continuity dou u by dou x plus dou v by
dou y is equal to 0, we know that dou u by dou x is not 0 so that means, that dou v by dou
y is not 0 ok. So that means, v can be function of y ok;
so, the v basically will be a function of y; that means, that at every point the fluid will not
only have an x velocity, it will also have a y velocity. And, if it does not have that y
velocity, it will have to violate the continuity equation or in other words it is basically a
consequence of mass conservation.
What is that?
No, we know this this dou u but, this mass conservation is true whichever you do right
whether it is laminar turbulent, whether you talk about average fluctuations and so on. So,
it will be true in general ok. So, the v velocity develops because it is because the mass has
to be conserved.
So, it must be have a better explanation in terms of the total flow rate, but I cannot think
of anything as such ok. So in fact, that is what we had used whenever we talked about our
fully developed flow. When we talked about a fully developed flow we said our dou u by
dou x is 0 and therefore, dou v by dou y is equal to 0 ok.
And therefore, v is 0 at some place and v will be 0 everywhere ok. So, we have gotten an
integral form, now what we are going to do is that we are going to assume that the velocity
profile in the boundary layer is quadratic. Now, if you ask why it is? There is no a priori
reason, but what we want good see is that it basically going to serve well. So, if you assume
that the velocity profile inside the boundary layer is quadratic, in other words you would
assume that u is given by U infinity times 2 y by delta minus y square by delta square.
2𝑦 𝑦 2
𝑢 = 𝑈∞ ( − 2 )
𝛿 𝛿
760
So, we are actually hypothesizing this form and this form one way of thinking it is that
yeah, it is quadratic in y ok. Now, this is because in pipe flow we know that it is quadratic
so, that is one of the reason. If you look at y is equal to 0, what is u? u is 0 ok, if you look
at it y is equal to delta what is u? U so, this profile essentially satisfies the boundary
condition. In the sense that this profile satisfies the fact that it is no slip at the so, on the
solid plate it actually gets a velocity y; it could have gone linearly, it could have gone
quadratically and so on.
In fact, one would be able to you know write down the full equation you will be able to
solve it and when you solve it you will find that the profile is very close to parabolic. In
reality the profile only v deviates a little bit from parabolic, but we do not want to go
through the entire calculation of doing it. So, we are basically taking a simpler approach
in a in which we are assuming that the velocity profile is parabolic.
761
Lecture - 67
Boundary Layer - Momentum Integral Analysis – 2
So we are looking at the boundary; sorry boundary layer flow and we started out by looking
at an integral approach ok. We basically looked at what does the mass conservation tell us,
what does the momentum conservation tell us. And that is we defined drag force and we
defined the drag force the force exerted by the plate on the fluid as an integral, so the first
integral that you see is what they wrote down is that the all right.
And then we said we do not know what the profile is going to be, but it is a reasonable
approximation that we you know use a parabolic profile for the velocity field. And if so
we can calculate what is the drag force on the party, on the plate or the force exerted by
the plate on the fluid. So, we said if you take that second expression U as a quadratic
expression that is something that satisfies the boundary condition, that is something that
varies quadratically with the with Re. So, we can use that expression and try to calculate
what is the drag force ok, so that is where we ended up, so we could continue.
We will first calculate this so called momentum thickness theta equal to integral 0 to delta
U, so we need U by U infinity. So, that is 2 y by delta minus y square by delta square times
762
1 minus U minus U by U del U infinity that is 1 minus 2 y by delta plus y square by delta
square d y.
Equal to you could go ahead multiply an integrate q y q by delta into y is y square by 2

minus 2 into 2 4 by delta square y square that is y cube by 3. Then plus 2 by delta cube y
cube that is y raised to 4 by 4 minus 1 by delta square y square that is y cube by 3 and so
on keep going all of them and the apply a limit and see what we get 2 by 15 delta.
𝛿 𝛿
2𝑦 𝑦 2 2𝑦 𝑦 2 2 𝑦3 4 𝑦3 2 𝑦4 2
𝜃 = ∫ [ − 2 ] [1 − + 2 ] 𝑑𝑦 = [ − 2 + 3 +⋯] = 𝛿
0 𝛿 𝛿 𝛿 𝛿 𝛿 3 𝛿 3 𝛿 4 0
15
So, that is theta, now we can calculate the wall shear stress tau w as mu into du by d y at
y is equal to 0. So, that is mu into d by d y of U that is U infinity into 2 y by delta minus y
square by delta square at y is equal to 0. So, what does is that give you ok, while we are
writing it down. Because we know tau w in terms of theta what is the expression what is
the connection between tau w and theta?
𝑑𝑢 𝑑 2𝑦 𝑦 2 2𝜇𝑈∞
𝜏𝑤 = 𝜇 |𝑦=0 = 𝜇 [𝑈∞ [ − 2 ]] |𝑦=0 =
𝑑𝑦 𝑑𝑦 𝛿 𝛿 𝛿
We derived tau w is equal to rho U square d theta by d x ok and we have got an tau w now
as this. That means, 2 mu infinity divide by delta is equal to rho mu infinity d by d x of 2
763
by 15 delta. So, one of them will go away this is equal to 2 by 15 rho U d delta by d x,
remember delta is a boundary layer thickness and it changes with x. Therefore, so we will
go we have 15 mu U infinity by rho U infinity d x is equal to d delta divided by delta 15
mu by rho infinity d x is equal to d delta by delta.
𝑑𝜃
𝜏𝑤 = 𝜌𝑈∞
𝑑𝑥
2𝜇𝑈∞ 𝑑 2 2 𝑑𝛿
= 𝜌𝑈∞ ( 𝛿) = 𝜌𝑈∞
𝛿 𝑑𝑥 15 15 𝑑𝑥
15𝜇
∫ 𝑑𝑥 = ∫ 𝛿 𝑑𝛿
𝜌𝑈∞
And integrating do not to get 15 mu by rho u infinity into x is equal to lon delta.
What is that, did I make a mistake?
It is a that way, so we can not be a log right, so it is delta d delta integral that is right. So,
that is delta square by 2 how could it that is because we have supposed to get a rule and
that is where I get delta equal to 30 rho U infinity x divided by mu square root. Now, there
is something else missing what is that I will missing.
I am ending within with a dimensional quantity equal to a non dimensional quantity.
764
We no fine ok, this x is down stress right, so I should be basically doing that when I move,
so on the f write basically you have a root of 30 times R e x ok. So, this is what we had
said when we started talking about boundary layer this thing that the thickness of the
boundary layer grows with x and for a laminar fluid laminar boundary layer it goes
approximately a square root of Reynolds x ok. So, if you assume that the profile is actually
parabolic you basically end up doing that or ended seeing that.
15𝜇 𝛿2
𝑥=
𝜌𝑈∞ 2
𝛿 30 𝜇 30
=√ =√
𝑥 𝜌𝑈∞ 𝑥 𝑅𝑒𝑥
So, where should I change is 30 by because I have in done mu delta by x is equal to x

square 13 by rho U infinity x this is right. So, therefore, I should really have root 30 divided
by root of R e x is this right. So, that we have delta going as x root 30 divided by x U
infinity rho by mu to the power of half and that means as x to the power of half.
So, the other thing is that you know actually the Reynolds as the Reynolds number
increases the boundary layer thickness should go down ok. So, it has to be a one by relation
that also it tells ok, the Reynolds number increases the thickness is there. So, that mean,
so see what we assume was this, this is what we assume for the laminar velocity profile.
So, if you could you know write down an expression for the turbulent velocity profile.
Then you would be able to see how does it actually change in a turbulent boundary layer
as well, what is this scaling in turbulent boundary layer? X to the power of 6 by 7 or
something, so you could a get now, the other thing is this stream line that we do the other
day. So, that stream line is getting displaced because the fluid below is getting retarded.
so that is a argument that we had ok. So, there is a, so how much it gets you know displaced
this distance ok. So, if comes at h and it reaches a delta, so that h minus del delta minus h
is called displacement thickness, so this thing is called displacement thickness.
765
So, homework for you, you see use this use these expressions and try to show that delta
star by x is equal to 1.83 by square root of R e x where delta star is the displacement
thickness. This is the equations that we have already written down and try to manipulate
you see whether, so you understand what is delta star right is just delta minus h. We know
how delta changes with R e x, so see whether you are able to do that, so useful expression
ok, any doubts.
𝛿∗ 1.83
=
𝑥 √𝑅𝑒𝑥
So, theta is called this is the momentum thickness, it is just a measure of the drag force but
express there is a length ok, so that is all it is. It is a in fact, if you look it, it just a introduced
in this calculation for convenience instead of dealing with that particular integral. Integral
0 to delta U by this thing you have just called a theta and if you look at it is basically d by
rho b U infinity square ok. So; that means, it is drag force, but non dimensionalized. So,
that quantity is called theta, but it has the dimensions of length now, so it is not completely
non dimensionlized which has the dimensions of length.
Delta minus h, so the displacement thickness.
It will change with.
766
H will not, h is the height in which a fluid element is coming and it is getting displace by
a distance of delta star ok. So, h, so what matters this how much it is dis, so delta minus h
is what matter. So, delta star will be a quantity which will be a function of x.
Because, it is going to match with a fine the real expressions ok, so one thing is that
whenever you have laminar flow confined between two things the it has to if the viscous
mechanism is important like in pipe flow we typically get something close at a parabolic
profile. So, here also if you actually had a complete derivation you get a profile which is
very close to parabolic. So, typically whenever you have flow in very thin sections is
reasonable the an approximation of a parabolic profile often gives you good results, so that
is the only reason.
So, we could have done the same calculation assuming linear and you would have gotten
a good estimate ok, parabolic is slightly better and a because the actual profile is very
closed to a parabolic profile. And if you want to do turbulent then you would not want to
do parabolic because you know that it is going to be a much flatter one.
So, it is basically is a better to assume something which is like you know going to the
power of n where n could be some number which gives you a much more flatter profile
ok. So, we understand a little bit about how to approach boundary layer and the thickness,
we will look at differential approach now.
767
Lecture - 68
Boundary Layer - Differential Approach
We will start down by writing down a Steady Incompressible 2D Navier Stokes Equations.
So, the x-momentum equation, so I will throw away the unsteady term u dou u by dou x
plus v dou u by dou y equal to minus dou p by dou x plus mu del square u by del x square
plus del square u by del y square dou u dou v by dou x plus v dou v by dou y equal to
minus del p by del y plus mu del square v by del x square plus del square v by del y square
ok.
𝜕𝑢 𝜕𝑢 𝜕𝑝 𝜕 2𝑢 𝜕 2𝑢
𝜌 [𝑢 +𝑣 ]=− +𝜇[ 2 + 2 ]
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑥 𝜕𝑦
𝜕𝑣 𝜕𝑣 𝜕𝑝 𝜕 2𝑣 𝜕 2𝑣
𝜌 [𝑢 +𝑣 ]=− +𝜇[ 2 + 2 ]
𝜕𝑥 𝜕𝑦 𝜕𝑦 𝜕𝑥 𝜕𝑦
So, this is these are equations that you should have really solved to get velocity profile
inside the boundary layer. I am going to make some assumptions which is basically on the
fact that if you go to higher and higher Reynolds number the boundary layer is suppose to
768
be very thin ok, its going to get smaller and smaller and smaller and also you know that
the largest velocity is in the u direction.
In other words the v velocity is going to be much smaller than u. So, I will say if my v is
much smaller than u and if I say my dou u by dou x is much smaller than dou u by dou y
because dou u y is the small number dou u by dou y is going to change from 0 to some
finite number in a small thickness, but dou u by dou x is what the change in u along x
which is changing very slowly ok.
So, therefore, this approximation of dou u dou x is smaller than dou u by dou y is good for
boundary layer also dou v by dou x is much smaller than dou v by dou by ok. So, these
three approximations are found to be good for a boundary layer if the boundary layer is
very thin. And if I use that and if I look at my y momentum equation I find that this quantity
as a dou v dou x in it this quantity as a v times dou v by dou y, this is also proportional to
v this is also you know proportional to v.
In other words this equation this left hand side or I am drawing it in this circle are going
to be extremely small quantities for a boundary layer. Now, you have to trust me there
because one has to actually do a formal way of doing it and sort of going to tell that is what
is going to hold for the boundary layer ok. And the reason I want to tell it is because if you
look at this entire equation many of these terms will come out to be so small that you
essentially ended with and with this expression that dou p by dou y is going to be 0 ok.
Or in other words pressure variation across the boundary layer is a small number or its
actually negligibly small. And so, in other words therefore, pressure variation is only in
the x direction ok, its not there in the y direction inside a boundary layer or to put it
physically if you have let us say we have a diverging channel and let us say you have a
fluid that is coming and a boundary layer is growing. So, this is boundary layer ok; so that
was our.
So, let us say we have two plates diverging and there is a fluid that is coming and let us
say boundary layer starts growing on the solid wall. Because you have fluid coming and
is you know going inside a diverging channel, what happens to the velocity? Velocity will
decrease right. So, that would mean pressure would increase.
769
So, you can really apply Bernoulli’s equation and let us say calculate pressure here and
pressure here you can use the same pressure here also because the pressure variation inside
a boundary layer is going to across the boundary layer is so small. The pressure variation
along the boundary layer is then determined by what is happening outside.
If pressure was changing with respect to y then you would have to worry about how does
the pressure change, but if you want to only determine pressure inside the boundary layer
you can actually calculate it from the numbers that is not outside. I have given a very sort
of a hand waving arguments if you do not believe it just do not worry it take for granted
that pressure variation across the boundary layer is going to be so small. Therefore you
can actually determine the pressure variation inside a boundary layer just by knowing the
pressure outside. This is.
In laminar meaning laminar boundary layer.
No, so, there we are basically saying that the pressure variation is only in a x direction
right.
No there we can actually show that right. So, there what happens is if you looked at v
velocity all of them will identically go to 0 and therefore, we will see the dou p by dou y
is 0 and therefore, p is not a function of y. Now, I know that its a v is not 0. So, none of
the terms are actually 0, but they are all small.
So, in other words the flow is going to be very similar to what we will have in your pipe
flow ok, v component is very small that will basically end up with a similar situation where
p with y variation is going to be small. So, the situation is similar; so you can see the
analogy already that its because of the reason that v is very small that you ending up with
a such a conclusion. Questions?
So, that would mean that this quantity dou p by dou x is simply dt by dx.
770
So, I will write down my x momentum equation again; my x momentum equation is going
to be rho times u dou u by dou x plus v dou u by dou y is equal to minus dp by dx plus mu
del square u by del x square sorry del x square u by del y square.
𝜕𝑢 𝜕𝑢 𝜕𝑣 𝜕𝑣
𝑣 ≪ 𝑢; ≪ ; ≪
𝜕𝑥 𝜕𝑦 𝜕𝑥 𝜕𝑦
𝜕𝑢 𝜕𝑢 𝑑𝑝 𝜕 2𝑢
𝜌 [𝑢 +𝑣 ]=− +𝜇 2
𝜕𝑥 𝜕𝑦 𝑑𝑥 𝜕𝑦
So, I am not writing del square u by del x square, but its del square u by del x square is
also expected very small and the only reason I want to write it down is because I want to
write this equation on the surface of a or on the wall I want to simplify this on the wall.
So, what is u on the wall? 0. What is v on the wall? 0. So, u is 0, v is 0 and therefore, if I
actually look at my x momentum equations these things go away on the surface of the wall
and therefore, I actually have mu v square u by d y square is equal to dp by dx on the
surface ok.
𝑑𝑝 𝜕 2𝑢
0=− +𝜇 2
𝑑𝑥 𝜕𝑦
All of you agree pressure is decreasing with x. So, what is the situation that I am going to
look at? I am looking at this kind of a diverging channel no I do not want to do that I want
771
to start with the other one let us look at converging channel. So, velocity decreases pressure
increases; so dp by dx is greater than 0 ok.
So, let us say we have a converging channel in the converging channel velocity decreases
that would mean pressure increases. So, if I have a boundary layer.
So, let us say you know you have a kind of a converging channel there is a fluid that is
coming and going. So, if there is a growth of boundary layer like that I am basically looking
at you know a point somewhere there and trying to write down what is it what is the
772
differential form telling me. So, I have taken a situation where dp by dx is greater than 0,
you have question.
So, let us say you have. So, far we have been talking about a flat plate and a fluid that is
coming right. Now, and we said of course, the outside the boundary layer the fluid is
moving with a constant velocity. Now, I’m expanding that case to which the outside fluid
is also either accelerating or decelerating.
So, the u infinity that I have been having I am think I am look at it as an accelerating fluid
or a decelerating fluid. So, it will accelerate or decelerate I mean to accelerate to decelerate
I am trying to put in a converging or a diverging channel. So, I have a uniform flow that
comes and then it is seeing a converging section. So, it’s basically going to enter the
converging section while it is enter entering the converging section it’s basically going to
see it is going to develop a boundary layer as if it’s like to going to have an entrance length
ok. So, its going to its going to develop a boundary layer and I want to I also said look if I
look at the pressure gradient outside that is going to be same as the pressure gradient inside
also ok.
The dp by dx inside the boundary layer is same as dp by dx outside the boundary layer.
So, if I simply apply my Bernoulli’s equation which is the fact that the velocity is
increasing I would see that my pressure is decreasing if it’s a converging otherwise its
increasing in a diverging channel ok. So, just by looking at whether u infinity is increasing
or decreasing I am able to tell what is going to happen for the pressure inside the boundary
layer and by looking at that I am actually going in to now predict what is going to be the
flow profile exactly on the wall.
So, what am I am going to first look at cases where it is a converging channel. So, u in is
smaller than u out. So, that would mean p in is greater than p out and therefore, dp by dx
should be less than 0, correct in a converging channel sorry we made mistake dp by dx is
less than 0 is the first situation that we are look at which also would mean du by dx.
So, u is increasing with sorry u e u is increasing with x; so d u by dx is u is increasing with

x du by dx is greater than 0 ok. So, we said on the surface this is the expression that should
hold ok. So, if I have a situation in which dp by dx is less than 0 my d square u by dy
square should also be negative. So, if my d square dy square is negative can we see, can
we tell the what the velocity profile should be? Should it be like that, should it be like this,
773
should it be like that, three different type is should it be like this should it be like that or
should it be like this.
Remember this expression this expression that we have written down is valid only at the
wall. So, I can only tell how it should be at the wall, but by looking at this where I have
dp by dx less than 0; that means, d square u by dy square less than 0 whether this green or
blue or this magenta will satisfy d square u by dy square less than 0. Green well blue cant
because blue means its a straight line. So, the second derivative should go to 0. So, blue is
ruled out now green or magenta.
How do you know that?
So, let us say d square u by d y square I will write using finite difference of approximation
du by dy at y plus delta y minus du by dy at y divided by delta y right in the limit of delta
y going to 0, I want d square u by d by square to be less than 0 so; that means, du by dy at
y plus delta y should be smaller than du by dy at y.
So, du by dy is the slope and that slope should be largest at the wall; so that satisfied by
the magenta line right. Remember u is on the x axis y is on the y axis ok. So, when I say
du by dy you should really be looking at the slope like that ok. So, its highest like this; so
its a magenta line that is going to be satisfying this situation d square u by dy square less
than 0 ok.
So, I can immediately say that therefore, the green cannot be and it should be something
like the magenta ok. So, this is what the boundary velocity profile should be if I say dp by
dx is less than 0 ok. So, what should happen is that the profile would and that is what we
had we have drawn and that is what we had analyzed also. So, its just we going to be
something of that is sort.
Yes.
Cannot be a du by dx is you are arguing du by dx is not small I am saying that du by dx is

going to be smaller than du by dy u changes from 0 to capital u in a small distance of y,
while u changing very slowly along the x direction.
So, it has to be a sort of a slowly converging I it cannot be like that it has to be something
where this approximation holds. So, that is because we cannot solve actual equation. So,
774
we are saying can be actually simplify it is in some way. So, that we can get some
qualitative arguments and then which might true hold in general equations and that is what
we are trying to aim at. So, this situation is called a favorable pressure gradient, we will
see why its favorable later let us call that favorable pressure gradient.
So, the next one we want to analyze is dp by dx is equal to 0. So, that is case 1, case 2, dp
by dx is equal to 0 what would be the profile? We know dp by dx is given by. So, d square
u by d y square the curvature is 0, so the velocity at the wall should be linear ok. So, it
should be should start linear and but then we do not know what is going to happen
something of that sort that is going to be if dp by dx is equal to 0.
And if dp by dx is equal to 0 what would you say about d u by d x the outside velocity d u
infinity if you like du infinity by dx is also 0 ok. So, outside fluid is know not changing its
velocity in any fashion pressure is also not changing.
So, see here if you look at this would be linear, but then we do not know because this
equation holds true only on the surface right this equation is not completely general we
put we got it when we put u is equal to 0 v is equal to 0. In the conservation equation in
the full we put u equal to 0 v equal to 0 threw away these terms and therefore, we go out
this. So, that can true hold only on the surface of the wall.
Yes.
775
This is only for the thin boundary layer yes if the. So, because we are saying that the this
dou dp by dx right you can say that its only a function of outside when the boundary layer
is so thin. So, that we can neglect you know all x all x gradient and so on and typically
boundary layers are thin like you know it may be like a millimeter or smaller things like
that.
On the surface.
On the surface y is equal to 0 yes.
On the surface y is equal to 0 u and v are also 0 right.
Yes, so, let me just tell you again. So, what; so see by looking at the v momentum equation
whether you believed me or not what I said is that pressure across the boundary layer,
pressure in y direction is not changing ok; that means, pressure in x direction is the only
thing that is changing. So, if I have a boundary layer and outside the boundary layer if I
know how pressure is changing I will know how pressure changes inside the boundary
layer also because pressure does not change in the y direction. Is that clear ok?
You have a boundary layer let us say this is the plate you have a boundary layer, the
boundary layer is very thin and saying that pressure variation across the boundary layer is
very small; that means, if I know pressure gradient outside I will know pressure variation
inside also. If it is one atmosphere 1.5 atmosphere, it will be 1 atmosphere and 1.5
atmosphere inside also because I know pressure is not changing across y ok.
Knowing a pressure gradient outside will help me to tell what is pressure gradient inside
also in the boundary layer that is a first thing that I said. Second thing that I have said is if
I look at the x momentum equation and if I write down the x momentum equation on the
wall; on the wall then I know because of no slip and no penetration my u component of
velocity is 0 on the wall; my v component of velocity is 0 on the wall and therefore, I am
able to simplify my governing equation in to this magenta circle ok, where I have only a
viscous term and a pressure term ok.
Now, remember that pressure term dp by dx I say I already know it from the external flow
field when I say external this external to the boundary layer. So, by knowing what is the
pressure gradient outside I can tell something about my velocity profile on the wall that is
776
where I am doing really ok. By knowing the pressure gradient outside I am able to tell
something about by velocity gradient on the wall is that clear. Does that answer your
question, what is your question?
Turbulent.
Yes it is only on the surface. So, remember I am drawing it only on the surface ok. So,
when I said you know d square u by dy square is less than 0, I only know it at this region
when I am saying d square u by dy square is 0 I only know about this the other part I just
connected and drew I do not know what it is and I am not going to care about that also at
the moment.
This is wall. The other is just if you draw the profile I am just drawing it can 0 distance.
Because I have pressure decreasing, so velocity increasing.
So, this is only up to boundary layer whatever I have drawn is up to. So, delta is going to
be much larger than that. So, let us say ok. So, this distance up to which I have drawn is
really inside the boundary layer its not even going to connect outside. See I cannot say
even anything because I am really looking at wall what is the; what is the governing
equation at the wall and making some deductions.
It will be actually continuous because it’s increasing and it will go on then join there and
outside then.
Outside again there could be some changes which will be along x direction not along y
direction.
This one at y is equal to 0. You have some doubts no ok. So, then if I have a case where
dp by dx is less than 0, in fact all three cases are going to be dp by dx greater than 0 ok.
What do we know about d square u by d y square at the wall? Should be greater than 0
correct. So; that means, my curvature cannot be like this my curvature should change the
sign.
So, initially it was initially it was like this then I said its going to be like this now it has to
be like that ok. So, in other words my velocity profile should start doing that ok, the
777
original one of those color green or magenta that we had wrote it down. So, like that and
then it would do that sign d square u by d by square should be greater than 0.
So, write it down again using your finite difference and remember u is your x axis, y is
your y axis ok. So, its typically opposite to the slope that we talk about, so.
Correct sorry that is right ok. So, this is what the velocity profile going to be if I have. So,
when would this happen? This would happen when dp by dx is greater than 0; that means,
pressure increases along the x direction that would happen when u decreases along the x
direction, u will decrease along the x direction and it is a diverging section ok.
So, if you have a diverging section then you should expect the velocity profile to curve in
a different way is that ok. So, if dp by dx is even larger, what would you expect this
curvature would be increase at the wall it would be almost become flat and do that. If dp
by dx is much larger it would even bulge a more and it would actually do that inside the
boundary layer and drawing my u velocity. What does the last figure tell? That on the wall
the flow is actually in the opposite direction, if dp by dx is very large.
So, this is you know dp by dx is small let us say this is some moderate and this is some
large ok. So, if outside fluid outside the boundary layer if pressure increases and if that
pressure increases very large on the wall the fluid will actually start flowing in the opposite
direction that is what we actually called flow separation.
So, whenever the talked about you know those diverging sections and so on I always drew
fluid flowing in the opposite direction right. So, this because of this particular reason this
is called actually boundary layer separation ok. The external pressure gradient will cause
a fluid flow to you know flow in the opposite direction on the wall.
778
Lecture - 69
Laminar and Turbulent Boundary Layer
So, if you look at the facts that we talked about, one of the things that we talked about was
that, so if you have you know let us say boundary layer and this is x-direction that is y-
direction, then I said pressure gradient across the boundary layer. So, across the boundary
layer meaning in y-direction is negligible compared to pressure gradient along the
boundary layer.
So, this is one of the things that we said in the last class. And therefore I argued that if
therefore if we know pressure gradient which is you know dp by dx outside the boundary
layer, then we know dp by dx inside the boundary layer as well ok. So, knowing the
pressure gradient outside the boundary layer essentially helps us to determine what is the
pressure gradient inside the boundary layer, so that is one of the things that we learned.
779
And then I also argued that on the wall the governing equation essentially simplifies to d
square u by d dou square u by dou y square is equal to dp by dx so on the wall. And
therefore knowing the pressure gradient outside the boundary layer helps us to determine,
how does the you know second derivative of velocity change on the wall ok. What is the
second derivative of velocity on the wall and that basically helps helped us to plot the
velocity profiles.
So, just to go through it once again I am interested in calculating the second derivative of
velocity. So, dou square u by dou y square at y is equal to 0, I can write it as dou u by dou
y at let us say delta y minus dou u by dou y at y is equal to 0 by delta y in the limit delta y
tending to 0 is my definition of second derivative of velocity. And if I say dou square u by
dou y square is greater than 0 that automatically implies that dou u by dou y at y is equal
to delta y should be larger than dou u by dou y at y is equal to 0. So, that would mean that
my velocity profile it is, I have to say that if it is at 0 the slope is smaller above the slope
is larger. So, it has to be that or this.
Student: First.
First one, I did not do the different colors. So, remember your that is here u that is your y
ok. So, if dou u by dou y is large, basically this is this it has to be of that form right. So, it
has to be that.
780
And if dou square u by dou y square is less than 0 that would imply that dou u by dou y at
y is equal to delta y should be smaller than dou u by dou y at y is equal to 0 and that would
happen if my profile actually has a opposite curvature. And therefore, we talked about few
different situations.
I talked about favorable pressure gradient.
It is velocity profile no. So, how do you draw velocity profile? That is your u, that is your
r or y or anything.
781
Did I do it wrong. So, let us look at the first case we have dou u by dou y should be correct
dou u by. So, this is my y is equal to 0. Here my slope is smaller, there my slope is larger,
because if there is a small change in y, u changes not here correct, this is ok. So, we have
a favorable pressure gradient. Second was 0 pressure gradient; third one was weak adverse
pressure gradient; fourth critical. So, I did not put all the names I think yesterday pressure
gradient by excessive adverse pressure gradient.
So, favorable pressure gradient has to be in the direction in which dp by dx is less than 0;
this is dp by dx is 0; this is dp by dx is greater than 0. This is dp by dx greater than 0; this
is dp by dx greater than 0. And therefore, d u infinity by dx should be greater than 0; d u
infinity by dx is 0; d u infinity by dx is less than 0; d u infinity by dx is less than 0; d u
infinity by dx is less than 0.
So, in the first case, where u infinity is increasing with x that is like a contraction, this is
just uniform flow. This is like an expansion; this also like an expansion and this also like
an expansion that the amount of divergence should be larger and larger as you go to the
right. And therefore, we said the flow profile here should be like that, here till increase her
like a straight line. And then smoothly join with U infinity, here you should have a
difference different curvature.
So, it has to be like, like that or you could actually have it almost like that and you could
have it like that. So, in a contraction flow sticks to the wall in an expansion on the surface,
you could have flow going the other way. So, this was the story that we had at in the last
class. Any questions on that again? So, what I am going to do today is, I am just going to
show you how do you see these things in the experiments.
782
So, what they are going to do is that you know they are basically going to have look at
both contracting, expanding as well as uniform flow and to see the flow they have actually
put hydrogen bubbles into the fluid flow ok. So, this hydrogen bubbles are extremely
small, they just get carried ok, they just go passively with the flow, but you can see it
whatever you going to see white are actually concentrated our hydrogen bubbles ok. So,
that is the way they visualize it.
So, first let us look at. So, this is; so, this thick band that you are seeing that is a whole lot
of hydrogen bubbles that has been inserted and there is a fluid flow from left to the right.
So, you see that the band is moving ok. And as the band moves, they will put a white solid
plate. So, the fluid will go above the plate and then you then the moment the fluid is going
on the plate you should see the formation of the boundary layer. So, that is the idea, so no,
so there.
783
So, that is the solid plate there you go.
So, you see the this basically the fluid near the plate is getting accelerated and then you
start seeing this growth of boundary layer. You will see a few more times and you can see
as boundary layer the thickness grows as you go downstream ok.
784
Now, what they do is, you have this, they will actually move the camera along with the
band ok. So, then you can see much more downstream, you can see that now, maybe they
are short, they are.
So, they are moving along with that and you can see the growth in the boundary layer now
right ok.
785
So, now what they have done is, they are actually going to take a diverging channel. So,
we remember we said in a diverging channel is where you basically have a you can expect
to see a separation, so that is what they want to demonstrate. So, you can see that this
channel is slightly diverging ok. So, slightly diverging is going to generate a big adverse
pressure gradient.
So, there will not be any separation, but as you will see that as divergence increases, you
will see the fluid flow separating. So, here this is going to be without. So, you can see the
boundary layer here ok. So, this black region is basically the representation of boundary
layer, the boundary layer grows, but nothing special happens.
786
Now, taking a bigger diverge more diverging channel.
So, they are in so this is a band of you know hydrogen bubbles that has been inserted. So,
here this is like you know more or less uniform, then you can see a bit further this
downstream and this is a bit more downstream ok.
787
And you can see that here because of because this is diverging, the pressure the velocity is
decreasing that would mean the pressure is increasing. And we know that the pressure
gradient outside is going to determine what is going to happen on the wall and the fluid is
actually flowing in the opposite direction.
So, I said in the boundary layer ok, so as soon as the boundary layer starts, it will be a
laminar boundary layer. And beyond a point, it will transition into turbulent boundary layer
ok. You can also see that. So, in this particular case, we are not going to do it on a wall.
788
So, there they actually have a pipe ok, and they and there is a fluid that is coming from the
left side and it is flowing on top of the pipe ok. So, it is a round thing. And therefore, a
boundary layer grows on top of the pipe and there they have actually characterized. So, on
the leftmost, you have a laminar boundary layer. And then what you see is some kind of a
waves and so on that essentially represents transition and here much further downstream
it basically turns into turbulent boundary layer.
So, this is what the boundary layer would look like. If you I mean if you basically put lots
of you know let us say something that you can visualize you will see how smooth the
laminar region is and how things basically turn to turbulent as the boundary layer
undergoes transition. So, this is the laminar turbulent transition in a boundary layer.
789
Now, something that I did not tell yet ok, we really did not distinguish laminar boundary
layer versus turbulent boundary layer. So, and we talked also about the separation in the
boundary layer. So, what happens is that the turbulent boundary layer, so turbulent
boundary layer would mean larger Reynolds number right and larger and Reynolds number
is basically a representation of larger inertia.
So, if they are in turbulent boundary layer, therefore, separation occurs at a later point not
as easily as in a laminar boundary layer. So, in laminar boundary layer, the fluid might you
know get separated, so this region that I am talking about ok. While if you are looking at
a turbulent flow, the separation is expected is you know there is less chance for separation
because it is more inertial, therefore, it will have a tendency to actually stick to the wall
and go ok, so that is called a delayed separation. So, in turbulent boundary layer use you
are less likely to see the separation than in laminar boundary layer if you were to do a
comparison. So, this can also be seen and that is what is going to be demonstrated.
790
Note this there. So, here you have a same diverging channel. And I forgot on one of the
side they will have. So, what they do is that, so it is the same channel and then the fluid is
coming on the top it is just smooth pipe; at the bottom they will introduced a little bit of
roughness ok. They basically will put a wire ok. And then what happens is that the flow
will basically be trip to turbulence because you see some you know disturbance. So, it was
trying to maintain the laminar flow, but you have given some disturbance. So, the fluid
flow will turn turbulent. So, this is a nice case in which the top, you will have laminar
turbulent layer; at the bottom, you will have a turbulent boundary layer ok.
791
So, this is laminar to turbulent, I mean in top and bottom, it is different. And you can see
that turbulent boundary layer sort of sticks to the wall much further downstream than the
laminar boundary layer ok. So, the top you can see separation the fluid flow is going the
other way so basically this already so see the white is coming from the left. So, if it was
occupying the full region that would mean that it was a unidirectional kind of flow ok, but
it is not. All this region it is actually occupied by the black, that means, that the fluid flow
is basically coming in a different way.
So, this black region should be thought of as region that are separated from the main stream
right ok. So, if you were to see what is there in the black you will see the flow in the
opposite direction as we saw earlier. While in turbulent you can see that the fluid flow is
almost sticking to the wall much further downstream ok. Therefore, turbulent boundary
layer sticks to the wall much more than the laminar boundary layer.
Yes, you mean the video or the what I said? So, what I am trying to say is the turbulent
boundary layer turbulence basically corresponds to larger inertia. So, the fluid that is going
to come basically have the tendency to maintain that velocity ok. So, instead of having an
opposite velocity, it would have a tendency to go with the same velocity ok. So, therefore,
turbulent boundary layer the chances of separation is less compared to laminar boundary
layer. So, this will we can do this comparison if you have the same Reynolds number and
you can still get laminar and turbulent.
So, that is the case here. You have the same Reynolds number the fluid flow can actually
be laminar or turbulent. If the fluid flow is turbulent, then the fluid flow will not that will
not separate from the wall, it will have it will go with you know in the same direction as it
was coming well the laminar might separate. So, upside you have laminar flow, it is
separating, but downside you see it is turbulent, so it is trying to stick to the wall black
color. So, everywhere it is fluid, the white region indicates the fluid that is coming from
the left.
So, wherever you are seeing white, you should imagine that is the fluid actually that is
incoming right. So, the black region is the one then it has not come from there. So, it has
come from somewhere else. So, it must represent regions either stagnant or it flow it fluid
flow from the other side.
792
No, that is the region of the boundary layer now or actually that is the region in which they
know. So, this is the wall; this is the wall. See this is the wall and this is the wall these two
white are walls it is a diverging channel. So, 99 percentage greater than 99 percentage
would be this region. This region will be boundary layer; this region will also be boundary
layer ok.
Now, this region you are not seeing any white fluid means that the fluid here has not come
from the left side, either it was just stuck there or it would have come from the right side
right. See here whatever is entering is coming as a white fluid, so if it was all going it
should have occupied the entire channel.
So, fluid has a velocity smaller than 99 percent is both up there and down there right ok.
So, now, what I am saying is that look here the fluid has already reached white color. So,
the fluid has not reached here, the white fluid has not reach here. You expect that the white
fluid will come and occupy the entire region.
If boundary layer was not there. So, here there is boundary layer. So, the fluid is the white
fluid is going to come at a slower, oh, I see what you are saying. So, it is going to come at
a slower velocity. Now, look here, so this region it has already reached white right, but
here it has not come white ok, so that means, that the fluid even though it was a slower
region this region has already turned white and therefore, we would expect this region also
to turn white.
No, so, yes, this is fluctuating velocity field, but the fluid that the main velocity is still in
the x-direction. So, the fact that you have gotten the white fluid here would mean that the
fluid has reached here, the incoming fluid. Similarly, you would have expected the
incoming fluid to reach here also, but it has not ok. In fact, it has not because if you look
at this, this entire region there is a black color. So, in this entire region, it is because the
fluid flow is separated. So, look here. So, this entire region is still occupied by the white,
so that is all incoming fluid. But this region is not incoming fluid ok, so that means, the
laminar flow has already separated, well turbulent flow has not separated yet.
793
They will draw a little bit of velocity profiles ok. So, you can see how here also they have
done the same thing. You have a plate above they have made it turbulent; below it is
laminar ok. So, laminar you can see that a boundary layer and it will be all fluctuating
velocity field. We will draw more. They show you a few snapshots of various profiles. So,
this is the effect of the fluctuations in the velocity field in a turbulent flow ok, but on an
average it would look something. So, they will plot an average also is like images put on
top of each other.
794
So, they are going to put all the images together of the turbulent, then you can see that
there is a nice average ok. So, up as the turbulent boundary layer, bottom is the laminar
boundary layer. And you can also see that in the turbulent boundary layer the profile is
much flatter than in the laminar boundary layer. So, that is what we had even argued in the
case of a pipe flow, till draw a mean profile ok.
This is not really anything to do with the boundary layer, but I have been telling you about
the you know the fluid flow getting separated and so on. So, this last portion actually they
show it very nicely in this region, you can see how you know this vortices form other
corners ok.
795
Yes, there is there are two corners up there up there I mean up there and down there and
you can see how near the corners the fluid flow is separated and then form this vortices
and so on. You can go back and check the video at least the images.
796
Lecture - 70
Tutorial - 09
So, welcome to the Tutorial section of Fluid Mechanics and myself Chaitanya.
So, the problem that we are going to solve today is this water at room temperature flows
at the same volumetric flow rate, cube equals to 9.4 into 10 power minus 4 meter cube per
second through two ducts. One is the round pipe, other is the annular pipe which you can
see in the figure and all walls are made of commercial speed and both the sort of same
length and the dimensions are given. So, our objective is to calculate the head loss in the
two pipes and comment on the efficiency of the two pipes. So, let us see how to solve this.
797
So, the geometry is one is a round one other is the annular pipe. So, if you look at the
dimensions R is 15 mm a is 25 mm. So, R is 15 mm, a is 25. So, the first step is to calculate
the unknown which is b. So, the information that is given is the cross sectional area of two
ducts is equal. So, therefore, the cross sectional area of round pipe is equal to cross
sectional area of the annular type.
So, this gives us pi R square is equal to pi into a square minus b square. So, we know R
we know a, which is 25 mm and this is 15 mm and therefore, we can calculate b which is
going to be 20 mm. So, therefore, we know all the dimensions in the problem, now we
need to calculate the head loss in the circular pipe and the annular pipe.
798
So, let us first calculate the head loss in the circular pipe. So, to do that first we need to
calculate the Reynolds number and for that we need the velocity. In the problem the flow
rate is given as 9.4 into 10 power minus 4 meter cube per second. And, we know the cross
sectional area which is pi R square from this we can calculate V as Q by A which gives us
1.33 meter per second.
So, with this information we can calculate the Reynolds number which is b rho by mu. So,
this is going to be 39700 and this is therefore, the flow is turbulent in this problem.
Therefore, one needs to use the moody chart to calculate the friction factor and use that to
calculate the head loss.
799
So, for the commercial steel as they have mentioned the pipe is made of commercial steel
and the epsilon by the value is 0.00153. So, one can use this epsilon by the value and the
Reynolds number, using both of these one can calculate the frictional coefficient from
Moody chart which is 0.0261. And, from this one can calculate the head loss using h f by
L is nothing, but f by d into V square by 2 g.
So, we have the information of f d and we know velocity and we know gravitational
constant 9.81 meter per second. So, substituting all this we can get h f by L as 0.0785. So,
therefore, for the circular pipe the head loss is 0.0785.
800
Yeah it is. So, it is meter per second square sorry. So, therefore, we have calculated the
head loss in a circular pipe, we will repeat the calculations for the annular pipe, but in the
case of an annular pipe we need to use the hydraulic diameter.
So, calculations for annular pipe; so, in the case of an annular pipe we need to use the
hydraulic diameter which is defined as 4 times of area by vector parameter. So, for this
geometry which is annular pipe this is a and this is what we have to b. The area is 5 times
a square minus b square and this is going to be pi times a plus b 2 pi R and for annular is
2 pi R a plus b.
801
So, this gives us 2 into a minus b. So, this is the hydraulic diameter and so therefore, we
calculate the Reynolds number based on this hydraulic diameter which is velocity D h by
nu. So, as we have; as it is given in the question, the cross sectional areas of both circular
and annular pipe are same; therefore the velocity fluid velocity in both the pipes will be
the same because the volumetric product remains same. And, we have we can calculate
the hydraulic diameter and we know the kinetic viscosity; substituting all this we will get
the Reynolds number as 26500.
Therefore, even in annular pipe the flow is turbulent and here it is going to be 0.023 epsilon
by D h value from which we can calculate f moody which is 0.0291. So, now we have the
factor from which we will be able to calculate the head loss in the same way as we the for
a circular pipe. So, by now it is going to be an approximate value because we are using the
hydraulic diameter concept. So, this will be around 0.131.
802
So, if I look at the compare the values of head loss for a circular pipe it is around 0.0785
and whereas, for annulus it is going to be around 0.131. Therefore, it is clear that in the
case of an annular pipe the head loss is more and it make sense because in the case of an
annular pipe, the fluid is in contact with the more wall area. Because, the fluid is in contact
with the inner wall as well as the outer wall therefore, there will be high friction. So, we
can say efficiency of circular pipe is greater than efficiency of the annular pipe. So, this is
question a.
803
Question b. So, in question b it was asked to assume; in question b it was asked to assume
in an annular duct, the flow in an annular duct is laminar and suggest the value of a and b
that gives a head loss same as that of a circular pipe. So, here what we do is that we have
an expression for head loss 2 g and we were asked that the flow assume that the flow in a
annular duct is laminar. Therefore, we can directly calculate f because we have a relation
between f and Reynolds number in the case of a laminar flow.
And, we need to suggest the values of a and b which is the radius of inner pipe and the
outer pipe that gives the head loss same as that of a circular pipe. So, for the circular pipe
we have obtained the head loss as 0.0785. And, in the case of laminar flow f is 64 by
Reynolds number which is defined based on the hydraulic diameter and we have one more
D H here and we have V square by 2 g.
So, if I substitute everything 64 D H velocity by kinematics viscosity into D H to V square

by 2 g. So, this V and this V got cancel off and we end up with an equation. So, if I
substitute V as Q by A because I know volumetric flow rate, but not the velocity and I
know area is unknown because we do not know what is a comma b in this problem.
Therefore, I will end up with an equation 64 Q by 2 g to nu by 4 pi into 1 by a square
minus b square into a minus b whole square 0.0785.
So, volumetric flow rate is given in the problem, we know the kinematic viscosity and we
know the gravitational constant and everything except a and b. So, I will take the unknowns
804
to one side and I will end up with, if I substitute the corresponding values I will get to 10
to the power minus 9. So, this is the equation that governs the values of a and b.
And, if we assume if you take the value of a equals to 25 which is given in the schematic,
the value of b that we get is around 21 mm. So, one can get if you want to have the
efficiency of an annular pipe same as that of a circular pipe then what one we need to have
a very thin annular ring. So, you can just see this is I get 21 mm and this is 25 mm. So, this
should be the annular parameter that one should use to get the head loss same as that of a
circular pipe. So, therefore, this is about this question.
805
Now, let us move on to then second question. So, in this question we use the concept of
displacement thickness to calculate the pressure gradient and the mean velocity. So, the
question goes like this air at 20 degree centigrade and 1 atmosphere enters a 40 cm square
duct as shown in the figure. And, using the displacement thickness concept we are asked
to estimate the mean velocity, the mean pressure in the core of the flow at the position x is
equal to 3 meters and what is the average gradient in Pascal per meter in this section.
806
So, let us try to answer this question. So, this is question number 2. So, as they have given
air at temperature 20 degree centigrade, the parameters can be obtained from the
corresponding tables where density is going to be 1.2 kg per meter cube and the viscosity
is 1.8 into 10 power minus 5 kg per meter second. So, we know the air properties, now we
were asked to calculate the velocity at the exit. So, if you look at the question so, we are
asked to calculate the mean velocity which means the velocity at the exit.
To calculate that we need to impose the continuity equation we will get to that, but first let
us calculate the Reynolds number which is defined for this problem as rho U x by mu and
we have an inlet velocity and we know density which is. So, density is 1.2, velocity is 2
meter per second which is given and the distance is 3 and viscosity is 1.8 into 10 power
minus 5. Substituting all this we will get the Reynolds number as 4 into 10 power 5 which
means the flow is laminar, as it is a plate problem.
Now, using the displacement thickness concept as we know the Reynolds number and we
know the distance which is x equal to 3 meters, we will be able to calculate the
displacement thickness which is delta star. So, there are two formulas to calculate delta
star; one is based on the exact calculation which is 1.721 x by R e x power 1 by 2. So, this
is based on the exact calculation and if we assume the profile in the boundary layer is
parabolic then what we get is something very close to that which is 1.83 x by R e x power
1 by 2. So, the relative error between these two will be 6 per around 6 percent.
807
So, now we will use the exact equation maybe you can try using the equation that we have
obtained assuming the flow is parabolic. So, we know x which is 3 meters and we know
Reynolds number so, that gives us the delta star as 0.0082 meters so, it is a rounded
decimal. So, now we need to impose the continuity equation, let us look at the geometry
again. We have a fluid entering so, even before that let us look at what is displacement
thickness and how we can use it to understand this how to impose the continuity equation.
So, suppose you have a flat plate and the fluid is entering and we know that there will be
a boundary layer development and displacement thickness is the measure to quantify the
amount by which the outermost streamline gets displaced. So, this is what delta. So, if so,
the streamline which is just outside the boundary layer is getting displaced because of the
boundary layer affects. So, this displacement is quantified by displacement thickness. So,
if you want impose the continuity equation so, if one wants to impose the continuity
equation. So, this is fluid which is entering or the mass flux rho u into A.
Now, here the area is changed. So, here the area is based on this height whereas, here area
is based on let say this is h so, this is delta star. So, we need to say that rho u h is equal to
rho u h plus delta star. So, if we assume it is a square duct it is like h square into h plus
delta square. So, this is the case for a flat plate.
Now, if we use the same concept for the given geometry which is a channel, what we will
end up with this and they have given the dimensions of the channel; this is a 2D channel
808
and they have given it is around 40 centimeters. So, here what we get is V into L naught
square equals to V exit which is an unknown into L naught minus 2 delta star whole square.
So, you can see so, this is the inlet velocity, we know inlet velocity, we know the
dimensions of the channel therefore, we can calculate the inlet mass flux which is rho u
into s. And, similarly we know the displacement thickness and using that we can write the
out mass; this is the mass flux out and this is the mass flux in. So, we know the entrance
velocity which is 2 meter per second, L naught is given in the problem which is 40
centimeters. So, V exit is an unknown, L naught and we have calculated delta star.
So, from this we can calculate V exit as 2.175 meter per second. So, therefore, we have
calculated the a part which is the mean velocity and then we were asked to calculate the
mean pressure in the core of the following at the position x equal to 3 meters. So, to
calculate the pressure because we have a velocity data and we have an inlet pressure so,
we can impose the Bernoulli equation. So, which is P exit plus rho V square by 2 rho V
exit square equals to P inlet or I will define as P 0 plus rho by 2 V 0 square. So, as the inlet
pressure is atmospheric pressure let us work with the gas pressure which means we are
subtracting atmospheric pressure from the given pressures.
Therefore, as inner pressure is already 1 atmosphere, this goes to 0 because we are looking
at the gas pressure. So, what we can get is the P exit which is rho by 2 times V inlet square
minus V exit square. So, this will be around 0.44 Pascal’s and you should remember that
809
this V exit corresponds to x equals to 3 meters. Therefore, this P exit corresponds to x
equal to 3 meters. So, therefore, we have calculated the pressure and the next question is
what is the average pressure gradient in Pascal per meter.
So, the pressure gradient is nothing, but delta P by x, as we are working with the gauge
pressures we can write delta P as 0.44 and x in the problem is 3 which gives us around 1.5
Pascal per meter. So therefore, using the displacement thickness concept we have
calculated the mean velocity and using that mean velocity and the Bernoulli’s expression,
we have calculated the pressure and from that we have calculated the pressure gradient.
So, in this way we can use the displacement thickness concept in the case of a channel and
calculated the exit velocity and the pressure gradient. So, this ends the tutorial.
Thank you.
810
THIS BOOK IS
NOT FOR SALE
NOR COMMERCIAL USE
(044) 2257 5905/08 nptel.ac.in swayam.gov.in

103106158

Uploaded by

Document Information

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

103106158

Uploaded by

Copyright:

Available Formats

INDEX

S. No Topic Page No.

(Refer Slide Time: 00:22)

(Refer Slide Time: 06:22)

(Refer Slide Time: 00:25)

So, I can write the net force:

So, that is the definition of the gradient operator. So, we get:

(Refer Slide Time: 09:43)

Therefore, P is only a function of z. This the equation of hydrostatic pressure.

Now, if you look:

So, I will use that as:

(Refer Slide Time: 00:31)

And from Newton’s law of viscosity:

(Refer Slide Time: 07:03)

𝐵𝑜𝑡𝑡𝑜𝑚 𝑙𝑎𝑦𝑒𝑟 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦, 𝑢 = 0

𝑇𝑜𝑝 𝑙𝑎𝑦𝑒𝑟 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦, 𝑢 = 𝛿𝑢

(Refer Slide Time: 11:19)

So, I can calculate tan δθ as:

(Refer Slide Time: 00:25)

(Refer Slide Time: 02:46)

⃗ (𝑥, 𝑦, 𝑧, 𝑡) = 𝑢(𝑥, 𝑦, 𝑧, 𝑡)𝑥̂ + 𝑣(𝑥, 𝑦, 𝑧, 𝑡)𝑦̂ + 𝑤(𝑥, 𝑦, 𝑧, 𝑡)𝑧̂

Now, when I have written V as a function of x, y, z and t; I am actually defining my

So, acceleration is going to be:

(Refer Slide Time: 08:01)

(Refer Slide Time: 10:22)

⃗ = 𝑢𝑥̂ + 𝑣𝑦̂ + 𝑤𝑧̂

(Refer Slide Time: 12:42)

Or combining all of them; I can write the acceleration vector:

FACE IN OUT DIFFERENCE

(Refer Slide Time: 26:00)

So, then we can calculate the rate of change of density:

𝑅𝑎𝑡𝑒 𝑜𝑓 𝑐ℎ𝑎𝑛𝑔𝑒 𝑜𝑓 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑖𝑛 𝑡ℎ𝑒 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑎𝑙 𝑣𝑜𝑙𝑢𝑚𝑒 𝑑𝑥 𝑑𝑦 𝑑𝑧 = 𝑚𝑖𝑛

We can write this equation in a much more convenient form:

⃗ = 𝜌𝑢𝑥̂ + 𝜌𝑣𝑦̂ + 𝜌𝑤𝑧̂

Now at this point I will talk about an incompressible fluid.

(Refer Slide Time: 20:20)

So, similarly we can do it for each face. So, let us do that.

FACE MOMENTUM MOMENTUM FLUX DIFFERENCE

The rate of change of momentum.

(Refer Slide Time: 11:47)

𝜕𝜌 𝜕(𝜌𝑢) 𝜕(𝜌𝑣) 𝜕(𝜌𝑤) ⃗

(Refer Slide Time: 19:13)

(Refer Slide Time: 20:33)

(Refer Slide Time: 00:13)

𝜎𝑥𝑥 𝜎𝑦𝑥 𝜎𝑧𝑥 −𝑃 + 𝜏𝑥𝑥 𝜏𝑦𝑥 𝜏𝑧𝑥

(Refer Slide Time: 06:54)

(Refer Slide Time: 12:31)

𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑢 𝜕𝑃 𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥

𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑣 𝜕𝑃 𝜕𝜏𝑥𝑦 𝜕𝜏𝑦𝑦 𝜕𝜏𝑧𝑦

𝜕𝑤 𝜕𝑤 𝜕𝑤 𝜕𝑤 𝜕𝑃 𝜕𝜏𝑥𝑧 𝜕𝜏𝑦𝑧 𝜕𝜏𝑧𝑧

(Refer Slide Time: 19:12)

And this can be further simplified.

𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥 𝜕 𝜕𝑢 𝜕 𝜕𝑢 𝜕𝑣 𝜕 𝜕𝑢 𝜕𝑤

𝜕𝜏𝑥𝑥 𝜕𝜏𝑦𝑥 𝜕𝜏𝑧𝑥 𝜕 2𝑢 𝜕 2𝑢 𝜕 2𝑢

And the simplified equations would be

(Refer Slide Time: 26:11)

Now, similarly for the y and z momentum equation

(Refer Slide Time: 02:03)

𝑂𝑢𝑡: (𝑃 + 𝑑𝑃), (𝜌 + 𝑑𝜌), (𝑉 + 𝑑𝑉), (𝐴 + 𝑑𝐴)

𝐿𝑖𝑛𝑒𝑎𝑟 𝑚𝑜𝑚𝑒𝑛𝑡𝑢𝑚 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛: ∑ 𝐹 + 𝐼𝑛 − 𝑂𝑢𝑡 = 𝑅𝑎𝑡𝑒 𝑜𝑓 𝑐ℎ𝑎𝑛𝑔𝑒

(Refer Slide Time: 12:53)

(Refer Slide Time: 17:25)

𝑑(𝜌𝐴𝑉 2 ) = 𝑑(𝜌𝐴𝑉)𝑉 + 𝜌𝐴𝑉𝑑𝑉

If you look at the continuity equation: