DIMENSIONAL ANALYSIS

AND
SIMILARITY

CHAPTER-5.
(FRANK WHITE)
 WHAT IS DIMENSIONAL ANALYSIS
Three Techniques For Solving Fluid Dynamics Problems
1. Analytic Fluid Dynamics (AFD)
2. Experimental Fluid Dynamics (EFD)
3. Computational Fluid Dynamics (CFD)

EXPERIMENTAL FLUID DYNAMICS (EFD)

In Experiments we study,
 Effect of one or more variables on other variables
 Example, 1 2
S  So  vot  gt 
 S  f S , v , t, g
o o 
2
Determine Experimentally the effect “ So, vo, t, g ” ON ‘S ’
Keep “So, vo, g ” constant, Do experiments by varying ‘t’, get ‘S ’
Suppose you do 10 experiments by varying time
Keep “vo, t, g ” constant and for each of the above ‘t’, Do another 10
experiments by varying ‘So’, get ‘S ’, and so on
 WHAT IS DIMENSIONAL ANALYSIS, contd.

Q.1 How many experiments you will perform?

Ans. 10000 OR 104.
Q.2 If one experiment costs Rs. 100. Find total cost?
Ans. Rs. 106.

IS THIS COST REASONABLE

Q.3 Now do experiments by varying 10 S/So for each vot/So
(for 10 vot/So) keeping  = gSo/vo2 constant. How many
experiments will be done?
Ans. 1000 and cost will be Rs 100000.
However, with dimensional analysis, we can immediately
reduce this information to an equivalent form
 WHAT IS DIMENSIONAL ANALYSIS, contd.

S  f  So , vo , t , g 
Two different
mathematical functions
but contain all the same S  vo t gt 2 
information =k , 
So  So So 

Nothing is lost in a dimensional analysis

Q.4 Will 103 experiments of S/So ,vo t /So &  = gSo/vo2

give the information of 104 experiments?
Ans. Yes.
 WHAT IS DIMENSIONAL ANALYSIS, contd.
Dimensional Analysis is the Planning, Presentation,
and Interpretation of Experimental Data
Dimensional Analysis, a Process of Formulating
Problems in Terms of Dimensionless Variables and
Parameters
BENEFITS OF DIMENSIONAL ANALYSIS
1. It helps our thinking & planning for an experiment or theory
2. Reduces the number of experiments & hence cost by reducing the
number of variables either by discarding them or by grouping
them in dimensionless groups
3. Useful in understanding the Physics of problems
4. Dimensional Analysis gives scaling laws
Data on cheap large/small scale models  large / small scale
expensive prototypes
WHAT IS DIMENSIONAL ANALYSIS, contd.
BUT BE VERY CAREFUL
1. Learning dimensional analysis is like learning to play
a game. There are levels of the game
2. One can establish some ground rules and do some
fairly good practice. But the level of expertise comes
with lots of practice
3. Similarly dimensional analysis in the broad view has
many subtleties and nuances which only time, practice
and maturity enable you to master
4. Although dimensional analysis has a firm physical and
mathematical foundation, considerable art and skill are
needed to use it effectively
PRINCIPLE OF DIMENSIONAL HOMOGENETY
Any equation describing a physical situation will only be true if
both sides have the same dimensions.
i.e, it must be Dimensionally homogenous.

EXAMPLE 1: Displacement of freely falling body

Governing equation,

S  So  Vot  12 gt 2

A Unit Balance gives:

m 1 m 2
m  m s 2 2 s m  m
s s
PRINCIPLE OF DIMENSIONAL HOMOGENETY
A Dimensional Balance gives:

        2       L    L 
L  L   LS 1
 S  1 
LS 2
  S 2

EXAMPLE 2: Flow over a rectangular weir 2

Q  B 2g H 2
3

Governing equation, 3

Writing the equations in terms of dimensions

 
1
3
1 2
3
 L LT L  L3T 1
2
LT 2

NOTE HOW THE DIMENSIONS

AND THEIR POWERS ARE EQUAL
PRINCIPLE OF DIMENSIONAL HOMOGENETY
IMPORTANT INFORMATION.
Every Dimensional Homogenous Equation has;
Dimensional Variables: Quantities which actually varies
during a given case. e.g. S, t
Dimensional Constants: May vary from case to case, but
constant for a given case. e.g. So, Vo, g
Pure constants: Which remain constant in every case
e.g. ½ etc

Mathematical Operations Such as Differentiation

or Integration May Change the Dimensions but
the Dimensional Homogeneity Does Not Change
PRINCIPLE OF DIMENSIONAL HOMOGENETY
Dimensionally homogeneous eqs. can be written in a more
compact entirely equivalent non-dimensional form

There is more than one method of presenting

one’s dimensionless data or theory.
 RESULTS OF DIMENSIONAL ANALYSIS
Performing dimensional analysis on a physical problem gives;
1. A single mathematical equation
2. Relates all the physical factors involved with each other
3. Rearrangement of variables into dimensionless groups
Example: Determine the force on a propeller blade:
Solution:
1. First determine the factors which influence this force.
2. This information comes from experience & observations
Force ‘F’ depends on the following parameters
a. Diameter of the propeller blade,
b. Forward velocity of propeller (velocity of plane), u
c. Fluid density, 
d. Revolutions per second, N
e. Fluid viscosity, 
 RESULTS OF DIMENSIONAL ANALYSIS
Hence the functional form will be
F    d , u,  , N ,  
OR 0  1  F , d , u ,  , N ,  
Where  and 1 are unknown functions

Functional form can be written as F  K d mu n  o N p  q

Where K is some constant & m, n, o, p, q are constant powers

Using Dimensional Analysis, ONE

1. obtains these powers (m, n, o, p, q)
2. forms the variables into several dimensionless
groups
Whereas K or  or 1 are to be found experimentally
 BUCKINGHAM’S  THEOREMS
2 Theorems accredited to Buckingham, known as  theorems
1st  theorem:
A relationship between m variables (physical properties
such as velocity, density etc.) can be expressed
GENERALLY as a relationship between m-n non-
dimensional groups of variables (called  groups), where n
is the number of governing or repeating variables.

If a physical problem can be expressed:

 ( Q1 , Q2 , Q3 ,………, Qm ) = 0
Then, as per the above theorem, this can be expressed as
 (1 , 2 , 3 ,………, m-n ) = 0
In fluids, we normally take n = 3
 BUCKINGHAM’S  THEOREMS, contd.

2nd  theorem
Each  group is a function of n governing or repeating
variables plus one of the remaining variables.

Repeating Variables:
Those which we think
1. Will appear in all or most of the  groups,
2. Are of major influence in the problem

BEFORE DOING ANALYSIS OF A PROBLEM

ONE MUST CHOOSE THE REPEATING
VARIABLES
 BUCKINGHAM’S  THEOREMS, contd.
How To Choose A Repeating Variables
There is considerable choice
1. From 2nd theorem there can be n ( = 3) repeating variables
2. When combined, these repeating variables must contain all
of dimensions (M, L, T). (That is not to say that each must
contain M, L and T only)
3. A combination of the repeating variables must not form a
dimensionless group
4. Repeating variables do not have to appear in all  groups
5. The repeating variables chosen should be measurable in an
experimental investigation.
e.g, pipe diameter (dimension L) is more useful and
measurable than roughness height (also dimension L)
6. They should be of major interest to the designer
 BUCKINGHAM’S  THEOREMS, contd.
An Example:
Shear stress, τw depends on various parameters, such as
Dimensional 8 w 
 w  F (  ,V ,  , d ,  )  f  F ( Red , )
Analysis gives V 2
d

Let’s solve his case:

0   ( w ,  , V ,  , d ,  ) 1st  theorem; m=6 , n=3
therefore  groups = m – n = 6 – 3 = 3
Repeating Variables: , d, V 0   ( 1 ,  2 ,  3 )
(sort of rule of thumb in fluid dynamics)

2nd  theorem; 3 ’s

1   d V  w
a1 b1 c1
 2   a2 d b2V c2   3   a3 d b3V c3 
 BUCKINGHAM’S  THEOREMS, contd.

     
a1 c1
 L
3 b1 1
1 in dimensions M L T  ML
0 0 0
LT ML1T 2

For dimension Mass M 0  a1  1

For dimension Length L 0  3a1  b1  c1  1

For dimension Time T 0  c1  2

 a1  1, b1  0, c1  2

 w 
Hence; 1   d V  w  
1 0 2
2 
 V 

Similarly repeat the procedure for 2

 BUCKINGHAM’S  THEOREMS, contd.

   L   LT   L 
a2 b2 c2
2 in dimensions M L T  ML
0 0 0 3 1

For dimension Mass M 0  a2

For dimension Length L 0  3a2  b2  c2  1

For dimension Time T 0  c2

 a2  0, b2  1, c2  0

 
Hence; 2   d V    
0 1 0

d 

Similarly repeat the procedure for 3

 BUCKINGHAM’S  THEOREMS, contd.

     
a3 c3
 L
b3
3 in dimensions M L T  ML
0 0 0 3
LT 1
ML1T 1

For dimension Mass M 0  a3  1

For dimension Length L 0  3a3  b3  c3  1

0  c3  1 Correlatio
For dimension Time T
 a3  1, b3  1, c3  1

  
Hence; 3   d V   
1 1

1
HENCE;
 Vd 

 w    w   Vd 
0    1 ,  2 ,  3   , ,  OR  f  1  , 
 V 2
d Vd  V 2
 d  
 Another Example
Force ‘F’ On A Propeller Problem
0   ( F ,  ,V ,  , d , N )

1st  theorem; m=6 , n=3 0   ( 1 ,  2 ,  3 )

therefore  groups = m – n = 6 – 3 = 3

Repeating Variables: , d, V
(sort of rule of thumb in fluid dynamics)

2nd  theorem; 3 ’s

1   d V F
a1 b1 c1
 2   a2 d b2V c2 N  3   a3 d b3V c3 
 Another Example, contd.
     
a1 c1
 L
3 b1 1
1 in dimensions M L T  ML
0 0 0
LT MLT 2

For dimension Mass M 0  a1  1

For dimension Length L 0  3a1  b1  c1  1

For dimension Time T 0  c1  2

 a1  1, b1  2, c1  2

 F 
Hence; 1   d V F   2 2 
1 2 2

 d V 

Similarly repeat the procedure for 2

 Another Example, contd.

     
a2 c2
 L
3 b2 1
2 in dimensions M L T  ML
0 0 0
LT T 1

For dimension Mass M 0  a2

For dimension Length L 0  3a2  b2  c2

For dimension Time T 0  c2  1

 a2  0, b2  1, c2   1

 Nd 
Hence;  2   0 d 1V 1 N   
 V 

Similarly repeat the procedure for 3

 Another Example, contd.
   L   LT   ML T 
a3 b3 c3
3 1 1 1
3 in dimensions M L T  ML
0 0 0

For dimension Mass M 0  a3  1

For dimension Length L 0  3a3  b3  c3  1

For dimension Time T 0  c3  1

 a3  1, b3  1, c3  1

  
Hence; 3   d V   
1 1

1
HENCE;
 Vd 
 F Nd   F  Nd  
0    1,  2 ,  3   , ,  OR  1  , 
  d 2 2
V V Vd   d 2 2
V  V Vd 
 BUCKINGHAM’S  THEOREMS, contd.

WRONG CHOICE OF PHYSICAL PROPERTIES

You choose extra but unimportant variable

• You end up with an extra , which plays little role on
physical problem
• Extra  is found by experiment

You ignore important influential variable

• You end up with one less ,
• Experimental analysis will miss significant behavioral
changes
 Manipulations Of Dimensionless Groups, ’s
Once  groups identified then
1. Manipulation of the  groups is permitted
2. Manipulations do not change the behaviour
3. May change their appearance drastically
If the defining equation is
(1 , 2 , 3 ,………, m-n ) = 0
 Manipulations Of Dimensionless Groups, ’s
1. Any number of groups can be combined by multiplication
or division to form a new group which replaces one of
the existing. e.g. 1 & 2 may be combined to form
1a = 1/2
2. Reciprocal of any dimensionless group is valid. 1a=1/1

3. Any dimensionless group may be raised to any power is

valid. 1a = (1)2
4. Any dimensionless group may be multiplied by a constant.
1a = constant  1
5. Any group may be expressed as a function of the other
groups, 1 = (2)
In General The Defining Equation Could Look Like
(1 , 1/2 , (3) i ,………,0.5 m-n ) = 0
 𝑈
 
√𝐺𝐿

SOME
COMMON
DIMENSIONLESS
NUMBERS
 SIMILARITY

Assume that for a particular problem you have.

1. Identified the variables involved
2. Performed dimensional analysis and listed the  groups

Now you do experiments to determines the ’s

Based on the relationship, you try to build your dream machine

Q. What should be the shape and size of the machine?

Q. Would it be wise to build it to the actual size and test?

Q. If yes then, what would happen if machine fails or does
not perform as expected?
Q. If no then, would you prefer to build a model of that
machine, smaller in size but exactly same shape?
 SIMILARITY, contd.

A small / large scaled model would be better, why

1. Cheaper
2. Can be modified, rebuild again and again
3. Easier to do experiments

Would the data of small / large scale model fit the actual size
1. If yes, then how would you ensure this

Suppose the data of model fits the prototype.

1. Functional dependence of variables on model and prototype
would be identical

MODEL PROTYPE
1m =  (2m , 3m , …) IDENTICAL 1p =  (2p , 3p , …)
 SIMILARITY, contd.

True only if,

1m = 1p

Experience tells that this happens if there exists between model and
prototype, similarities;
GEOMETRIC
KINEMATIC
DYNAMIC

GEOMETRIC SIMILARITY

MODELS ARE OF TWO TYPES

TRUE AND DISTORTED

HAVE EXACTLY THE SAME PHYSICAL

TRUE MODELS
FEATURES BUT TO A SCALE
 GEOMETRIC SIMILARITY, contd.

DISTORTED DO NOT HAVE EXACTLY THE SAME

MODEL PHYSICAL FEATURES AT ANY SCALE

GEOMETRIC SIMILARITY EXISTS BETWEEN MODEL AND

PROTOTYPE IF THE RATIO OF ALL CORRESPONDING
DIMENSIONS IN THE MODEL AND PROTOTYPE ARE EQUAL.
OR
IF ALL BODY DIMENSIONS IN ALL THREE COORDINATES
HAVE THE SAME LINEAR-SCALE RATIO.

Lmodel Lm
  L  Scale factor for length
Lprototype L p

Amodel A
 m   A  Scale factor for area
Aprototype Ap
 GEOMETRIC SIMILARITY, contd.

1. All angles are preserved

2. All flow directions are preserved

3. The orientations of model and prototype w.r.t surroundings must

be identical.

Example of model of 1/10 scale

 GEOMETRIC SIMILARITY, contd.

Example of
model of 1/10
scale

1. The model nose radius must be one-tenth as large

2. The model surface roughness must be one-tenth as large
3. If the prototype has a 5-mm boundary-layer trip wire 1.5 m from
the leading edge, the model should have a 0.5-mm trip wire 0.15 m
from its leading edge
4. If the prototype is constructed with protruding fasteners, the model
should have homologous protruding fasteners one-tenth as large
 GEOMETRIC SIMILARITY, contd.
1. Departure from these details is a violation of geometric similarity

a. Must be justified by experimental comparison to show that

the prototype behavior was not significantly affected by the
discrepancy

2. Models appearing similar in shape but clearly violating geometric

similarity should be compared at your own risk.
 KINEMATIC SIMILARITY

Similarity of time as well as geometry.

It exists between model and prototype
1. If the paths of moving particles are geometrically similar
2. If the ratio of the length scale are same
3. If the ratio of the time scale are same
Consequence (velocity scale is also same)
4. The kinematic dimensionless groups must be same
Kinematic similarity exists between model and prototype if,
HOMOLOGOUS PARTICLES LIE AT HOMOLOGOUS POINTS
AT HOMOLOGOUS TIMES
AN EXAMPLE

FREE SURFACE FLOWS

 KINEMATIC SIMILARITY, contd.
Vm Lm TM L
Velocity    V
Vp Lp Tp T
am Lm T 2 M L
Acceleration   2  a
a p Lp T p  T
2

Qm L3 m TM  3 L
Discharge  3   Q
Qp L p Tp T

Example: Low Speed Free surface flow

1 1

Vm  Lm    Lp 
2 2

  
     
V p  Lp   Lp 
Vm2 V p2
Frm    Frp
gLm gLp Tm  Lm Vm    
      

Tp  Lp V p    
 KINEMATIC SIMILARITY, contd.
Fluid Properties Viscosity, Surface Tension, Compressibility etc
are same then
KINEMATIC SIMILARITY DEPENDS ON DYNAMIC SIMILARITY

Dynamic Similarity:
Reynolds and/or
Mach Number and/or
Froude Number
etc are involved
 DYNAMIC SIMILARITY
Similarity of time, geometry as well as forces
It exists between model and prototype
1. If the ratio of the length scale are same (first requirement)
2. If the ratio of the time scale are same
3. If the ratio of the force-scale or mass-scale are same

Kinematic and Dynamic similarities exists simultaneously if;

1. Both model and prototype have identical all types of force
coefficients
How this is ensured for compressible flow for the prototype
and model. If
1. Reynolds numbers are equal
2. Mach number are equal
3. Specific Heat ratio are equal
 DYNAMIC SIMILARITY
How this is ensured for incompressible flow for the prototype
and model. If
1.With No Free Surfaces
1. Reynolds numbers are equal,
2.With Free Surfaces
1. Reynolds, Froude, Weber & cavitation number are equal
Examples

DISCREPANCIES WITH AIR AND WATER TESTING

PERFECT DYNAMIC SIMILARITY OF MODEL & PROTOTYPE

REALITY OR DREAM
A TEST CASE: Hydraulic model testing with free surface

For Dynamic Similarity, there should be

Equivalent Froude numbers and Equivalent Reynolds
 DYNAMIC SIMILARITY, contd.

For Equivalent Froude numbers between Model and Prototype

Lmodel Lm
  L  Scale factor for length
Lprototype Lp

Vm Lm V p L p Vm2 V p2
Re m    Re p Frm    Frp
m p gLm gL p

 m Vm Lm
1 1

Vm  Lm    Lp 
2 2

  2
3

 p V p Lp       
V p  L p   Lp 

Implication of this result:

Say  = 0.1  3/2 = 0.032

model=(1/30)prototype
 DYNAMIC SIMILARITY, contd.

If Prototype is to work with water then this kinematic viscosity

ratio is almost impossible to achieve, what to do now ?
Now you have to choose which of the dimensionless group is
more important. Neglect the other and then scale your
experiment. In this case
1. Neglect Re Number.
2. Equate Froude Number
Vm2 V p2
Frm    Frp
gLm gL p
 EXAMPLE-1
The valve coefficient K =∆p/(V2) for a 60 cm diameter valve are
to be determined from tests on a geometrically similar 30-cm
diameter valve using air at 80oF. The ranges of tests should be
for flow of water at 70oF at 1 to 2.5 m/s. What ranges of airflow is
needed?
Solution This needs dynamic similarity: Hence Rem=Rep

VD   1m / s   0.6m  
 Remin  water     610, 000

  5 2

 1.056 10 ft s  0.3048m / ft  
2

VD
 Remax  water   2.5  610, 000  1,525, 000


For testing with air at 80oF, air = 1.67210-5 m2/s

Vmin Dm  Vmin  0.3m  

 Remin  air  610000    Vmin  30.6m / s
 air
 
 1.627 10 m s 
5

2

 EXAMPLE-1, contd.
Vmax Dm  Vmax  0.3m  
 Remax  air  1525000    Vmax  85m / s
 air
 
 1.627  10 m s 
5 2
 
Similarly

 Qmin  air   0.3m   30.6m / s   2.16 m3
2
s
4


 Qmax  air   0.3m   85m / s   6.0 m3
2
s
4
 Example-2
An underwater missile, diameter 2m and length 10m is tested in
a water tunnel to determine the forces acting on the real
prototype. A 1/20th scale model is to be used. If the maximum
allowable speed of the prototype missile is 10 m/s, what should
be the speed of the water in the tunnel to achieve dynamic
similarity?
Solution This needs dynamic similarity: Hence Re =Re m p

 Vm  m Dm   Vp  p Dp   p Dp m 1
      Vm  V p  10  1  1  200m / s
 m   p   m Dm  p 1 20

Very High velocity for water,

How can this be reduced ??
 Example-3
A model aeroplane is built at 1/10 scale and is to be tested in a
wind tunnel operating at a pressure of 20 times atmospheric.
The aeroplane will fly at 500 km/h. At what speed should the
wind tunnel operate to give dynamic similarity between the
model and prototype? If the drag measure on the model is
337.5 N. What will be the drag on the plane?
Solution For dynamic similarity: Hence Rem=Rep

 Vm  m Dm   V p  p D p   p Dp m 1 1
      Vm  V p  500    1  250km / h
 m    p   m Dm  p 20 1 10

pm  m RT  m 20
     m  20  p
p p  p RT  p 1

Relationship for resistance of a body moving through a fluid

The resistance, R, depends on the following physical properties:
: ML-3 V: LT-1 l: (length) L m: ML-1T-1
 Example-3, contd.
 R  V D 
Using Dimensional analysis:  2 2 
   
 V L    

 Rm   Rp 
Thus for dynamic similarity  2 
 2 
 Rem  Re p
 V 2
L   V 2
L 
 m m m   p p p 

2 2
Rm   mVm 2 Lm 2   20  p   0.5V p   0.1L p 
        0.05
R p   pV p 2 L p 2    p   Vp   Lp 

So The Drag Force On The Prototype Will Be

Rp 337.5
Rp    6750 N
0.05 0.05