DIMENSIONAL ANALYSIS

Fluid flow is influenced by several parameters like, the geometry, fluid

properties ,fluid velocity etc. In our earlier studies analytical methods used
in fluid flow studies were looked at in details.
In the study of flow of real fluids analytical methods alone are found
insufficient. Experimental methods and results have contributed heavily for
the development of fluid mechanics.
The solution of realistic problems usually involves both analytical and
experimental studies.
Experiments are used to validate analytical results as well as generalize and
extend their applications. Depending either solely on analytical methods or
experiments for the design of systems is found to lead to inadequate
performance and high cost.
Experimental work is rather costly and time consuming, particularly when
more than three parameters are involved. Hence it is necessary to plan the
experiments so that most information is obtained from fewest experiments.
Dimensional analysis is found to be a very useful tool in achieving this
objective. The mathematical method of dimensional analysis comes
to our help in this situation. The number of parameters can be reduced
generally to three by grouping relevant variables to form dimensionless
parameters. In addition these groups facilitate the presentation of the results
of the experiments effectively and also to generalize the results so that
these can be applied to similar situations.
Flow through pipes can be considered as an example. Viscosity, density,
flow velocity and diameter are found to influence the flow. If the effect of
each of these parameters on flow is separately studied the number of
experiments will be large.

Also these results cannot be generalized and its usefulness will be limited.
When the number of these variables are combined to form a dimensionless
group like (u D /) few experiments will be sufficient to obtain useful
information. This parameter can be varied by varying one of the variables
which will be the easier one to vary, for example velocity u. The results will
be applicable for various combinations of these parameters and so the
results can be generalized and extended to new situations. The
results will be applicable also for different fluids and different diameters
provided the value of the group remains the same. Lets look at a situation
The drag force F on a stationary sphere in flow is found to depend on
diameter D, velocity u, fluid density and viscosity . Assuming that to
study the influence of a parameter, 10 experimental points are necessary, to
estimate the total experimental points needed to obtain complete
information. We need to see how this could be reduced.
To obtain a curve of F vs u, for fixed values of , and D, experiments
needed are estimated to be 10.
To study the effect of these 10 experiments should be repeated 10 times
with 10 values of the total now being 102.
The 102 experiments have to repeated 10 times each for different values of .
Total experiments for u, and = 103.
To study the effect of variation of diameter all the experiments have to be
repeated 10 times each.
Hence total experiments required = 104.
We can however combine these parameters to obtain two dimensionless
parameters. These parameters could be

uD
F

u 2D 2

Now only 10 experiments are needed to obtain a comprehensive information

about the effect of these five parameters.

uD
F

u 2D 2

Experiments can be conducted for obtaining this information by varying the
parameter (uD/)and determining the values for F/u2D2.
Note : It will be almost impossible to find fluids with 10 different densities
and 10 different viscosities.
This illustrates the advantage dimensional analysis has in experiment
planning. The use of the results of dimensional analysis is the basis for
similitude and modeling studies.

METHODS OF DETERMINATION OF DIMENSIONLESS GROUPS

1. Intuitive method: This method relies on basic understanding of the
phenomenon and then identifying competing quantities like types of forces
or lengths etc. and obtaining ratios of similar quantities.
Some examples are: Viscous force vs inertia force, viscous force vs gravity
forces or roughness dimension vs diameter. This is a difficult exercise and
considerable experience is required in this case.
2. Rayleigh method: A functional power relation is assumed between the
parameters and then the values of indices are solved for to obtain the
grouping. For example in the problem shown above we can write

1, 2 F a b D c d u e
The values of a, b, c, d, and e are obtained by comparing the dimensions on
both sides the dimensions on the L.H.S. being zero as terms are
dimensionless. This is also tedious and considerable expertise is needed to
form these groups as the number of unknowns will be more than the number
of available equations. This method is also called indicial method.

3. Buckingham Pi Theorem method: The application of this theorem

provides a fairly easy method to identify dimensionless parameters
(numbers). However identification of the influencing parameters is the job of
an expert rather than that of a novice. This method is illustrated extensively
in fluid mechanics.

8.3 BUCKINGHAM PI THEOREM

The statement of the theorem is as follows : If a relation among n
parameters exists in the form

f ( q 1 , q 2 , q 3 ,.....) 0
then the n parameters can be grouped into n m independent
dimensionless ratios or parameters, expressed in the form

g(1, 2,3,..........nm ) 0
and 1 g1(2,3,.......... nm )
where m is the number of dimensions required to specify the dimensions of
all the parameters,q1, q2, .... qn. It is also possible to form new dimensionless
parameters as a discrete function of the (n m) parameters. For example
if there are four dimensionless parameters 1, 2, 3and 4 it is possible
to obtain 5, 6 etc. as

1
10.5
5
or 6 0.667
3 4
2
The limitation of this exercise is that the exact functional relationship
cannot be obtained from the analysis. The functional relationship is
generally arrived at through the use of experimental results

Determination of Groups
Irrespective of the method used the following steps will systematize the
procedure.
Step 1. List all the parameters that influence the phenomenon concerned.
This has to be very carefully done. If some parameters are left out, terms
may be formed but experiments then will indicate these as inadequate to
describe the phenomenon. If unsure the parameter can be added. Later
experiments will show that the term with the doubtful parameters as
useful or otherwise. Hence a careful choice of the parameters will help in
solving the problem with least effort.
Usually three type of parameters may be identified in fluid flow namely fluid
properties, geometry and flow parameters like velocity and pressure.
Step 2. Select a set of primary dimensions, (mass, length and time), (force,
length and time), (mass, length, time and temperature) are some of the sets
used popularly.
The primary dimensions are specified as
Mass = M in kg
Length = L in meters
Time = T in sec
Temperature = in C or K
Step 3. List the dimensions of all parameters in terms of the chosen set of
primary dimensions.
Force = kg m/sec2 =ML/T2
Area = m2 = L2
Volume = m3 = L3
Volume Flow rate = m3/sec = L3/T
Mass flow rate = kg/sec = M/T
Velocity = m/sec = L/T
Angular Velocity = Radians/sec = 1/T
5

Pressure = N/m2=( ML/T2 )/L2 = M/(LT2)

Moment = N(m) = ( ML/T2 )L =ML2/T2
Work, Energy(J) = N(m) = ML2/T2
Power (J/sec) = ML2/T3
Density = kg/m3 = M/L3
Dynamic Viscosity = (kg/m sec) = M/LT
Kinematic Viscosity = ( m2/sec ) = L2/T
Specific Heat = (J/kg K) = (ML2/T2)/M = L2/(T2)
We will develop such relations using primary dimensions to signify all
parameters.
Step 4. Select from the list of parameters a set of repeating parameters equal
to the number of primary dimensions. Some guidelines are necessary for the
choice.
(i) the chosen set should contain all the dimensions
(ii) two parameters with same dimensions should not be chosen. say L, L2,
L3,
(iii) the dependent parameter to be determined should not be chosen.
Step 5. Set up a dimensional equation with the repeating set and one of the
remaining parameters, in turn to obtain n m such equations, to determine
terms numbering n m.
The form of the equation is,
1 = qm+1 . q1 a . q2 b . q3 c ..... qmd
As the LHS term is dimensionless, an equation for each dimension in terms
of a, b, c, d can be obtained. The solution of these set of equations will give
the values of a, b, c and d. Thus the term will be defined.
Step 6. Check whether terms obtained are dimensionless. This step is
essential before proceeding with experiments to determine the functional
relationship between the terms.
6

Lets see how we do this

The force on a propeller blade depends on the following parameters;
c. Density in kg/m3

a. Diameter d in m.

b. Forward velocity u in m/s.

d. RPM N in 1/s.

e. Fluid Viscosity in kg/m- s while F is in kg-m/s2

Here m = 6 and n = 3 so we can get 3 groups.

We choose ,u, d as the repeating variables . They are measureable , and in
combination contain all dimensions of M,L and T.
Lets designate our Groups;
1 =a1ub1dc1F

2 =a2ub2dc2N

2 =a3ub3dc3

All these groups are dimensionless we now we use dimensional

homogeneity to find the groups.

For 1 a1ub1dc1F
a1

b1

c1 ML
M L
M LT 3 L 2
L T
T
0 0

we have

0 0
M0LT
M1a1 L3a1b1c11 T b12 or

For M

0 1 a1 which gives a1 1,

For T

0 b1 2 gives b1 2

For L

0 3a1 b1 c1 1 gives c 2

F
and so 1 2 2
u d

Groups;
1 =a1ub1dc1F

Fo r

2 =a2ub2dc2N

2 =a3ub3dc3

2 a 2 u b2 d c 2 N
0

w e h a ve

M LT

M 0 L0T 0 M a1

M
3
L

3 a 2 b2 c 2

a2

L
T

b2 1

b2

c2 1
L

T

or

Fo r M

0 a2

Fo r T

0 b 2 1 g ive s b 2 1

Fo r L

0 3 a 2 b 2 c 2 g ive s c 2 1

a n d so 2
1 =a1ub1dc1F

Nd
u
2 =a2ub2dc2N

2 =a3ub3dc3

For 3 a3 u b3 d c3
a

3
M

L
M 0 L0T 0 3
L T

we have

b3

c3

LT

M 0L0T 0 M a3 1 L3 a3 b3 c3 1 T b3 1 or
For M

0 a3 1

For T

0 b3 1 gives b3 1

For L

0 3a3 b3 c3 1 gives c3 1

and so 3

gives a3 1

ud

Thus we can study the behavior of the three non-dimensional groups to

solve our problems.
We can resolve our difficulties by relating them as below and then find an
empirical relationship

Nd
F

f
,

2 2
u d
u

ud

Choice of repeating variables

Repeating variables are those which we think will appear in all or most of the
pi groups, and are a influence in the problem. Before commencing analysis
of a problem one must choose the repeating variables. There is considerable
freedom allowed in the choice.
Some rules which should be followed are
i. From the theorem there can be only n ( = 3) repeating variables.
ii. When combined, these repeating variables variable must contain all of
dimensions (M, L, T) (That is not to say that each must contain M,L and T).
iii. A combination of the repeating variables must not form a dimensionless
group.
iv. The repeating variables do not have to appear in all p groups.
v. The repeating variables should be chosen to be measurable in an
experimental investigation. They should be of major interest to the designer.
For example, pipe diameter (dimension L) is more useful and measurable
than roughness height (also dimension L).
Wrong choice of physical properties.
If, when defining the problem, extra - unimportant - variables are introduced
then extra p groups will be formed. They will play very little role influencing
the physical behavior of the problem concerned and should be identified
during experimental work.

If an important / influential variable was missed then a pi group would be

missing. Experimental analysis based on these results may miss significant
behavioral changes. It is therefore, very important that the initial choice of
variables is carried out with great care.
Manipulation of Groups
Once identified manipulation of the pi groups is permitted. These
manipulations do not change the number of groups involved, but may
change their appearance drastically.
Taking the defining equation as: f ( 1 , 2 , 3 m-n ) = 0
Then the following manipulations are permitted:
i. 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 , and 2 may be
combined to form 1a = p1 / p2 so the defining
equation becomes
f ( 1a , 2 , 3 m-n ) = 0
ii. The reciprocal of any dimensionless group is valid. So f (1 ,1/ 2 , 3
1/ m-n ) = 0 is valid.
iii. Any dimensionless group may be raised to any power. So f ( 12 , 20.5 ,
33 m-n ) = 0 is valid.
iv. Any dimensionless group may be multiplied by a constant.
v. Any group may be expressed as a function of the other groups, e.g.
2= f ( 1 ,3 m-n )
In general the defining equation could look like
f ( 1 , 1/2 , 3n 0.5m-n ) = 0

10

Common groups

11

12

Similarity
The underlying concept of dimensional analysisthe principle of similarity.
There are three necessary conditions for complete similarity between a
model and a prototype.
The first condition is geometric similaritythe model must be the same
shape as the prototype, but may be scaled by some constant scale factor.
Geometric similarity exists between model and prototype if the ratio of all
corresponding dimensions in the model and prototype are equal

LModel
L
m L where L is the scale factor
LPr ototype LP
AModel
L2m
2 2L where A is the scale factor
APr ototype L P
All corresponding angles are to be the same.
The second condition is kinematic similarity, which means that the velocity
at any point in the model flow must be proportional (by a constant scale
factor) to the velocity at the corresponding point in the prototype flow as
shown below

13

Specifically, for kinematic similarity the velocity at corresponding points

must scale in magnitude and must point in the same relative direction. You
may think of geometric similarity as length-scale equivalence and kinematic
similarity as time-scale equivalence. Geometric similarity is a prerequisite
for kinematic similarity. Just as the geometric scale factor can be less than,
equal to, or greater than one, so can the velocity scale factor. Some ratios
are

L
V e lo c it y

V
V

LP

L
am
aP

A c c e le r a t io n

Q
D is c h a r g e
Q

T
LP
T

L3 m
m

L
=u
T

L3P

14

2
m
2

L
=
2T

3L

= Q
T

You may recall that streamlines are kinematic phenomena; hence, the
streamline pattern in the model flow is a geometrically scaled copy of that in
the prototype flow when kinematic similarity is achieved.

The third and most restrictive similarity condition is that of dynamic

similarity. Dynamic similarity is achieved when all forces in the model
flow scale by a constant factor to corresponding forces in the prototype flow
(force-scale equivalence). As with geometric and kinematic similarity, the
scale factor for forces can be less than, equal to, or greater than one.
Kinematic similarity is a necessary but insufficient condition for dynamic
similarity. It is thus possible for a model flow and a prototype flow to
achieve both geometric and kinematic similarity, yet not dynamic similarity.
All three similarity conditions must exist for complete similarity to be
ensured.
Here we worry about Force Ratio

F o rc e

R a t io

Fm
FP

m
P

a
a

m
P

2
L

m L3 m
L

X
P L3P
2T
L

2
L

2
u

Models
When a structure is build it undergoes some analysis in the design stage.
Often the structures are too complex for simple mathematical analysis and a
model is build. Usually the model is less than full size but it may be greater.
The real structure is known as the prototype. The model is usually
built to an exact geometric scale of the prototype but in some cases -notably
river model - this is not possible. Measurements can be taken from the
model and a suitable scaling law applied to predict the values in the
prototype.
15

To illustrate how these scaling laws can be obtained we will use the
relationship for resistance of a body moving through a fluid.
The resistance, R, is dependent on the following physical properties:

M
L
M
, u
, l L,
3
L
T
LT
s o f (R , ,u , l, ) 0 m 5, n 3
W e c a n h a v e 2 g ro u p s

For
we
M

a1

have
L 0T

l c1 R

L 0T

a1 1

b1

M
3
L

3 a1 b1 c 1 1

a1

L
T

b1

b1 2

For M

0 a2 1

For

0 b1 2

For

0 3 a 1 b 1 c 1 1 g iv e s

and

so

For

we
M

a2

have
L 0T

R
u 2l

g iv e s

L T

a2 1

M
3
L

0 a2 1

For

0 b 2 1 g iv e s

For

0 3a2 b2 c

and

so

W e c a ll 1 /

b2 2

a2

3 a2 b2 c 2 1

For M

or
c1 2

l c2

M L
T

so a2 1

b2

c1

L
T

b2

b2 2

c2

M
LT

or

so a2 1

1/
ul
2a

16

b2 1
1 g iv e s

c2 1

R e y n o ld s N o

The defining equation becomes f 1, 2a 0

ul
ul
R
2 2

f
or
R

u
l
)
f

u 2 l 2

This equation will apply for all body sizes and is

or

applicable to model and prototype

mumlm
P uP lP
Rm
RP
f
f
and

2 2
2 2
mum lm
PuP lP
m
P
mumlm
Rm
f

2 2

u l
m

or m m m
RP
P uP lP
f

PuP 2lP 2
P
At this point we can go no further unless we make
some assumptions.
We assume bith model and Prototype have same
muml m PuP lP
Reynolds No

m
P
Rm mum2lm2
so now

RP P uP 2lP 2
Which gives the scaling law as R 2u 2l
17

We made the assumption that Rem ReP

mumlm PuP lP
or

and we get
m
P

um P m lP

or u
uP m P lm
l
If we subsitute it in the scaling law for Resistance

R 2u 2l
we get

R (

( )2

( )2

2
)

l
2
2
( ) (l )
( )2

so the force on the prototype can be predicted from

measurement of the force on the model. But only if
they both have the same Reynolds Number, and
um P m lP

uP m P lm
In such a case both model and prototype are
Dynamically Similiar.
5.11.1 Dynamically similar model examples
18

An underwater missile, having 2 m diameter and length of 10 m 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?
For dynamic similarity the Reynolds number of the model and prototype
must be equal:

F o r D y n a m ic S im ila r it y R e

Re

or

ud
ud

m
s o m o d e l v e lo c ity = u m u P
F o r m o d e l a n d P ro to ty p e
so

uP

P dP m
m dm P
P,P

dP
1

10
200m / s.
dm
1 / 20

Note that this is a very high velocity. This is one reason why model tests are
not always done at exactly equal Reynolds numbers. Some relaxation of the
equivalence requirement is often acceptable when the Reynolds number is
high. Using a wind tunnel may have been possible in this example. If this
were the case then the appropriate values of the and ratios need to be
used in the above equation.

19

A model airplane is built at 1/10 scale and is to be tested in a wind tunnel

operating at a pressure of 20 times the atmospheric pressure. The airplane
will fly at 500km/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? The drag is given
by the relation:

ul
D rag ( u 2 l 2 )f
( u 2 l 2 )f R e l

For dynam ic sim ilarity R e m R e P
ul

m
T he value

ul
P lP m
or
u

u
m
P

m lm P

P
of does not change m uch w ith P ressure

If air is taken as an ideal gas then P = R T

so

m
P
m
P
PP

Pm 20 PP

as T does not change

so

20 PP

m so m 20 P
PP
P

P lP
1 1
500
1 / 10 250 km / hr
m lm
20

Dm
( u 2 l 2 )m
T he ratio of forces is given by

DP
( u 2 l 2 )P

um uP

Dm
20 0.5 .1
or

1 1 0.05
DP
1

D P 6 750 N
20