Similitude and Model Testing

1. Introduction
Fluid flow analysis is involved in the design of aircrafts, ships, submarines, turbines, pumps,
harbors and tall buildings and structures. Fluid flow is influenced by several factors and because of
this the analysis is more complex. For many practical situations exact solutions are not available. The
estimates may vary by as much as ± 20%. Because of this it is not possible to rely solely on design
calculations and performance predictions. Experimental validation of the design is thus found
necessary. Consider the case of a hydraulic turbine of 50 MW size. It will be a very costly failure if
the design performance and the actual performance differ. If we can predict its performance before
manufacturing the unit it will be very useful. Model testing comes to our aid in this situation.
Constructing and testing small versions of the unit is called model testing. Similarity of features
enables the prediction of the performance of the full size unit from the test results of the smaller
unit. The application of dimensional analysis is helpful in planning of the experiments as well as
prediction of the performance of the larger unit from the test results of the model.

2. Model and Prototype

In the engineering point of view model can be defined as the representation of physical system
that may be used to predict the behavior of the system in the desired aspect. The system whose
behavior is to be predicted by the model is called the prototype. Our discussion is about physical
models that resemble the prototype but are generally smaller in size. These may also operate with
different fluids, at different pressures, velocities etc. As models are generally smaller than the
prototype, these are cheaper to build and test. Model testing is also used for evaluating proposed
modifications to existing systems. The effect of the changes on the performance of the system can
be predicted by model testing before attempting the modifications. Models should be carefully
designed for reliable prediction of the prototype performance.

3. Conditions for similarity between models and prototype

Dimensional analysis provides a good basis for laying down the conditions for similarity. The 
theorem shows that the performance of any system (prototype) can be described by a functional
relationship of the form given in equation 1.
π 1 p =f ( π 2 p , π 3 p ,⋯, π np )
(1)
The  terms include all the parameters influencing the system and are generally ratios of forces,
lengths, energy etc. If a model is to be similar to the prototype and also function similarly as the
prototype, then the  terms for the model should also have the same value as that of the prototype
or the same functional relationship as the prototype. (equation 1)
π 1m=f ( π 2 m , π 3m ,⋯, π nm )
(2)
For such a condition to be satisfied, the model should be constructed and operated such that
simultaneously
π 1m=π 1 p , π 2m=π 2 p ,⋯,π nm=π np (3)
Equation 3 provides the model design conditions. It is also called similarity requirements or
modeling laws.
3.1 Geometric Similarity
Some of the  terms involve the ratio of length parameters. All the similar linear dimension
of the model and prototype should have the same ratio. This is called geometric similarity. The

1
 ratio is generally denoted by the scale or scale factor. One tenth scale model means that the
similar linear dimensions of the model is 1/10 th of that of the prototype. For complete
similarity all the linear dimensions of the model should bear the same ratio to those of the
prototype. There are some situations where it is difficult to obtain such similarity. Roughness is
one such case. In cases like ship, harbour or dams distorted models only are possible. In these
cases the depth scale is different from length scale. Interpretation of the results of the tests on
distorted models should be very carefully done. Geometric scale cannot be chosen without
reference to other parameters. For example the choice of the scale when applied to the
Reynolds number may dictate a very high velocity which may be difficult to achieve at a
reasonable cost.
3.2 Dynamic Similarity
Similitude requires that  terms like Reynolds number, Froude number, Weber number etc.
be equal for the model and prototype. These numbers are ratios of inertia, viscous gravity and
surface tension forces. This condition implies that the ratio of forces on fluid elements at
corresponding points (homologous) in the model and prototype should be the same. This
requirement is called dynamic similarity. This is a basic requirement in model design. If model
and prototype are dynamically similar then the performance of the prototype can be predicted
from the measurements on the model. In some cases it may be difficult to hold simultaneously
equality of two dimensionless numbers. In such situations, the parameter having a larger
influence on the performance may have to be chosen. This happens for example in the case of
model tasting of ships. Both Reynolds number and Froude number should be simultaneously
held equal between the model and prototype. This is not possible as this would require either
fluids with a very large difference in their viscosities or the use of very large velocities with the
model.
3.3 Kinematic Similarity
When both geometric and dynamic similarities exist, then velocity ratios and acceleration
ratios will be the same throughout the flow field. This will mean that the streamline patterns will
be the same in both cases of model and prototype. This is called kinematic similarly. To achieve
complete similarity between model and prototype all the three similarities - geometric, dynamic
and kinematic should be maintained.

4. Types of model studies.

Model testing can be broadly classified on the basis of the general nature of flow into four types.
These are
(1) Flow through closed conduits
(2) Flow around immersed bodies
(3) Flow with free surface and
(4) Flow through turbomachinery

4.1 Flow through Closed Conduits

Flow through pipes, valves, fittings and measuring devices are dealt under this category. The
conduits are generally circular, but there may be changes along the flow direction. As the wall
shear is an important force, Reynolds number is the most important parameter. The pressure
drop along the flow is more often the required parameter to be evaluated. Compressibility
effect is negligible at low Mach numbers. (M<0.3).
From dimensional analysis the pressure drop can be established as

2
 ΔP
ρu
2
=f(ρuL ε D
, ,
μ L L ) (4)
The geometric scale is given by the ratio, scale = Lm/Lp.
D m ε m Lm
= = =λ
This requires D p ε p L p

Reynolds number similarity leads to the condition for velocity ratio as

u m ρ m Lm u p ρ p L p u m μm ρ p L p
= =
μm μ p  u p μ p ρ m Lm (5)
If the fluid used for the model and prototype are the same, then
um Lp up
=
u p Lm or um= λ

As  is less than one, the velocity to be used with the model has to be higher compared to
the prototype. Otherwise a different fluid with higher viscosity should be chosen to satisfy the
requirements.
The pressure drop in the prototype is calculated as in equation 6.
From equality of, P/u2,

( )
2
ρ u
ΔP p = p p ΔPm
ρ m um
(6)
As Pm is measured, using the model, the pressure drop in the prototype can be predicted.
When Reynolds numbers are large the inertia forces are predominant and viscous forces will
be small in comparison. In such cases, the Reynolds number similarity becomes unimportant.
However, the model should be tested at various Reynolds numbers to determine the range at
which its effect on pressure drop becomes negligible. After this is established the model test
results can be applied without regard to Reynolds number similarity, in this range.
Another condition is the onset of cavitation at some locations in the flow, particularly in
testing components where at some points the local velocity may become high and pressure may
drop to a level where cavitation may set in. Unless cavitation effects are the aim of the study,
such condition should be avoided. In case cavitation effects are to be studied, then similarity of
cavitation number should be established. i.e. (pr – pv)/(u2/2). Where pr is the reference pressure
and pv is the vapor pressure at that temperature.

4.2 Flow Around Immersed Bodies

Aircraft, Submarine, cars and trucks and recently buildings are examples for this type of
study. In the sports areas, golf and tennis balls are examples for this type of study. Models are
usually tested in wind tunnels. As viscous forces over the surface and inertia forces on fluid
elements are involved in this case also, Reynolds number of the model and prototype should be
equal. Gravity and surface tension forces are not involved in this case and hence Froude and
Weber numbers need not be considered. Drag coefficient, defined by [Drag force /(1/2) u2 l2)]
is the desired quantity to be predicted. Generally the following relationship holds in this case.

C D=
D
2 2
( 1/2 ) ρu l [l 1 ε ρ ul
=f , ,
l l μ ] (7)

3
 where l is a characteristic length of the system and l1 represents the other length parameter
affecting the flow and  is the roughness of the surface.
When the flow speed increases beyond Mach number 0.3 compressibility effect on similarity
should be considered. Using the similitude, measured values of drag on model is used to
estimate the drag on the prototype.

( )
2
ρp u p lp
D p =Dm ×
ρ m um l m
(8)
From Reynolds number similitude
μm ρ p l p νm l p
um = u p= up
μ p ρm l m ν p lm (9)
When same fluid is used for both prototype and model

()
l
um= p u p
lm
(10)
The model velocity should be higher by the geometric scale.
If the prototype is to operate at 100 kmph and if the scale is 1:10, then the model should
operate at 1000 kmph, which will mean a high Mach number. The model will be influenced by
compressibility effect due to the operation at high Mach numbers. The prototype however will
be operating at low Mach numbers where compressibility effect is negligible. Hence the
performance prediction will be in error.
This may be overcome by using different fluids say water in place of air. Using equation 9, as
kinematic viscosity of air is about 10 times that of water, the velocity will now be at a reasonable
level. Another method is to pressurize the air in the wind tunnel, thus increasing the density,
and reducing the required velocity of the model.
Where expense is of no consideration due to the requirement of utmost reliability as in
space applications and development of new aircraft, full scale models are also used.

4.3 Flow with Free Surface

Flow in canals, rivers as well as flow around ships come under this category. In these cases
gravity and inertia forces are found to be governing the situation and hence Froude number
becomes the main similarity parameter.
In some cases Weber number as well as Reynolds number may also influence the design of
the model.
Considering Froude number, the velocity of the model is calculated as below.
um up
=
√ glm √ gl p (11)

um=u p
√ lm
lp
=u p √ scale
(12)
In case Reynolds number similarity has to be also considered, substituting this value of
velocity ratio, the ratio of kinematic viscosities is given as
νm
=( scale )3/2
νu (13)

4
 As these situations involve use of water in both model and prototype, it is impossible to
satisfy the condition of equations 12 and 13 simultaneously. In such a case distorted model
may have to be selected.
If surface tension also influences the flow, it is still more difficult to choose a fully similar
model.
In many practical applications in this type of situation the influence of Weber and Reynolds
number is rather small. Hence generally models are designed on the basis of Froude number
similarity.
A special situation arises in the case of ships. The total drag on the ship as it moves is made
up of two components: (1) The viscous shearing stress along the hull, (2) Pressure induced drag
due to wave motion and influenced by the shape of the hull.
As it is not possible to build and operate a model satisfying simultaneously the Reynolds
number similarity and Froude number similarity ingenious methods have to be adopted to
calculate the total drag. The total drag on the model is first measured by experiment. The shear
drag is analytically determined and the pressure drag on the model is calculated by subtracting
this value. The drag on the prototype is determined using Froude number similarity. The
calculated value of viscous drag is then added to obtain the total drag.
In case of design of river model, if the same vertical and horizontal scales are used, the
depth will be low for the model and surface tension effects should be considered. But the use of
distorted model, (vertical scaling smaller than horizontal scaling) overcomes this problem.

4.4 Models for Turbomachinery

Pumps as well as turbines are included in the general term turbomachines. Pumps are
power absorbing machines which increase the head of the fluid passing through them. Turbines
are power generating machines which reduce the head of the fluid passing through them.
The operating variables of the machines are the flow rate Q, the power P and the speed N.
The fluid properties are the density and viscosity. The machine parameters are the diameter and
a characteristic length and the roughness of the flow surface. Power, head and efficiency can be
expressed as functions of  terms as in equation 14

Power=f 1 ( l ε Q ρ ND 2
, , ,
D D ND 3 μ ) (14)
The term /D is not important due to the various sharp corners in the machine. The
dimensionless term involving power is defined as power coefficient, defined as Cp = P/(N3D3).
The head coefficient is defined as Ch = gh/(N2D2) . The term Q/(ND3) is called flow coefficient. If
two similar machines are operated with the same flow coefficient, the power and head
coefficients will also be equal for the machines. This will then lead to the same efficiency.
Combining flow and head coefficients in the case of pumps will give the dimensionless specific
speed of the pump.
N √Q
N sp =
( gh )3/4 (15)
Popularly used dimensional specific speed for pumps is defined as
N √Q
N sp=
h3 /4 (16)
In the case of turbines, combining power and flow coefficients, the specific speed is
obtained as

5
 N √P
N st = 1/2 5/ 4
ρ ( gh ) (17)
Popularly used dimensional speed for turbines is
N √P
N st =
h5/4 (18)
In model testing at a particular speed, the flow rate at various delivery heads can be
measured. This can be used to predict the performance of the pump at other speeds using the
various coefficients defined. The procedure for turbines will also be similar. The model can be
run at a constant speed when the head is varied, the power and flow rate can be measured. The
performance of the prototype can be predicted from the results of the tests on the
geometrically similar model.

5. Sample Problems
5.1 To determine the pressure drop in a square pipe of 1 m side for air flow, a square pipe of 50 mm
side was used with water flowing at 3.6 m/s. The pressure drop over a length of 3 m was
measured as 940 mm water column. Determine the corresponding flow velocity of air in the
larger duct and also the pressure drop over 90 m length. Kinematic viscosity of air = 14.58 × 10–
6 m2/s. Density = 1.23 kg/m3. Kinematic viscosity of water = 1.18 × 10 –6 m2/s
Given:
Water:
a w =0 .050 m
uw =3 . 6 m/s
ΔP w =940 mm w .c .
Lw =3 m
−6 2
ν w=1. 18×10 m / s
Air:
a air =1 m
Lair=90 m
3
ρair =1. 23 kg/m
−6 2
ν air =14. 58×10 m / s

Required: uair , ΔP air

Solution:
For pipe flow, Reynolds number analogy should be used. Also the drag coefficients will be
equal.
For square section hydraulic mean diameter , Dw = 4A/P = 4aw2/4aw = aw (side itself)
u w Dw 3 . 6×0 . 05
Re= = =152 ,542
νw 1 .18×10−6
For air:
u air Dair u air ×1
Re= = =152 , 542
ν air 14 .58×10−6

6
 uair =2. 224 m/ s
Drag coefficient F/(u) should be the same for both pipes
F air ρair u2air
=
F w ρ w u2w
The pressure drop equals the shear force over the area. For square section, area = a2, perimeter
= 4a.
4 FL ΔP = 4 F air Lair ΔP = 4 F w Lw
ΔP= air
a , a air , w aw
Dividing and substituting for Fair/Fw.

( )
2
ΔPair Lair aw F air L air a w ρair u air
= × × = × ×
ΔP w Lw a air F w Lw a air ρ w uw
ΔPair 90 0 .05 1.23 2 .224 2
ΔP w 3
= × ×
1 1000 3 .6 ( ) =7 .04143×10−4

ΔP air =( 7 . 04143×10−4 ) ( 940 )=0. 662 mm w . c .

5.2 In order to predict the flow conditions after the turbine outlet (tail race) of a hydroelectric plant
delivering 2400 m3/s, a model of 1/75 scale is proposed. Determine the flow rate required.
Given:
3
Q p =2400 m /s
Lm 1
=
L p 75
Required:
Qm
Solution:
This is a free surface flow. Hence Froude number similarity is to be maintained.

√
um up um lm
= =
Fr m=Fr p or √l m √l p or u p lp
As flow (Q = Au) depends on area which varies as L2
Q m Am um L2m Lm Lm
√ ()
2. 5
= = =
Q p A p u p L2p L p L p

( )
2. 5
1
Qm=2400 =0 .04927 m3 /s
75
5.3 A centrifugal pump with dimensional specific speed (SI) of 2300 running at 1170 rpm delivers 70
m3/hr. The impeller diameter is 0.2 m. Determine the flow, head and power if the pump runs at
1750 rpm . Also calculate the specific speed at this condition.
Given:
D1 = D2 = 0.2 m
N1 = 1170 rpm
Ns1 = 2300 rpm
Q1 = 70 m3/hr
N2 = 1750 rpm
Required:

7
Q2
H2
P2
Ns2
Solution:
The head developed and the power at test conditions are determined first. (At 1170 rpm).
N 1 √ Q 1 1170 √ 70
N s 1= 3/4 = =2300
H1 H 3/4
1
H 1=6 .9 m
Power

P 1 =ρ gQ 1 H 1 =( 9810 N /m3 )( 70 m3/hr ) ( 6.9 m ) ( )

1 hr
3600 s
=1316 W
When operating at 1750 rpm, using flow coefficient Q/(ND3), as D is the same

Q2 =70 (1750
1170 )
=104 . 7 m /hr3

Using head coefficient, H/(N2D2),

H 2 =H 1
( ) ( )
N2 2
N1
=6 . 9
1750 2
1170
=15 . 44 m

Using power coefficient : P/(N3D5),

( )
3

( )
N 1750 3
P2 =P 1 1 =1316 =4404 W
N2 1170
Specific speed for the model
N 2 √ Q2 1750 √ 104 . 7
N s 2= 3/4 = =2300
H2 (15 . 44 )3/4
Note: Specific speeds are the same
- End -