Order of Magnitude Analysis
Order of Magnitude Analysis
Order of Magnitude Analysis
The boundary layer concept, initiated by Prandtl in 1904 has been widely
used to obtain solutions to many important fluid mechanics and heat transfer
problems. His analysis simplifies the complicated Navier Stokes and energy
equations and make it possible to obtain solutions to the problems that do not have
analytical solutions. The base of this concept lies in the principle that for many
applications, the flow of viscous fluid over a surface, the effects of fluid viscosity and
of the no slip condition are important only in a very narrow region adjacent to the
surface. This region is called the Boundary Region and the layer of affected fluid is
called boundary layer.
The boundary layer interpretation is valid only for the portions of the surface
for which the main flow remains attached, that is unseparated. We will see that the
boundary layer is thin compared to the significant dimension of the body for high
Reynolds Number flows.
If the conditions in the boundary region hypothesis are met, the flow field may
be separated into two regions: the main flow, which is inviscid and is described by the
simpler Euler equations; and the boundary layer region, in which Navier Stokes
equations / conditions apply. Prandtl showed that because of the nature of flow in thin
layers, the Navier Stokes equations could be reduced to much simpler form, called
Boundary Layer Equations. To obtain complete solution of a problem, we solve the
main inviscid flow field, by the methods of hydrodynamics. The resulting pressure
distribution is then applied to the boundary layer to solve the boundary layer
equations.
The boundary layer equations are developed as shown in the below sections.
For simplicity consider 2 D flow over a flat plate at zero angle of incidence. The
fluid approaches the plate at a uniform velocity u. A boundary region forms in which
the fluid is decelerated by viscous forces/action. A y component of velocity (v)
arises. The local thickness of the boundary layer is denoted by . The relevant Navier
Stokes equations, the continuity equation and the boundary conditions are shown
below. [Heat Transfer B. Gebhart]
Let us consider steady, 2 D flow with constant property without any body force. The
continuity equation and Navier Stokes equation in Cartesian coordinates are given
by
u v
+ =0
x y
u
u
x
+v
u
y
1 p
=
x
+
(
2 u 2 u
+
x2 y 2 )
Boundary Conditions are:
as y , u u.
Let us express above equations in non dimensional form. The reference quantities
taken for non - dimensionalization are
i. reference length, L, for the flow over plate, length of the plate is taken as
characteristic length, where as for the case of pipe, diameter is taken as the
characteristic length.
u* v *
+ =0
x* y*
u*
u* * u *
x*
+ v
y*
= +
(
p* 1 2 u* 2 u *
+
x* R e x *2 y *2 )
u*
v* * v*
x*
+ v
y*
= +
(
p* 1 2 v * 2 v *
+
y * R e x *2 y *2 )
x: distance in x (or length) direction. The length of the plate varies from 0 L.
Therefore, we take the order of magnitude of x is equal to the of the order of length of
plate.
y: boundary layer is very thin compared to length of the plate. The velocity gradients
in y direction are limited to a small region, which is known as boundary layer
thickness, . The region in y direction over which the gradients present are from y =
0 and y ~ . Therefore, its (y - ) order can be taken to be small quantity in comparison
to the plate length.
Before the plate, there is no y component of velocity. But, this velocity component
(v) is present in boundary layer. Due to the shear forces, the mass entering a
differential control volume within boundary layer in x direction is not same as the
mass leaving the control volume in x direction. To satisfy the continuity, the mass
leaves the control volume in y direction, its value being the difference between the
mass entering and mass leaving. There is no mass flow through the bottom surface of
the control volume (at the surface), due to the assumption of no penetration condition.
If the control volume is taken away from the surface within the control volume, we
may have mass flow passing through both the faces whose normals are in y
direction.
Then how to find the magnitude or order of v*? we can make use of continuity
equation for finding v. In the continuity equation, we have u, v, x, y. we know the
orders of x, y, u. From these values, we can find the order or v. In continuity equation,
we can not neglect any term. All terms in continuity equation are equally important.
u* v *
+ =0
x* y*
u* O(1)
~ ~ O(1)
x * O(1)
*
v ()
~ O(1)
y O( *)
*
O (1) ?
O (1) O ( * )
* *
O (v ) ~ O( )
( )
* * * 2 * 2 *
u * * u p 1 u u
x momentum equation: u *
+v *
= * + *2
+ *2
x y x Re x y
LHS represents inertia forces. We want the left side as the fluid motion is
represented by the momentum terms and hence by inertial forces. Both inertial terms
are of same order. So we cannot neglect any of the inertial term. If we compare the
inertia and viscous forces, the magnitude of viscous forces are small compared to
inertia forces, when Re is high. Prandtl thought that for high Reynolds number flows,
1 2 u* 2 u*
We found/assumed that << L or * << 1. 1 . , which
*2
x *2 y *2
2 u* 2 u*
means that may be neglected in comparison to . physically also, this
x *2 y *2
u*
is consistent. In x direction, the velocity gradient is much smaller in
x*
*
u
comparison with the gradient in y direction * . In this way, we are not
y
completely ignoring the viscous terms. As a result, viscous term in x direction is
1 2 u*
. As we considered that the viscous forces are to present, magnitude of
Re y *2
viscous forces should be equal magnitude of inertia forces. Pressure is function of
geometry. Then the order of x momentum reduces to
1 1
O(1)
R e O( *2 )
1
O ( *2 )
Re
* 1 1 /2
O ( ) 1 /2
R e
Re
x ( )
O R e x
1/ 2
1 /2
O ( )x R e x
For flat plate, the pressure gradient in x direction is zero. For the analysis purpose,
we may assume the pressure gradient is also of the same order of the inertial terms. In
the momentum equation, we can't straight away neglect the pressure term in
comparison with inertia forces.
u*
v* * v*
x*
+ v
y*
=
p* 1
+
y* R e ( 2 v * 2 v *
+
x *2 y *2 )
Order of Magnitude Analysis 5 P. Anil Kishan
( )
* * * *
O( ) O( ) 1 O( ) O( )
O(1) ; O(*) ; ; ;
O(1) O(*) R e O( 1) O( *2)
* * * 3 *
O( ); O( ) ; ; O( ) ; O( )
1
Last two terms are obtained by substituting O(*2)
Re
Except the pressure term, other terms are of the order of *. If we calculate the order
of magnitude of Pressure term in y direction, we can observe that its order of
magnitude is *. Therefore,
p
O ( *)0 pf ( y)
y
This implies physically that the pressure gradient within the boundary layer in the
direction normal to the flow is negligible in comparison to the pressure gradient in
flow (x ) direction. This means that the pressure varies only in x direction.
Therefore,
p( x , y )p ( x )
p dp
=
x dx
u* v *
Conservation of mass: + =0
x* y*
* * * 2 *
* u u p 1 u
x Momentum Equation: u *
+ v * * = * +
x y x R e y *2
p*
y momentum equation: =0
y*
After combining both the momentum equations, we get the momentum equation of
the following form, with continuity equation remaining the same.
u* * u * dp* 1 2 u *
u* + v = +
x* y* dx * R e y *2
u u 1 dp 2 u
u +v = + 2
x y dx y
dp/dx includes the effect of the geometry. For flat plate, it is zero. For the curved
surface, dp/dx is not equal to zero.
For the case of flat plate, we can find dp/dx from the flow in the outer region, where
we can use Euler's equation.
p1 = p2 for flat plate. Therefore, the pressure gradient inside the boundary layer
is also equal to the pressure gradient outside the boundary layer. In the outer region,
we can apply Bernoulli equation or Euler's Equation as Prandtl had shown that the
viscous effects can be neglected in the region above boundary layer. Considering the
Euler's equation,
1 dp d u
dx +u dx =0
dp du
= u =0
dx dx
dp
0
dx
As an engineer, we are interested in the shear forces at the wall and the
coefficient of friction. Let us apply the order of magnitude analysis for the shear stress
and the coefficient of friction. in scaling terms, flow friction / shear stress is given by:
=
( uy )
y=0
= ( )
U u *
L y* *
y =0
~
L ( )
U u *
y* *
y =0
=
U u*
L y* ( ) *
y =0
~
U 1
L *
To calculate the wall shear stress, we must evaluate * (or ) (i.e., extent of
of the imaginary boundary layer region).
~ u2 Re1 /2
L
The above relation gives the shear stress. But this value is dependent on the
fluid and flow properties. We need a quantity, which is fluid and flow independent.
Therefore, we define dimensionless skin friction coefficient, Cf, which is a
dimensionless number. Hence, it may be useful for any fluid and flow condition. Skin
friction coefficient is defined as:
1
Cf= 2
~ ReL1 /2 ~ .
1/2 U Re L
We found that skin friction coefficient is function of ReL-1/2 or Rex-1/2 (if we are
finding Reynolds number at a location).
This relation does not give exact value for Cf. But this relation tells that Cf is a
function of Reynolds number and gives the exact nature of the function. We need
coefficient that relates the Cf and Re. This coefficient is found from exact solutions or
experiments. This outcome from order of magnitude analysis is in accordance with the
derivation using Blasius solution or from Momentum Integral Method solution, which
are cumbersome and lengthy in nature. The importance of order of magnitude analysis
seems to be very simple in current days. But the insight it gave during the last century
where there are no computers, is worth appreciating. This order of magnitude analysis
can be applied in other fields also. It will tell which terms are important and which
terms can be neglected without changing the final result much. Magnitude analysis
simplifies the equation so that the solutions may be obtained relatively easily
compared to solving the full equations.
x
u
This tells that larger the free stream velocity, thinner the boundary layer.
In other words, x 1/ 2 . This means the boundary layer thickness increases with x.
This relation tells us that the coefficient of friction decreases with x. this is
obvious or expected as we saw that the velocity gradients decrease as we move along
the plate. This decreases the shear stress and the coefficient of friction.
Compare this result with the result obtained using the similarity studies.
C f , x =0.664 R e 1/x 2 Using similarity studies/analysis.
We can see that at the leading edge of the plate, boundary layer thickness
increases sharply and coefficient of friction decreases sharply. This may be predicted
as: as the boundary layer starts the velocity gradients are very high. Hence Cf will be
high and will be low near the leading edge.
Energy Equation:
for 2 D, steady flow with constant properties and without heat generation and
viscous dissipation, energy equation is given by
C p u (
T
x
+v
T
y
=k 2) (
2 T 2 T
x y
+ 2
)
In non dimensional form, the energy equation is given as
( ) 1
( )
* *
u* * + v * * = 2 * + 2 *
x y R e Pr x *2 y *2
For the case of Pr >> 1, >> t, the x velocity within the TBL is smaller compared
to the velocity scale of HBL. Hence, we have to consider order of magnitude of x
velocity within the TBL accordingly. This can be calculated as: velocity scale in
HBL : 0 u. Velocity within TBL : 0 u. Order of magnitude of y for TBL is
u
against (since the velocity changes only from 0 u; not from 0 u). Order
u
of magnitude for y velocity needs to be calculated accordingly.
From the boundary layer analysis similar to HBL, (L >> t), we may write for the
diffusion terms,
2 * 2 *
*2 *2
x y
( ) 1 2 *
* *
u* * + v * * =
x y R e Pr y *2
(
u* * +v * * =
x y
1
)2 + 2
R e Pr x *2 y *2 ( )
O(1)
O(1)
O(1)
O( )
* O(1)
*
O( t )
1 O(1) O(1)
(
R e Pr O(1) O( t )* 2 )
T
where = ; T = ; L length scale
* *
L L
2 2 1
1 . Therefore,
From magnitude analysis, *2 *2 as t
x y
2 *
conduction term on RHS is represented by only . Therefore, the above
y *2
1 2
equation reduces to ( u* * + v * * =
x y )R e Pr y *2
2 T
(
C p u
T
x
+v
T
y
=k
y2) .
L k
h= f
(T sT )
T
( )
L y* *
y =0
hL
*
kf ( )
y *
y =0
hL
*
Nu
kf y ( ) *
y =0
1
Nu *
( )
y *
y =0
*
t
Conduction ~ convection
T T T
u ,v ~ 2
L T T
T T
Magnitude of 2nd term is given as: v ~ u
T L T
here, / t <<1. Therefore, 2nd term on left side can be neglected in comparison to 1 st.
Now the energy equation reduces to
*
1 2 *
u * * ~
x R e Pr y *2
O(1) 1 O(1)
O(1)
O(1) Re Pr O( *t )2
*T
* ~ Pr1 /2 1
Relative size of , t depends on Pr (Pr = /).
The assumption of t >> is valid in the limit Pr1/2 <<1, which represents the
range occupied by the liquid metals.
As engineers, we need the convective heat transfer coefficients rather than any
other terms. From the above analysis and the definition of h, we can write,
T
k f *
( )
y y=0 L k
~ f *
h=
T L y ( ) y=0
hL
~ *
K y( ) *
y =0
hL 1
~ ~ Re1 /2 Pr1 /2
K *T
Nu ~Re 1/2 1/ 2
L Pr Pe1/ 2 .
By doing similar analysis for the thin thermal boundary layers (Pr >> 1), we
1/2 1/3
obtain Nu ~Re L Pr . For deriving this, you need to think about the order of
magnitudes of velocity in thermal boundary layer. This has been explained in the
previous sections. From the Boundary layer solution and the numerical analysis, we
1 /2 1 /3
get Nux =0.332 Re x Pr
References: