Turbulence Prof. E.G.

Tulapurkara
Chapter-3

Chapter 3

Lecture 19

Equations for turbulent flows - 4

Appendix D Elements of laminar boundary layers and transition

Appendix E Equation for dissipation

Appendix-D

Elements of laminar boundary layers and transition

D.1 Introduction
The concept of boundary layer has been introduced in Appendix ‘C’ on “Basic concept
and equations of fluid dynamics”. Recall that when a fluid moves past a stationary
body, there is a thin region near the surface of the body in which the velocity changes
from, zero at the surface, to the velocity outside. This region is called the boundary layer
(BL).

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 1

Turbulence Prof. E.G. Tulapurkara
Chapter-3

Fig. D.1 Schematic of boundary layer (a) Over streamlined body (not to
scale) (b) Over bluff body (not to scale)

Some features of the boundary layer over the surface of a stream lined body are shown
in Fig D.1a. The features of the flow are as follows.
1. Near the leading edge or the nose of the body, the flow is brought to rest. This point
is called the stagnation point. Subsequently, a laminar boundary layer develops (a)
on the
surface.
(b) the
2.Depending on the Reynolds number, the pressure gradient and other parameters,
boundary layer becomes turbulent after undergoing transition. The turbulent boundary
layer may continue till the trailing edge of the body or may separate. In some cases the
laminar boundary layer may separate before the transition. In some other cases, the
turbulent boundary layer may separate before the trailing edge.

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 2

Turbulence Prof. E.G. Tulapurkara
Chapter-3

The features of the boundary layer on a bluff body are shown in Fig.D.1b. In this case
the boundary layer develops upto a certain distance along the body and then separates.
Though Fig.D.1b shows separation of turbulent boundary layer, in some cases the
laminar boundary layer may separate before undergoing transition.

The Nature of boundary layer decides the drag and the heat transfer from the body. If
the boundary layer separates, the drag increases manifold.

For these reasons a study of boundary layer and its transition are important topics in
fluid dynamics.

D.2 Laminar boundary layer

The simplification of N-S equations with the assumption that the boundary layer is thin
has been carried out in Section C.9.5. The equations for the 2-D incompressible flow
are (Eqs. C.119 and C.120):

U V
+ =0 (D.1)
x y

U U 1 dp  2U
U +V = - +v 2 (D.2)
x y ρ dx y
The pressure p is given by Bernoulli equation (Eq. C.119) as:
1 1
p + ρ V2 = p+ ρU2e (D.3)
2 2
Hence,
1 dp dUe
- = Ue (D.4)
ρ dx dx
The boundary conditions are: (a) y = 0, U = 0 and V = 0; (b) y = , U = Ue (D.5)
Remarks :
i) Ue is given either from potential flow analysis or from experimental data. Equations
D.1 & D.2 are solved to obtain U and V.

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 3

Turbulence Prof. E.G. Tulapurkara
Chapter-3

ii) Equation (D.2) is parabolic in nature whereas N-S equations are elliptic in nature.
From theory of differential equations, it is known that for a problem governed by
parabolic equation, one can start with profiles of U and V at a point and then compute
the downstream values.

 2U
iii) Even if Reynolds number is very high, the term v in equation (D.2) cannot be
y 2
neglected. If it is neglected, the order of the equation decreases and the treatment of
flow becomes inviscid. Consequently, the equations would reduce to Euler equations.
For a viscous flow, both the tangential as well as the normal velocity components must
be zero on the surface. For an inviscid flow, only the later condition is needed.

D.2.1 Analytical solution of laminar boundary layer

Equations (D.1) and (D.2) are partial differential equations. For the case of boundary
layer developing in a uniform external stream i.e. Ue = constant, Blasius, in 1908, was
able to reduce the system of equations to ordinary differential equations by employing
similarity hypothesis. Later, it was shown that this approximation is valid also for
external velocity given by Ue = c xm. Note m = 0 is the case of uniform external velocity.

Similarity hypothesis
The thickness of BL () and the external velocity Ue change with distance(x) along the
body surface. However, if the flow is self-similar, the velocity profiles at different x
locations can be made to coincide with each other by making the profiles non-
dimensional with suitable choice of reference velocity and length scale.
Let, the reference velocity be Ue(x) for U, and the reference length be g(x) for y. Then,
the self-similarity hypothesis relates the velocity profiles at stations x1 and x2 by the
following relationship.
 y   y 
U x1 ,  U x 2 , 
 g  x1    g  x 2  
= (D.6)
Ue  x1  Ue  x 2 

The solution procedure can be illustrated by considering the case of boundary layer
over a flat plate with uniform external stream i.e. Ue= const .
Dept. of Aerospace Engg., Indian Institute of Technology, Madras 4
Turbulence Prof. E.G. Tulapurkara
Chapter-3

To simplify the equations the stream function , defined below is used.

ψ ψ
U= ,V=- (D.7)
y x
With this definition of , the continuity equation is automatically satisfied. Using
expression for U and V from Eq.(D.7), in Eq.(D.2), it can be rewritten as:
ψ  2ψ ψ  2ψ dUe 3 ψ
- = U + (D.8)
y xy x y 2 y 3
e
dx
The boundary conditions given by Eq.(D.5) appear as:
ψ ψ ψ
y =0; = 0, =0;y =, = Ue  x  (D.9)
x y y
When the flow is self-similar,
U V y
= F η , = G  η  with η = (D.10)
Ue Ue δ

Let,  = 0 at wall i.e at  = 0. Then, from equation (D.7)

y η η

ψ =  Udy = Ue  F η  δ dη = Ue δ  F η  dη = Ue δ f η  (D.11)
0 0 0

Consequently,
ψ
U= = Ue f   η  (D.12)
y
ψ 
-V = = Ue δ f η = Ue δ f -Ue δ f  (D.13)
x x
f dδ
where, f  = ; δ = (D.14)
η dx

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 5

Turbulence Prof. E.G. Tulapurkara
Chapter-3

Substituting these in equation (D.8) and simplifying gives :

Ue
f  + δ.δ ff  = 0 (D.15)

f  U
Or = - e δ.δ (D.16)
ff  

Since, f and x are the independent variables, for similar solution, the left hand side and
the right hand side of equation (D.16) must be constant. i.e.:
Ue δ.δ U δ2
= const or e = cx +c1
  2
Taking the coordinates such that  = 0 at x = 0, yields :

Ue δ2 2cxv
= cx or δ =
 2 Ue

x
Choosing c = ½, renders δ =
Ue

Ue
Or the similarity variable η becomes y .
x
Equation (D.15) now becomes :
1
f  + ff  = 0 (D.17)
2

Remarks :

i) Blasius obtained an analytical solution for equation (D.17) in 1908.

ii) Subsequently, others also obtained its solution. Results of the numerical solution
obtained by Howarth, are given in chapter 7 of Ref.D.1, 1968 edition. The same
reference be consulted for bibliographic details of the article by Howarth. The quantities.
f, f and f are tabulated for  = 0 to 8.8. Variations of f and f are shown in Fig. D.2.
Important features are:
At  = 0, f = 0, f = 0.332 ; at  = 5, f = 0.9916.

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 6

Turbulence Prof. E.G. Tulapurkara
Chapter-3

Fig. D.2 Variations of f and f with η

iii) Following features of the boundary layer are noted.

a) Boundary layer thickness :
The velocity in boundary layer reaches the free stream value gradually.
U
At = 5, f    0.99 .
Ue

U
If δ0.99 is the height at which is 0.99 (Fig.D.3), then noting definition of :
Ue
Ue
5 = δ0.99
x
δ0.99 5 5 U x
Or = = ; Rx = e (D.18)
x Ue x Rx 


b) Displacement thickness and momentum thickness :

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 7

Turbulence Prof. E.G. Tulapurkara
Chapter-3

The displacement thickness (1 or *), is defined as:


 U
δ1 or δ =   1 -
*
 dy (D.19)
0 Ue 

The momentum thickness (2 or ) is defined as:


U U
δ2 or θ = 0 U  1-  dy
e  Ue 
(D.20)

Fig. D.3 Boundary layer

Using the values of f and integrating gives :

δ / x  or δ
1
*
/ x =
1.7208
Rx
(D.21)

0.664
 δ2 / x  or θ/ x  = (D.22)
Rx

The shape parameter (H) is defined as :

δ1 δ*
H= = (D.22a)
δ2 θ
From Eqs.(D.21) and (D.22), the shape parameter (H) for this boundary layer is 2.6.
c) Skin friction drag coefficient :
 U 
Shear stress at wall =  w = μ   (D.23)
 y wall

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 8

Turbulence Prof. E.G. Tulapurkara
Chapter-3

U Ue  (U/ Ue ) U (f ) U
  e  e f 
y x  Ue  x  x
 y 
Ue  x  Ue Ue

Ue
Hence w   fw
x
Ue

Note : fw  0.332

The local skin friction drag coefficient ( Cf ) is defined as :

w
Cf = (D.24)
1
ρU2e
2
Ue fw 0.664
Hence, Cf    (D.24a)
x 1 v 2 Rx
e
ve 2

The drag per unit width for a plate of length L over which the boundary layer is laminar,
is given by:
L
Drag = D =   w dx,
0

Hence, the skin friction drag coefficient (Cf) of the plate of length ‘L’ is given by :
D 1.328 U L
Cf = = , RL = e (D.25)
1 2
ρUe L RL 
2

Example D.1
Consider Ue = 10 m/s, x = L = 0.5 m,  = 14x10-6 m2/sec.

10×0.5
Then, Rx = RL = -6
= 3.57x 105.
14×10

Using Eqs. (D.18), (D.21), (D.22) and (D.25):

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 9

Turbulence Prof. E.G. Tulapurkara
Chapter-3

0.99 = 4.19 mm, 1 = 1.44 m, 2 = 0.557 mm, H = 2.6 and Cf = 0.0022

Note that  δ0.99 / x  in the present case is (4.19 / 500) i.e. 0.0084. This confirms the

assumption that the boundary layer thickness is much smaller than the length x.

iv). Chapter 7 of Ref. D.1, 2000 edition, gives similarity solutions when external velocity
is given by Ue = cxm.

v). Besides the aforesaid analytical approach based on the similarity hypothesis, there
are many other approaches to obtain boundary layer over practical shapes. In the
approach called “Integral methods” the boundary layer parameters like Cf , H and θ are
obtained for flow past airfoils by numerically solving the following momentum integral
equation, whose derivation is given in section C.10 of Appendix C.

Cf dθ θ dU
= + H+2  e (D.26)
2 dx Ue dx

It may be pointed out that Ue in Eq.(D.26) is either prescribed or known from the
potential flow solution. Further, Eq.(D.26) involves three unknown quantities Cf , H and
θ . Hence, two empirical relationships involving these three quantities need to be
prescribed. Chapter 7 of Ref.D.1, 2000 edition, be consulted for details.

Presently, numerical solutions using computational methods are being employed to

calculate the boundary layers.

vi).Separation:

When the flow takes place over shapes like airfoils, Ue is not constant. When Ue is
decreasing, from Bernouli equation, the pressure is increasing. In this situation the
pressure gradient is said to be adverse. When the external velocity is increasing or the
pressure is decreasing, the situation is called flow with favorable pressure gradient.
When pressure gradient is adverse, the boundary layer is subject to deceleration due to
the effects of both viscosity and the pressure. Figure D.4 shows the changes in the
shape of the boundary layer profiles in the presence of adverse pressure gradient. At

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 10

Turbulence Prof. E.G. Tulapurkara
Chapter-3

 U 
point S, the velocity gradient   becomes zero at the wall. At further downstream
 y 
locations, the velocity near the wall is negative and there is a region of recirculating
flow. The flow is said to be separated. The pressure recovery stops and the body has
an additional drag called pressure drag.

Fig. D.4 Flow in boundary layer near point of separation

vii). For discussion on thermal boundary layers, see Schlichting and Gerston (Ref.D.1,
chapters 9 and 10).

D.3 Transition
In a laminar boundary layer, the flow variables at a point have either constant values
or the values show a definite variation with time.
 U x
However, as R x  = e  increases, it is found that the flow variables inside the
  
boundary layer show a chaotic variation with time. But, the motion is organized such
that statistical averages can be taken. Such a boundary layer is called turbulent
boundary layer.
The change over from laminar to turbulent nature is called transition and takes place
over a distance called transition length.
Dept. of Aerospace Engg., Indian Institute of Technology, Madras 11
Turbulence Prof. E.G. Tulapurkara
Chapter-3

Transition in a boundary layer can be studied as an instability phenomenon. In this

approach the flow is perturbed by giving a small disturbance and it is examined whether
the disturbance grows or is damped out. Herein, a two-dimensional incompressible flow
is considered to illustrate this approach. Since, the development of disturbance is being
examined, the governing equations for unsteady laminar flow are used. These are :

U V
+ =0 (D.26)
x y

U U U 1 p   2U  2U 
+U + V =- +  2 + 2  (D.27)
t x y ρ x  x y 

V V V 1 p   2 V  2 V 
+U + V =- +  2 + 2  (D.28)
t x y ρ y  x y 

Let, the mean flow be simplified such that, U = U y ,V=W =0, p = p x,y 

On this flow, a two-dimensional disturbance which is a function of time is imposed. It

can be represented as :
u = u  x,y,t ,v = v  x,y,t , p = p  x,y,t  (D.29)

Thus, the instantaneous motion can be written as :

U = U+u', V = v', p = p +p' (D.30)
Substituting from Eq.(D.30) in Eq.(D.27), gives :

  U+u  U+u   
t
  
U+u + U+u
x

+ v
y


1  p+p  
  2 U+u'  2 U+u' 

+ 
  
=- +  (D.31)
ρ x  x
2
y 2

 
In a linear stability analysis, the terms involving the products of fluctuating quantities
are ignored. With this simplification Eq. (D.31) reduces to:

u u U 1 p 1 p   2 U  2u  2u 

+U + v =- - +  2 + 2 + 2  (D.32)
t x y ρ x ρ x  y x y 

When a time-average of Eq. (D.32), is taken it yields:

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 12

Turbulence Prof. E.G. Tulapurkara
Chapter-3

1 p   2 U 
0=- +  2  (D.33)
ρ x  y 
Subtracting Eq. (D.33) from Eq.(D.32) and carrying out similar operations on
Eqs. (D.26) and (D.28) gives the following equations for u’, v’ and p’.
u' v'
+ =0 (D.34)
x y

u' u U 1 p'   2u'  2u' 

+U + ν =- +  2 + 2  (D.35)
t x y ρ x  x y 

v v V 1 p'   2 v  2 v 

+U +v  =- +v 2 + 2  (D.36)
t x y ρ x  x y 

To solve equations (D.34) to (D.36), u' and v' are expressed in terms of stream function
(ψ) i.e.
ψ ψ
u' = , v = - (D.37)
y x
Disturbance is assumed to be composed of waves propagating in the x-direction. This
can be expressed in terms of ψ in the following generalized manner:
ψ x,y,t = (y)ei( αx -ct )
  (D.38)

Note:
(a) (y) = r + i i
2π
(b) Wave length of the disturbance = λ = ,
α
(c) c = cr + cii , cr = wave speed , ci = amplification factor ; ci < 0 means the disturbance
is damped out ; ci > 0, means the disturbance grows or the flow is unstable.

Substitution of (D.38) in equation (D.34) to (D.36) and eliminating pressure gives the
following non-dimensional equation:
  
 +U   ψxx +ψyy  -Uyy ψx = ψyyyy +ψxxyy +ψxxxx 
1
(D.39)
 t x  Re

ψ
It may be mentioned that the suffix indicates a derivative i.e. ψ x = and
x

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 13

Turbulence Prof. E.G. Tulapurkara
Chapter-3

2 U
Uyy =
y 2
Since, the flow is assumed to be nearly parallel to the wall, gives:
1
 D2 - α2   + ( U - c) ( D2 - α2 )  - Uyy  = 0 ; D =
d
2
(D.40)
αRe dy

Equation (D.40) is known as Orr-Sommerfeld equation.

The boundary conditions for the flat plate boundary layer case are:
 (0) = 0 i.e. wall is streamline
y (0)= 0 i.e. no slip condition at wall
 y (  )= 0 i.e. flow is undisturbed far away from wall
The solution of Eq.(D.40) depends on the velocity profile and Re. One looks for the
values of  when disturbance is amplified. Chapter 15 of Ref. D.1, 2000 edition, may be
referred to for further details.

Remarks:
i) Neutral stability curves :
The curve in Fig. D.5. is called neutral stability curve. When the values of  and R are
such that the point corresponding to these values, lies on this curve, the damping factor
ci is zero; R is the value of Reynolds number based on the boundary layer height ( δ ).

When the values of  and R are such that the point lies inside the curve, the
disturbance is amplified or the flow is said to be unstable. On the other hand, when the
values of  and R are such that the point lies outside the curve, the disturbance is
damped and the flow is said to be stable. Figure D.5 schematically shows the neutral
stability curve for a boundary layer on a flat plate developing under uniform external
stream.
It is seen that all disturbances are damped out below a certain Reynolds number. This
Reynolds number is called critical Reynolds number and denoted by Rcrit.

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 14

Turbulence Prof. E.G. Tulapurkara
Chapter-3

Fig. D.5 Neutral stability curve for a flat plate boundary layer (schematic)

For a flat plate boundary layer developing in a uniform external stream, Rcrit is
5x105. Above this Reynolds number some disturbances are amplified. Thus, if the
flow is free of disturbances, Rcrit can be higher. For a commercial pipe, Rcrit, based
pipe diameter (d) and mean velocity in the pipe, may be 2000 but for a very smooth
pipe it can be as high as 40000. As regards the boundary layer, over a flat plate
developing in a uniform stream, chapter 15 of Ref.D.1, 2000 edition shows
experimental results which indicate that the flow can be laminar even at a Reynolds
number of 2.8 x 106 when the free stream turbulence level (Tu) is very low. The
quantity Tu is given by the following equation.
100 1 2 2
× k ; k =Turbulent kinetic energy  u +  v +  w 
2
Tu =
Ue 2 
ii) Length of transition region :
Actual process of transition is complex and is spread over a distance called
transition region. It consists of the following stages. These stages are based on the
description in Ref.D.2, chapter 5.
a) After the critical Reynolds number is exceeded, some disturbances grow. These
are called Tollmien Schlichting (TS) waves.
b) The TS waves are unstable and develop into three-dimensional hairpin vortices.

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 15

Turbulence Prof. E.G. Tulapurkara
Chapter-3

c) The vortices breakdown, resulting in regions of high localized shear and three-
dimensional fluctuations.
d) Turbulent spots are formed at various locations in the flow. These spots are local
regions of highly turbulent flow. The spots appear randomly.
e) The turbulent spots grow as they move downstream. Various spots coalesce and
form fully turbulent flow over the entire width of the plate.

As the transition process takes place, the boundary layer thickness grows. A pictorial
depiction the five stages can be seen in chapter 5 of Ref. D.2 or chapter 15 of Ref.
D.1, 2000 edition.

Remarks:
(i) The length of the transition region depends on turbulence level in the external
stream. Chapter 15 of Ref. D.1., 2000 edition, be referred to for details.
(ii) It was mentioned earlier that the critical Reynolds number depends on the level of
freestream turbulence (Tu). However, the critical Reynolds also depends on (a) pressure
gradient in the external flow, (b) surface roughness, (c) free stream Mach number, and
(d) the curvature of the wall. Chapter 15 of Ref.D.1, 2000 edition be referred to for
details.
(iii) There is uncertainty regarding the onset of transition and the length of transition
region. Various computer codes dealing with computation of boundary layer over
surfaces, have their own criteria for the onset of transition and the extent of transition
region. Reference D.3, chapter 4 be consulted for additional information.

References
D.1 Schlichting.H. “Boundary layer theory”, McGraw Hill, New York 1968. See also
Schlichting.H and Gersten.K.”Boundary layer theory”, Springer, Berlin 2000.
D.2 White.F.M “ Viscous fluid flow” McGraw Hill, New York 1991
D.3 Wilcox, D.C. “Turbulence modeling for CFD”, DCW industries, La. Canada, U.S.A.,
2006

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 16

Turbulence Prof. E.G. Tulapurkara
Chapter-3

Appendix E
Equation for dissipation

The isotropic dissipation is given by :

ui' ui'
εi = 
xm xm
The derivation of an equation for ε i is presented below. The aim is not to scare the

students but to demonstrate that equations for turbulent quantities can be obtained by
patiently going through the steps. The derivation begins with differentiation of
expression for ui . The equation for ui is given by Eq.(3.23) . It is reproduced below :

 Ui u  1 p  2 ui

t
ui + uk
xk
+Uk i +
xk xk
 
ui uk -ui uk = -
ρ xi
+
xk xk
(E.1)

u i'
Differentiating Eq.(E.1) with respect to xm and multiplying by 2 gives:
x m
ui'  ui' ui'  ui'
2 ( ) + 2 Uk ( )
xm t xm xm xk xm
Ui ui' uk' Uk ui' ui'
= - 2 - 2
xk xm xk xm xm xm

 2 Ui ui' ui' 2 2 ui'

-2 uk' - 2 ' '
(uu )+2 ' '
(uu )
xmxk xm xm xmxk xmxk xm
i k i k

ν ui'  2 p' u'  3u'i

-2 +2 2 i  E.2 
ρ xm xmxi xm xm xk xk

ui'  ui'  ui'  ui' ε

Now, 2 ( ) = ( ( )= i
xm t xm t xm t xm t
ui'  ui'  ui' ui' ε
2 Uk ( ) = Uk ( ) = Uk i
xm xk xm xk xm xm xk
Since k, m and i are dummy indices, the second term on r.h.s of Eq. (E.2) can be written
as:
Uk ui' ui' Ui um' um'
- 2 = - 2
xm xm xk xk xi xk
Then, the first and second term on the r.h.s of Eq.(E.2) can be written as:

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 17

Turbulence Prof. E.G. Tulapurkara
Chapter-3

ui' 2 ui'   ' ui

'
' uk 
'
- 2 (ui' uk' ) = - 2  (u +u )
xm xmxk xm  xm xk xk 
k i

The second term inside the brackets is zero from continuity. Hence, this term reduces
to:

ui' ui' uk' ui' '  2ui'

- 2 - 2 uk
xm xk xm xm xmxk
 2ui' uk'
Adding - , which is anyway zero, gives the following equation on
xmxm xk
averaging.

ui'  2 (ui' uk' ) ui' ui' uk'  ' ui

'
ui'
- 2 =- 2 - ( uk )
xm xm xk xm xk xm k xm xm

The fifth term on r.h.s of Eq.(E.2) vanishes when the averaging is done.

2 ui'  2p' 2 p'  ui'

The sixth term -  when added by  ( ) gives on averaging.
ρ xm xmxi ρ x i x m x i

2 ui'  2p' 2  p' u'i

- =- ( )
ρ xm xmxi ρ xi xm xm

The last term on r.h.s. of Eq.(E.2) when added and subtracted with
 2u'i  2u'i
2 2 becomes:
xmxk xmxk
u'i 3u'i u'i 3u'i  2u'i  2u'i  2u'i  2u'i
2 2 = 2 2 +2 2 - 2 2
xm xm xk xk xm xmxk xk xmxk xm xk xm xk xm xk
  ui' ui'  2u'i  2u'i
= ( )- 2 2
xk xk xm xm xmxk xmxk

On averaging and noting the definition of εi , gives the last term as:
ui  3ui  2εi  2ui'  2ui'
2 2 = - 2 2
xm xmxk xk xk xk xmxk xmxk

Thus, equation (E.2 ) on averaging becomes:

εi ε Ui  u'k u'i u'm u'm  u'k u'i u'i   u'i u'i 
+Uk i = -2  +  -2 -  u'k 
t xk xk  xm xm xi xk  xm xk xk xk  xm xm 

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 18

Turbulence Prof. E.G. Tulapurkara
Chapter-3

2  p' u'i  2ui  2ui  2 Ui ' u'i  2εi

- ( )- 2 2
-2 uk + (E.3)
ρ xi xm xm xm xk xmxk xk xm xm xk xk

Which is the desired equation.

Dept. of Aerospace Engg., Indian Institute of Technology, Madras 19