0

Chapter-2
External Flows

2.1 Blasius Solution

The simplest example of boundary layer flow is flow over a flat plate is shown in Fig. 2.1.

U U Free stream
8

Potential flow
y
Boundary layer
δ(x)
u(x,y)
x
L

Figure 2.1: Flat plate boundary layer flow

The governing equations are :

∂u ∂u ∂2u
u +v =ν 2 (2.1)
∂x ∂y ∂y

∂u ∂v
+ =0 (2.2)
∂x ∂y
The boundary conditions are : (a) y = 0, u = v = 0, (b) y = ∞, u = U∞
dp
Substitution of − ρ1 dx in the boundary layer momentum equation in terms of free stream
velocity produces U∞ (dU∞ /dx) which is equal to zero. Hence the governing equation
does not contain any pressure gradient term. The characteristics parameters of this
problem are u, U∞ , ν, x, y. Before we proceed further, let us discuss laws of similarity.
• u component of velocity has the property that two velocity profiles of u(x, y) at
different x locations differ only by a scale factor.

1
 • The velocity profiles u(x, y) at all values of x can be made same if they are plotted
in coordinates which have been made dimensionless with reference to the scale
factors.

• The local free stream velocity U(x) at x is an obvious scale factor for u, because the
dimensionless u(x) varies with y between zero and unity at all sections. The scale
factor for y, denoted by g(x), is proportional to local boundary layer thickness so
that y itself varies between zero and unity . Finally

u[x1 , (y/g(x1))] u[x2 , (y/g(x2))]

= (2.3)
U(x1 ) U(x2 )

Again, let us consider the statement of the problem

u = u(U∞ , ν, x, y) (2.4)

Five variables involve two dimensions. Hence it is reducible for a dimensionless relation
in terms of 3 quantities. For boundary layer equations a special similarity variable is
available and only two such quantities are needed. When this is possible, the flow field is
said to be self similar. For self similar flows the x-component of velocity has the property
that two profiles of u(x, y)/U∞ at different x locations differ only by a scale factor that
is at best a function of x.
!
u y y
=F = F p νx (2.5)
U∞ δ U∞

For Blasuis problem the similarity law is

" #
u y
= F p νx = F (η) (2.6)
U∞ U∞

where
r
y νx
η= , δ∼ (2.7)
δ U∞
or,
y
η = p νx (2.8)
U∞

Z Z r Z
νx p
ψ= udy = U∞ F (η) dη = U∞ νx F (η)dη (2.9)
U∞
or,
p
ψ= U∞ νx f (η) + C(x) (2.10)

2
 R
where f (η) = F (η)dη, and C(x) = 0 if the stream function at the solid surface is set
equal to 0.

∂ψ ∂ψ ∂η ′
u= = = U∞ f (η) (2.11)
∂y ∂η ∂y

  r
∂ψ ∂ψ ∂η 1 νU∞ h ′ i
v=− + = ηf (η) − f (η) (2.12)
∂x ∂η ∂x 2 x
Similarly,
∂u U∞ η ′′
=− f (η) (2.13)
∂x 2 x
r
∂u U∞ ′′
= U∞ f (η) (2.14)
∂y νx

∂2u 2
U∞ ′′′

2
= f (η) (2.15)
∂y νx
Substituting these terms in Eq. (2.1) and simplifying we get
′′′ ′′
2f (η) + f (η)f (η) = 0 (2.16)

The boundary conditions are :

We know that at y = 0, u = 0 and v = 0. As a consequence, we can write

′
@ η = 0, f (η) = 0 and f (η) = 0

Similarly at y = ∞, u = U∞ results in

′
@ η = ∞, f (η) = 1

Equation (2.16) is a third order nonlinear differential equation. Blasius obtained this
solution in the form of a series expanded around η = 0.
Let us assume a power series expansion of the (for small values of η)
A2 2 A3 3 A4 4
f (η) = A0 + A1 η + η + η + η + .........
2! 3! 4!
′ A3 2 A4 3
f (η) = A1 + A2 η + η + η + .........
2! 3!
′′ A4 2 A5 3
f (η) = A2 + A3 η + η + η + .........
2! 3!
′′′ A5 2 A6 3
f (η) = A3 + A4 η + η + η + ......... (2.17)
2! 3!
3
 ′
Boundary conditions @ η = 0, f (η) = 0 and @ η = 0, f (η) = 0 applied to above will
produce A0 = 0; A1 = 0

We derive another boundary condition from the physics of the problem: @ y = 0, (∂ 2 u/∂y 2 ) =
′′′
0 which leads to @ η = 0 : f (η) = 0; invoking this into above we get A3 = 0. Finally
′′ ′′′
Eq. (2.17) is substituted for f, f , f into the Blasius equation, we find
   
A5 2 A6 3 A2 2 A4 4 A5 5
2 A4 η + η + η + .... + η + η + η + .........
2! 3! 2! 4! 5!
 
A4 2 A5 3
× A2 + η + η =0
2! 3!
Collecting different powers of η and equating the corresponding coefficients equal to zero,
we obtain
A5 A22 A6
2A4 = 0, 2 + = 0, 2 =0
2! 2! 3!
A7 A2 A4 A2 A4
2 + + =0
4! 4! 2! 2!
A8 A2 A5 A2 A5
+2 + =0
5! 5! 2! 3!
This will finally yield : A4 = A6 = A7 = 0
A22 11 375 4
, A8 = A32 , A11 = −
A5 = − A
2 4 8 2
Now substituting the series for f (η) in terms of η and A2
A2 2 1 A22 5 11 A32 8 1 375 4 11
f (η) = η − η + η − A η
2! 2 5! 4 8! 8 11! 2
" #
1/3 1/3 1/3 1/3
1/3 (A2 η)2 1 (A2 η)5 11 (A2 η)8 1 375(A2 η)11
f (η) = A2 − + − + .... (2.18)
2! 2 5! 4 8! 8 11!
or,
1/3 1/3
f (η) = A2 F (A2 η)
Equation (2.18) satisfies boundary conditions at η = 0. Applying boundary conditions
at η = ∞, we have
h i
2/3 ′ 1/3 ′
limη→∞ A2 F (A2 η) = f (∞) = 1
or,
 3/2
1
A2 = (2.19)
limη→∞ F ′ (η)
The value of A2 can be determined numerically to a good degree of accuracy. Howarth
found A2 = 0.33206.

4
Numerical Approach
Let us rewrite Eq. (2.16)

′′′ ′′
2f (η) + f (η) f (η) = 0

as three first order differential equations in the following way:

′
f =G (2.20)
′
G =H (2.21)
′ 1
H = − fH (2.22)
2
′ ′
The condition f (0) = 0 [or η = 0, f (η) = 0] remains valid and f (0) = 0 [or η = 0, f (η) =
′
0] , means G(0) = 0. Finally η = ∞, f (η) = 1 gives ⇒ G(∞) = 1

Note that the equations for f and G have initial values. The value for H(0) is not known.
This is not an usual initial value problem. We handle the problem as an initial value
problem by choosing values of H(0) and solving by numerical methods f (η), G(η) and
H(η). In general G(∞) = 1 will not be satisfied for the function G arising from the
numerical solution. We then choose other initial values of H so that we find an H(0)
which results in G(∞) = 1. This method is called SHOOTING TECHNIQUE. In Eqs.
(2.20 - 2.22) the primes refer to differentiation with respect to η. The integration steps
following a Runge-Kutta method are given below
1
fn+1 = fn +(k1 + 2k2 + 2k3 + k4 ) (2.23)
6
1
Gn+1 = Gn + (l1 + 2l2 + 2l3 + l4 ) (2.24)
6
1
Hn+1 = Hn + (m1 + 2m2 + 2m3 + m4 ) (2.25)
6
as one moves from ηn to ηn+1 = ηn + h. The values of k, l and m are as follows
df ′
= f = F1 (f, G, H, η)
dη
dG ′
= G = F2 (f, G, H, η)
dη
dH ′
= H = F3 (f, G, H, η) (2.26)
dη

k1 = h F1 (fn , Gn , Hn , ηn )
l1 = h F2 (fn , Gn , Hn , ηn )
m1 = h F3 (fn , Gn , Hn , ηn ) (2.27)

5
  
1 1 1 h
k2 = h F1 fn + k1 , Gn + l1 , Hn + m1 , ηn +
2 2 2 2
 
1 1 1 h
l2 = h F2 fn + k1 , Gn + l1 , Hn + m1 , ηn +
2 2 2 2
 
1 1 1 h
m2 = h F3 fn + k1 , Gn + l1 , Hn + m1 , ηn + (2.28)
2 2 2 2

In a similar way k3 , l3 , m3 and k4 , l4 , m4 are calculated following standard formulae for

Runge-Kutta integration.

The functions F1 , F2 and F3 are G, H and −f H/2

Then at a distance ∆η from the wall, we have,

f (∆η) = f (0) + G(0)∆η

G (∆η) = G(0) + H(0)∆η
′
H (∆η) = H(0) + H (0)∆η
′ 1
H (∆η) = − f (∆η) H (∆η) (2.29)
2
′′
As it has been mentioned f (0) = H(0) is unknown. H(0) = S must be determined such
′
that the condition f (∞) = G(∞) = 1 is satisfied. The condition at infinity is usually
approximated at a finite value of η (around η = 10).

The value of H(0) now can be calculated by finding H̃(0) at which G(∞) crosses unity
(Fig. 2.2).

Refer to Fig. 2.2(b)

H̃(0) − H(0)1 H(0)2 − H(0)1

=
1 − G(∞)1 G(∞)2 − G(∞)1

repeat the process by using H̃(0) and better of two initial values H(0). Thus the correct
initial value will be determined.

Table 2.1

————————————————————————————————————-
′ ′′
η f f =G f =H ‘
————————————————————————————————————-
0 0 0 0.33206
0.2 0.00664 0.06641 0.33199
0.4 0.02656 0.13277 0.33147
0.8 0.10611 0.26471 0.32739
1.2 0.23795 0.39378 0.31659

6
 G( )

8
2

1
Initial value H(O) 2
G

G( )
8
1

Initial value H(O) 1

0 η

(a)

G( ) 2
8

G( ) 1
8

G( )
8

1 1

~
H(O)

0 H(O) 1 H(O) H(O)

2
(b)

Figure 2.2: Correction of the initial guess for H(0)

1.6 0.42032 0.51676 0.29667

2.0 0.65003 0.62977 0.26675
2.4 0.92230 0.72899 0.22809
2.8 1.23099 0.81152 0.18401
3.2 1.56911 0.87609 0.13913
3.6 1.92954 0.92333 0.09809
4.0 2.30576 0.95552 0.06424
4.4 2.69238 0.97587 0.03897
4.8 3.08534 0.98779 0.02187
5.0 3.28329 0.99155 0.01591
8.8 7.07923 1.00000 0.00000
————————————————————————————————————-

7
Shear Stress
∂u
τwall = µ |y=0
∂y
or,
∂ ′ ∂η
τwall = µU∞ f (n) |η=0
∂η ∂y
or,
1
τwall = µU∞ [0.33206] p
νx/U∞
or,
2
0.332ρU∞
τwall = √
Rex
Each time examine G(η) versus η for proper G(∞). Compare the values with Schlichting.
The values are available in Table 2.1

Local skin friction coefficient

τwall
Cf x = 1 2
2
ρU∞

0.664
Cf x = √
Rex

Total friction force per unit width

 L
Z L Z L 2 2 1/2
0.332 ρ U∞ dx 0.332 ρ U∞ x 
F = τwall dx = q √ = q
0 0 U∞ x U∞ 1
ν ν 2 0

r
2 νL
F = 0.664 ρ U∞
U∞

The average skin friction coefficient

F 1.328
Cf = 2 L
= √
1/2ρU∞ ReL
′
From the table, it is seen that f = G = 0.99 for η = 5. So, u/U∞ reaches 0.99 at η = 5.
It is possible to write
δ δ 5.0
p νx = 5; or =√
U∞
x Rex

8
2.2 Temperature distribution over a flat plate bound-
ary layer

T8
U T T U

8
Tw <T Tw >T

8
θ =1.0
1.0
1.0
T−T x
8

θ= θ
Tw −T
8

y η
0 θ = θ (x, y) θ = θ (η)

Figure 2.3: The growth of thermal boundary layer

Thermal Boundary layer equation

∂T ∂T k ∂2T
u +v = (2.30)
∂x ∂y ρcp ∂y 2
r
′ 1 νU∞ n ′
o
u = U∞ f (η), v = − f (η) − ηf (η) (2.31)
2 x

T − T∞
θ= (2.32)
Tw − T∞
Figure 2.3 describes the growth of thermal boundary layer on a heated flat plate. The
boundary condition for the situation described above is θ(0) = 1, θ(∞) = 0. Therefore,
T = T∞ + θ(Tw − T∞ )
Let us assume Tw = Tw (x)
p [ a general case]
On substituting η = (y/ Uνx∞ ), we can write different derivatives as,

∂T dTw  y  r U dθ
∞
=θ + (Tw − T∞ ) − (2.33)
∂x dx 2x νx dη

9
 r
∂T ∂T ∂η ∂θ U∞
= = (Tw − T∞ ) (2.34)
∂y ∂η ∂y ∂η νx

 
∂2T ∂ ∂T ∂η ∂ 2 θ U∞
= = (Tw − T∞ ) (2.35)
∂y 2 ∂η ∂y ∂y ∂η 2 νx

On substituting the derivatives in thermal boundary layer equation and multiplying by

x
U∞ (Tw −T∞ )
on both sides of the equation, we get
′ ′
r ′ ′′
f θx dTw f y U∞ ′ (f − ηf ) ′ k θ
− θ − θ = (2.36)
(Tw − T∞ ) dx 2 νx 2 ρcp ν

′ ′ ′ ′
′ d f ηθ (f − ηf )θ k ′′
f θx [ln(Tw − T∞ )] − − = θ (2.37)
dx 2 2 µcp
′
Dividing by f θ
′
d fθ k ′′ 1
x [ln(Tw − T∞ )] − ′ = θ (2.38)
dx 2f θ µcp f ′ θ

This equation could be put in the form

′′ ′
x(dTw /dx) kθ (η) f (η)θ (η)
= ′ + ′ =λ (2.39)
Tw (x) − T∞ µcp f (η)θ(η) 2f (η)θ(η)

This equation can be solved by using the method of separation of variables. This leads
to
x(dTw /dx)
=λ (2.40)
Tw (x) − T∞

Wall temperature has to follow this relation

dx
d {ln(Tw − T∞ )} = λ (2.41)
x
Integrating we get,

ln (Tw − T∞ ) = λ lnx + C (2.42)

So,

(Tw − T∞ ) = Cxλ

When

λ = 0, Tw = T∞ + C (constant wall temperature)

10
Again,
′′ ′
kθ (η) f (η)θ (η)
′ + ′ =λ (2.43)
µcp f (η)θ(η) 2f (η)θ(η)
or,
′′
θ (η) 1 ′ ′
+ f θ = λf θ (2.44)
Pr 2
or,
′′ Pr ′ ′
θ + f θ = λP rf θ (2.45)
2

′′ Pr ′ ′
θ + f θ − λP rf θ = 0 (2.46)
2
In this equation f is known from Blasius solution. The boundary conditions are:

y = 0, T = Tw → η = 0, θ = 1
y = ∞, T = T∞ → η = ∞, θ = 0
T = T∞ + θ(Tw − T∞ )

The final form of the boundary conditions are θ(0) = 1 and θ(∞) = 0.
′
Let Y1 = θ and Y2 = θ
then
′ Pr ′
Y2 = − f Y2 + λP rf Y1 (2.47)
2
with Y1 (0) = 1 and Y1 (∞) = 0

The numerical solution using Runge-Kutta method can be obtained via following steps
′
1. Guess a value of Y2 (0) or θ (0)

2. Solve for Y1 , Y2

3. Check if Y1 (∞) = 0, i.e., θ(∞) = 0 ?

4. If yes, stop. The calculated θ(η)’s are correct solution

′
5. If no, correct θ (0) or Y2 (0) and repeat the calculation.

11
2.3 Analytical solution, for λ = 0
′′ Pr ′
θ + fθ = 0 (2.48)
2
or,
′′
θ Pr
′ = − f (2.49)
θ 2
or,
d ′ Pr
(ln θ ) = − f
dη 2
or,
Z
′ Pr
ln θ = − f (η)dη + ln C
2
or,
′ Pr
R
θ = Ce− 2
f (η)dη

or,
Z R
Pr
θ=C e− 2
f (η)dη
dη + D

or,
Z η Rη
Pr
θ(η) − θ(0) = C e− 2 0
f (η)dη
dη
0

or, Z β Rβ
Pr
θ(η) = 1 + C (e− 2 0
f (r)dr
)dβ (2.50)
0

Now θ(∞) = 0 will result in

Z ∞ Rβ
Pr
−1 = C (e− 2 0
f (r)dr
)dβ (2.51)
0
or,
1
C = −R ∞ Rβ (2.52)
− P2r f (r)dr
0
(e 0 )dβ
Rη Pr
Rβ
0
(e− 2 0
f (r)dr
)dβ
θ(η) = 1 − R ∞ Rβ (2.53)
− P2r f (r)dr
0
(e 0 )dβ
hx
Nux = Nusselt number = k
h i
∂T ∂η
−k
−k(dT /dy)y=0 ∂η ∂y
η=0
h= = (2.54)
(Tw − T∞ ) (Tw − T∞ )

12
We know,
∂T ∂θ
T = T∞ + θ (Tw − T∞ ) ⇒ = (Tw − T∞ )
∂η ∂η
r
∂θ ∂η ′ U∞
h = −k |η=0 = −kθ (0) (2.55)
∂η ∂y νx
r
hx ′ U∞ x ′
Nux = = −θ (0) = −θ (0)(Rex )1/2 (2.56)
k ν

Nux ′

1/2
= −θ (0) ≃ 0.332 P r 1/3 (0.6 ≤ P r ≤ 10) (2.57)
(Rex )

The solution was obtained by Pohlhausen.

2.4 More Discussions on solution via numerical route

The heat transfer rate at the wall is given by
∂T
qw = −k |y=0 (2.58)
∂y
Hence,
qw ∂θ ∂θ ∂η
= − |y=0 = − |y=0 (2.59)
k(Tw − T∞ ) ∂y ∂η ∂y
i.e.,
qw x ′
p
= −θ |η=0 Rex (2.60)
k(Tw − T∞ )
i.e., p
′
Nux = −θ |η=0 Rex (2.61)
Nux and Rex , are, the local Nusselt and Reynolds numbers based on x. Because θ
′
depends only on η for a given P r, θ |η=0 depends only P r and its values can be obtained
from the solution for the variation θ with η for any value of P r. It is convenient to define.
′
A(P r) = −θ |η=0 (2.62)

In terms of this function A, eqn (2.61) can be written as

p
Nux = A Rex (2.63)

Values of A for various values of P r are shown in Table 2.2.

13
 Table 2.2
Values of A for various values of P r
————————————————————————————————–
′
Pr A = −θ |η=0 0.332P r 1/3

—————————————————————————————————

0.6 0.276 0.280

0.7 0.293 0.295
0.8 0.307 0.308
0.9 0.320 0.321
1.0 0.332 0.332
1.1 0.344 0.343
7.0 0.645 0.635
————————————————————————————————–
Over the range of Prandtl numbers covered in the table, it has been found that A varies
very nearly as P r 1/3 and, as will be seen from the results given in the table, is quite
closely represented by the approximate relation:

A = 0.332P r 1/3 (2.64)

We calculate the total heat transfer rate from a plate of length, L. Unit width of the
plate is considered because the flow is, by assumption, two-dimensional. The total heat
transfer rate per unit width Qw will, of course, be related to the local heat transfer rate,
qw by: Z L
Qw = qw dx (2.65)
0

But Eq. (2.60) gives the local heat transfer rate as

r
U∞
qw = Ak(Tw − T∞ ) (2.66)
xν
Substituting this result Eq. (2.65) then gives on carrying the integration:
r
UL
Qw = 2Ak(Tw − T∞ ) (2.67)
ν
If a mean heat transfer coefficient for the whole plate, is defined such that
Qw
h̄ = (2.68)
L(Tw − T∞ )

Then, since unit width of the plate is being considered, Eq. (2.67) gives
r
2Ak U∞ L
h̄ =
L ν

14
It is possible to calculate average Nusselt number as

h̄L p
Nu = = 2A ReL (2.69)
k
Let us revisit the thermal conditions for the wall again.
We consider the case where:

Tw = T∞ + Cxλ

i.e.,

Tw − T∞ = Cxλ (2.70)

C and λ being constants.

The heat transfer rate at the wall is as before given by

∂T
qw = −k |y=0 (2.71)
∂y

∂θ ∂η
qw = −k(Tw − T∞ ) |η=0
∂η ∂y

r
c λ U∞ ∂θ
qw = −k x . |η=0
νx ∂η

√
−k c U∞ ′
qw = √ θ (0)xλ−1/2 (2.72)
ν
For the constant heat flux along the wall, the expression should be independent of x or
in other words, we can say
1
λ= (2.73)
2

2.5 Approximate Methods for Flat Plate Boundary

Layer
h i
δ δ 1/3 δ 1/2
Flow over a heated flat plate P r ∼ δT
from δT
∼ (P r) or δT
∼ (P r) has been
illustrated in Fig. 2.4.
Governing Equations are
∂u ∂v
+ =0 (2.74)
∂x ∂y

15
 Pr<1 (δ < δ ) Pr>1 ( δ> δ )
T T

U T U T

8
8

8
8
y
y

δT δ
δT
δ
x x
Tw Tw

Figure 2.4: Velocity and Thermal Boundary Layers (a) for δ < δT (b) for δ > δT

∂u ∂v 1 dp ∂2u
u +v =− +ν 2 (2.75)
∂x ∂y ρ dx ∂y
∂T ∂T k ∂2T
u +v = (2.76)
∂x ∂y ρcp ∂y 2
The boundary conditions are:

@y = 0, u = 0 = v, T = Tw
@y = δ, u = U∞ , @y = δT , T = T∞

Integral method due to von Karman and Pohlhausen

Form of Eq. (2.75)
Z δ Z δ Z δ Z δ 2
∂u ∂u 1 dp ∂ u
u dy + v dy = − dy + ν 2 dy (2.77)
∂x ∂y ρ dx ∂y
| 0 {z } | 0 {z } | 0 {z } | 0 {z }
I II III IV
Term III is zero for flow over flat plate (because dp/dx = 0)
II term:
Z δ h i Z δ ∂u
δ ∂v
[vu]0 − udy = vu|δ − vu|0 + u dy (from continuity)
0 ∂y 0 ∂x
From continuity equation, one can write
∂u ∂v
=−
∂x ∂y
or,
Z δ Z δ
∂v ∂u
dy = − dy
0 ∂y 0 ∂x

16
or,
Z δ
∂u
v|y=δ = − dy
0 ∂x
II term finally becomes
Z δ Z δ
∂u ∂u
−U∞ dy + u dy
0 ∂x 0 ∂x
Equation (2.77) can be written as
Z δ Z δ
∂u ∂u ∂u
2 u dy − U∞ dy = −ν |y=0
0 ∂x 0 ∂x ∂y
or, Z δ Z δ
d 2 d ∂u
u dy − U∞ udy = −ν |y=0 (2.78)
dx 0 dx 0 ∂y
Equation (2.78) is Momentum Integral Equation for flow over a flat plate. Momentum
Integral Equation for a non-zero pressure gradient surface will be discussed later.
Let us assume
u = C0 + C1 y + C2 y 2 + C3 y 3
u
= C0 + C1 η + C2 η 2 + C3 η 3 , where η = y/δ
U∞
u
η is called similarity parameter; despite the growth of buoyancy layer in x direction, U∞
remains similar for same yδ at any x.
Boundary conditions:

u
at y = 0, u = 0, or, at η = 0, = 0, ⇒ C0 = 0
U∞
u
at y = δ, u = U∞ , or, at η = 1, =1
U∞
 
or, C1 η + C3 η 3 η=1 = 1, or, C1 + C3 = 1
∂2u  ∂u ∂u ∂2u 
at y = 0, 2 = 0 comes from u +v =ν 2 ,
∂y ∂x ∂y ∂y
2
∂ (u/U∞ ) = 0
or, at η = 0, = 0,
∂η 2
or, [2C2 + 6C3 η]at η=0 = 0 ⇒ C2 = 0
 
∂u ∂ u
at y = δ, = 0 or, at η = 1, =0
∂y ∂η U∞
or, [C1 + 3C3 η 2 ]η=1 = 0, ⇒ C1 + 3C3 = 0
Finally we get,
3 1 u 3 1
C1 = , C3 = − So, = η − η3
2 2 U∞ 2 2

17
Integral of the first term in Eq. (2.78)
Z δ 2 Z 1
2 U∞
u dy = [9η 2 + η 6 − 6η 4 ] δdη
0 4 0

2
  2
δU∞ 9 1 6 68 δU∞
= + − =
4 3 7 5 35 4
Integral associated with the second term of Eq. (2.78)
Z δ Z
U∞ 1 5 U∞ δ
udy = [3η − η 3 ] δdη =
0 2 0 4 2

The third term of equation (2.78) −ν ∂u |

∂y y=0
  
ν ∂ 3 1 3
=− U∞ η− η |η=0
δ ∂η 2 2

3 νU∞
=−
2 δ
Finally Eq. (2.78) becomes
2
 
U∞ dδ 68 5 3 U∞ ν
− =−
4 dx 35 2 2 δ
which reduces to
dδ 140 ν
δ =
dx 13 U∞
On integration this gives
1 13 2 νx
δ = +C (2.79)
2 140 U∞
Initial condition : at x = 0, δ = 0 gives C = 0.
Finally,
280 νx
δ2 =
13 U∞
which gives,
r
νx
δ = 4.64
U∞
U∞ x
On substituting Rex = ν
we get,
x
δ = 4.64 √ (2.80)
Rex

18
Casting the integral form of thermal boundary layer equation
Z δT Z δT Z δT
∂T ∂T k ∂2T
u dy + v dy = dy (2.81)
0 ∂x 0 ∂y 0 ρcp ∂y 2

Integrating the second term by parts and applying continuity, we get

Z δT Z δT  δ
∂T δT ∂v k ∂T T
u dy + [vT ]0 − T dy =
0 ∂x 0 ∂y ρcp ∂y 0
Z δT Z δT
∂T ∂u k ∂T
or u dy + T dy + T∞ v|δT = − |y=0
0 ∂x 0 ∂x ρcp ∂y
R δT ∂u
where v|δT = − 0 ∂x
dy
or,
Z δT   Z δT
∂T ∂u ∂u k ∂T
u +T dy − T∞ dy = − |y=0
0 ∂x ∂x o ∂x ρcp ∂y
or,
Z δT
∂ k ∂T
[u(T − T∞ )]dy = − |y=0
0 ∂x ρcp ∂y
or,
Z δT
d k ∂T
[u(T − T∞ )]dy = − |y=0 (2.82)
dx 0 ρcp ∂y

This equation is called Energy Integral Equation. In order to solve this, let us assume
the temperature profile , θ = TT∞−Tw
−Tw
= C0 + C1 ζ + C2 ζ 2 + C3 ζ 2. Where ζ = δyT is a
similarity parameter.

Applying thermal boundary conditions.

at y = 0; T = Tw , ⇒ at ζ = 0, θ = 0 : C0 = 0
∂2T d2 θ
at y = 0; 2 = 0, ⇒ at ζ = 0, 2 = 0 : C2 = 0
∂y dζ
at y = δT ; T = T∞ , ⇒ at ζ = 1, θ = 1 : C1 + C3 = 1
∂T dθ
at y = δT ; = 0, ⇒ at ζ = 1, = 0 : C1 + 3C3 = 0
∂y dζ

Solving for C1 and C3 we get C1 = 23 , and C3 = − 12 .

Therefore,
T − Tw 3 1
= ζ − ζ3 (2.83)
T∞ − Tw 2 2

19
Now,
T − Tw − T∞ + Tw T − T∞
θ−1 = =
T∞ − Tw T∞ − Tw
or,

(T − T∞ ) = (θ − 1) (T∞ − Tw )

First term of the energy integral (Eq. (2.82)) is

Z δT
d
(T∞ − Tw ) u(θ − 1)dy
dx 0

Z δT   
d 3 1 3 1 3
U∞ (T∞ − Tw ) η − η3 ζ − ζ − 1 dy
dx 0 2 2 2 2

Z δT    "    3 #
d 3 y 1  y 3 3 y 1 y
U∞ (T∞ − Tw ) − − − 1 dy
dx 0 2 δ 2 δ 2 δT 2 δT

δT
Substituting ξ = and rearranging, we get
δ
  
d 3 2 3 4 3 2 1 4 3 2 1 4
U∞ (T∞ − Tw ) δ ξ − ξ − ξ + ξ − ξ + ξ
dx 4 20 20 28 4 8

  
d 3 2 3 4
= U∞ (T∞ − Tw ) δ − ξ + ξ
dx 20 280

The RHS of energy integral equation (Eq. (2.82)) is evaluated in the following way:
"    3 #
3 y 1 y
(T − Tw ) = (T∞ − Tw ) −
2 δT 2 δT

 
∂T 3 1 3 y2
= (T∞ − Tw ) − 3
∂y 2 δT 2 δT
or,
k ∂T 3 (T∞ − Tw )
− |y=0 = −α
ρcp ∂y 2 δT

Now Eq. (2.82) becomes

  
d 3 2 3 4 3 α(T∞ − Tw )
(T∞ − Tw )U∞ δ − ξ + ξ =− (2.84)
dx 20 280 2 δT

20
 δT δ
δ δT

Pr ~
δ Pr ~ δ , Pr > 1 δ > δT
δT , Pr < 1 δT > δ δT

Figure 2.5: Different Situations with P r < 1 and P r > 1

For P r > 1, δ > δT , ξ = δδT becomes small, and so ξ 4 << ξ 2

For P r > 1 (Fig. 2.5), Eq. (2.84) becomes
 2
d ξ α
δU∞ δ =
dx 20 2ξ
or,
10α dδ d
= δ ξ 2 + δ2 ξ 2
U∞ ξ dx dx

dδ 140 ν 280 νx
where δ = , and δ 2 =
dx 13 U ∞ 13 U∞
Finally, on substitution of the above two relations, Eq. (2.84) becomes
d 2 13 1
ξ 3 + 2xξ (ξ ) = (2.85)
dx 14 P r
dχ dξ
Put χ = ξ 3 ; dx
= 3ξ 2 dx ; substituting in Eq. (2.85)

4x dχ 13 1
χ+ =
3 dx 14 P r
or,
dχ 3 13 1 3
+ χ=
dx 4x 14 P r 4x
3 dx
This linear differential equation for which the integrating factor is evaluated as e| 4 x =
e(3/4)ln x = x3/4
dχ 3 χ 3/4 13 × 3 1 −1/4
x3/4 + x = x (2.86)
dx 4 x 14 × 4 P r
or,
d 3/4 39 1 −1/4
(x χ) = x (2.87)
dx 56 P r
21
 y

δT

x
x0

Figure 2.6: The Flat Plate Situation with Unheated Initial Length

13 1
χ= + bx−3/4 (2.88)
14 P r
where b is the constant of integration.
Applying the boundary conditions as described in Fig. 2.6, we consider a short unheated
length x0 so that at x = x0 , δT = 0, ξ = 0; and so, at x = x0 , χ = 0. Substituting in
Eq. (2.88), we get

13 1 3/4
b=− x (2.89)
14 P r 0
After substituting in Eq. (2.88) for b, we get

13 1 13 1  x0 3/4
ξ3 = χ = −
14 P r 14 P r x
or,
  x 3/4 1/3
δT 1 −1/3 0
=ξ= (P r) 1−
δ 1.026 x

For x0 → 0, we finally have

δ
δT = (P r)−1/3 (2.90)
1.026
Again we know that
∂T
h(Tw − T∞ ) = −k |y=0
∂y
∂T
On substituting for | ,
∂y y=0
we get

−k (T∞δ−Tw) 3
2
h= T

(Tw − T∞ )

22
or,
3 1.026
h=k (P r)1/3 (Rex )1/2
2 4.64x
(after invoking Eqs. (2.80) and (2.90))
The above expression leads to
hx
= 0.332(Rex )1/2 (P r)1/3
k
The local Nusselt number can be expressed as

Nux = 0.332(Rex )1/2 (P r)1/3 (2.91)

RL RL
Now, the average heat transfer coefficient h̄ = 0
hx dx/ 0
dx and so the average Nusselt
number
h̄L
NuL = = 0.664(ReL )1/2 (P r)1/3 (2.92)
k
ρU∞ L
where ReL = µ

23
2.6 Viscous Dissipation Effects on Laminar Bound-
ary Layer Flow Over a Flat Plate
For the high speed flows, the p
effects due to viscous dissipation have to be considered.
′
Now, u = U f (η) and η = y/ (νx)/U , and so the viscous dissipation term
"  #
2
∂u
µ + ......
∂y

" 2 #
∂u ∂η
=µ + ......
∂η ∂y

 r !2 
′′ U
= µ U f + ......
νx

′′ U
= µU 2 (f )2
νx
The general energy equation is simplified as
  "  #
2
∂T ∂T k ∂2T µ ∂u
u +v = + + ...... (2.93)
∂x ∂y ρcp ∂y 2 ρcp ∂y

or,
"   r #
′ dTw ∂θ y U
U f θ + (Tw − T∞ ) −
dx ∂η 2x νx

r ( r )
1 νU ′ ∂θ U
− (f − ηf ) (Tw − T∞ )
2 x ∂η νx

k ∂2θ U µ 2 ′′ 2 U
= 2
(Tw − T∞ ) + U (f )
ρcp ∂η νx ρcp νx
x
Multiply by U (Tw −T∞ )
to obtain
′ ′ ′′
f θx dTw f θ θ ′′
− = + Ec(f )2 (2.94)
(Tw − T∞ ) dx 2 Pr

If we assume (dTw /dx) = 0, the General equation becomes

24
 T T T

8
y δT
δT

x
Tw Taw

Figure 2.7: Temperature profile within the boundary layer (a) for low speed flows (b) for
high speed flows

′′ ′
θ θ ′′
+ f + Ec(f )2 = 0 (2.95)
Pr 2
As defined earlier, θ = (T − T∞ )/(Tw − T∞ ) and the boundary conditions are θ(0) =
1, θ(∞) = 0 with Tw = wall temperature. If the heat flux at the wall is assumed to be zero,
i.e., qw = 0, the wall temperature becomes adiabatic wall temperature.
In other words, Taw = Adiabatic wall temperature, if the surface temperature is the
temperature of a perfectly insulated surface. As a consequence of the dissipation in the
boundary layer, even if there is no heat transfer to a body in a flow, a thermal boundary
layer forms at the body. If the surface of the body is impermeable to heat, i.e., adiabatic,
the dissipation dictates the distribution of the wall temperature is such a way that it is
above the surrounding temperature.
θ
For the temperature field at an adiabatic body, if φ ∼ Ec/2 , the general equation reduces
to the form
Ec 1 ′′ Ec f ′ ′′
φ + φ + Ec(f )2 = 0 (2.96)
2 Pr 2 2
or,
′′ ′
φ fθ ′′
+ + 2(f )2 = 0
Pr 2
′
with the adiabatic boundary conditions φ (0) = 0, and φ(∞) = 0.
now,

θ T − T∞ 2cp (Tw − T∞ ) T − T∞
φ∼ = 2
= 2
Ec/2 Tw − T∞ U U /2cp

Thus we get,

U2
T = T∞ + φ(η) (2.97)
2cp

25
which becomes,
U2
Taw = T∞ + φ(0) (2.98)
2cp
Taw can be determined if φ(0) is known, where φ(0) is also called recovery factor, r and
U2
Taw = T∞ + r
2cp
It is found from experiments,
√
φ(0) = fn(P r), φ(0) = Pr (see Fig. 2.8)

For the low speed flows, φ(0) = r = 0 as no kinetic energy is recovered as thermal energy.

Pr
Φ

Figure 2.8: φ as a function of P r and η

For ideal gases: i.e., i = e + pv = e + RT

or,
di de
= +R
dT dT
or,

cp = cv + R

which gives on simplification,

γR
cp =
γ−1
Now,
 
rU 2 (γ − 1)
Taw = T∞ 1 +
2γRT∞
i.e.,
 
r(γ − 1) 2
Taw = T∞ 1+ M∞ (2.99)
2

26
 √
For air, P r = 0.72, φ(0) = 0.72 = 0.85 = r. Putting γ = 1.4, we get,
2
Taw = T∞ [1 + 0.17M∞ ]

Now for T∞ = 400K, and M∞ = 0.1, we obtain Taw = 400.68K. Similarly for
M∞ = 0.8, Taw = 443.52K; for M∞ = 3.0, Taw = 1012K.

Consider the general solution for

′′
θ 1 ′ ′′
+ f θ + Ec(f )2 = 0
Pr 2
with η = 0, θ = 1 and η = ∞, θ = 0

Now, let us propose,

Ec
θ (η) ∼ φ (η)
2
at η = 0, we get
Ec
θ (0) ∼ φ (0)
2
We can write
 
Ec Ec
[θ (η) − θ (0)] ∼ φ (η) − φ (0)
2 2

 
Ec Ec
θ (η) ∼ 1 − φ (0) + φ (η) (2.100)
2 2

If the viscous dissipation is neglected, the dimensionless temperature can be defined as θ̃

Now the proportionality (2.100) can be written as

 
Ec 1
θ = 1− φ(0) θ̃ + Ec φ(η) (2.101)
2 2
The derivatives terms can be written as
 
′′ Ec ′′ 1 ′′
θ = 1− φ(0) θ̃ + Ec φ (η) (2.102)
2 2
 
′ Ec ′ 1 ′
θ = 1− φ(0) θ̃ + Ec φ (η) (2.103)
2 2
On substitution in the general equation we get
  ′′    ′′ 
Ec θ̃ 1 Ec ′ 1 φ (η) f ′ ′′
1− φ(0) + f 1− φ(0) θ̃ + Ec + φ (η) + Ec(f )2 = 0
2 Pr 2 2 2 Pr 2

27
or,
 " ′′
#  ′′ 
Ec θ̃ 1 ′ Ec ′′ 2 φ f ′
1− φ(0) + f θ̃ + 2(f ) + + φ =0
2 Pr 2 2 Pr 2

We know
′′
φ f ′ ′′
+ φ + 2(f )2 = 0 (2.96, rewritten)
Pr 2
′
With the adiabatic condition φ (0) = 0 and φ(∞) = 0. The above equation with the
boundary conditions can be solved by Shooting Technique.

After simplification and reduction we get,

′′
θ̃ 1 ′
+ f θ̃ = 0 (2.104)
Pr 2
Now we try for the General Solution of the homogeneous part

Known conditions are θ(0) = 1 and θ(∞) = 0. Let us revisit the relationship between θ
and θ̃,using equation (2.101). We obtain the following:

θ(0) = 1 means θ̃(0) = 1; and θ(∞) = 0 means θ̃(∞) = 0 [since φ(∞) = 0]

The solution of this equation that results finally is of the form

Rη R
Pr η


exp − f dη dη
θ̃(η) = 1 − R ∞0 2R 0
Pr η

(2.105)
0
exp − 2 0 f (η)dη dη
Now the heat transfer for the wall
 
∂T ∂θ  ∂θ ∂η 
qw = −k = −k(Tw − T∞ ) = −k(Tw − T∞ )
∂y ∂y y=0 ∂η ∂y η=0
or,
r
U ′
qw = −k(Tw − T∞ ) θ (0) (2.106)
νx
  ˜′
Following the adiabatic condition, φ (0) = 0, we get θ (0) = 1 − Ec
′ ′
2
φ(0) θ (0). Substi-
′
tuting θ (0) and Ec, we get
r  
U U2 ′
qw = −k(Tw − T∞ ) 1− φ(0) θ̃ (0)
νx 2cp (Tw − T∞ )
After rearranging and replacing appropriately using value of Taw from before, we get
r  
U Tw − Taw ′
qw = −k(Tw − T∞ ) θ̃ (0)
νx Tw − T∞

28
or,
qw x ′
p
= −θ̃ Rex = Nux (2.107)
(Tw − Taw ) k

This expression is same as for the case without viscous dissipation, but the only difference
is that here we have (Tw − Taw ) in place of (Tw − T∞ ) As a result, there will be occurrence
of heat transfer reversal.

2.7 More about similarity solution (Velocity Bound-

ary Layer)
A systematic study of flow past a wedge shaped body (Fig. 2.9) suggests
u ′
η = Ayxa and = f (η) (2.108)
U
1 −a
ψ is obtained by integration as ψ = A
x [Uf (η)]. The momentum equation for a flat
plate boundary layer reduces to

y
x
β

Figure 2.9: Flow past a wedge

νA2 2a ′′′ a ′′
x f − ff = 0 (2.109)
U x
p
Choosing a = −1/2 cancels x throughout the equation and choosing A = U/ν keeps
the equation free of the flow and fluid parameters.
Similarity solution can be obtained for a wider class of problems where U = Cxm . This

U=C

Figure 2.10: Flow past a flat plate

29
form of U represents flow past wedge shaped surface as shown earlier (Fig. 2.9). The
β/π
relationship between m and β is m = 2−β/π . For β = 0 and m = 0 and the flat plate
problem is recovered (Fig. 2.10). For β = π, m = 1 and this is stagnation point flow
(Fig. 2.11). The pressure field is p(x) = β − 21 ρU 2 (x), hence the pressure gradient

U=Cx

Figure 2.11: Stagnation point flow

dp dU m−1 ρU 2 m
= −ρU = −ρUCmx =− (2.110)
dx dx x
and the x momentum equation becomes

∂u ∂u mU 2 ∂2u
u +v = + ν 2 , U = Cxm
∂x ∂y x ∂y
Note that x and y do not refer to Cartesian coordinates any more. Instead they are
boundary layerpcoordinates, parallel and perpendicular√to the solid surface. The similarly
variable η = y U/νx and the stream function ψ = νxU f (η) reduce the momentum
equation to the two point boundary value problem

′′′ (m + 1) ′′ ′
f + f f + m[1 − f 2 ] = 0 (2.111)
2
′ ′
with f (0) = f (0) = 0, f (∞) = 1
The equation for f with m as a wedge parameter is called as the Falkner-Skan equation.
Similarity solution can be obtained for a wide class of problem where U = Cxm . This
form of U represents flow past wedge shaped surfaces. The relationship between β and
m is given by
β
m= = wedge parameter
2π − β

m(2π − β) = β

m < 0 signifies adverse pressure gradient. Angle β is negative and U is not a constant
for wedge flow (Fig. 2.12).

30
 U

Figure 2.12: Flow over a wedge that may cause flow separation

2.7.1 Derivation of Falkner-Skan Equation

We have U = C xm , dUdx
= C m xm−1
x dU
Therefore U dx = U C m xm−1 = C xxm C m xm−1 = m
x

Also we have
η = yh(x), ψ(x, η) = g(x)f (η) (2.112)

∂u ∂u 1 ∂p ∂2u
u +v =− +ν 2
∂x ∂y ρ ∂x ∂y
∂p
Here − 1ρ ∂x
becomes U dU
dx
for wedge flows. We know that
 
∂ψ  ∂ψ 
u= , v=−
∂y x ∂x y

Again, if (x, y) = fn (χ, η), the transformation is

∂ ∂ ∂χ ∂ ∂η
= · + ·
∂x ∂χ ∂x ∂η ∂x
∂ ∂ ∂χ ∂ ∂η
= · + ·
∂y ∂χ ∂y ∂η ∂y
If (x, y) = fn (x, η), the von Misses transformation is
          
∂  ∂  ∂x ∂  ∂η ∂  ∂  ∂η
= + = + (2.113)
∂x y ∂x η ∂x y ∂η x ∂x y ∂x η ∂η x ∂x y
and,
       
∂  ∂  ∂η ∂  ∂x  ∂  ∂η 
= + = (2.114)
∂y x ∂η x ∂y x ∂x η ∂y x ∂η x ∂y x
Substituting, we get
 
∂ψ  ∂  ∂[y h(x)] ′
u= = [g(x) f (η)] =f g h
∂y x ∂η x ∂y

31
  "    #
∂ψ  ∂  ∂η  ∂(g f ) 
 = −[g f ′ y h′ + f g ′ ]
v= −  =− [g(x) f (η)]  +
∂x y ∂η x ∂x y
 ∂x η

Choose g(x)h(x) = U(x)

  
∂u ∂u  ∂u  ∂η 
= +
∂x ∂x η ∂η x ∂x y

 
∂ ′  ∂ ′  ′
= [f gh] + [f gh] yh
∂x η ∂η x

′ ′ ′ ′′ ′
= f (gh + hg ) + f ghyh

   
∂u ∂  ∂η  ∂u  ∂η 
= =
∂y ∂η x ∂y x ∂η x ∂y x


∂ ′  ′′
= (f gh) h = f gh2
∂η x


∂2u ∂ ′′

2  ∂η 
 ′′′ ′′′

2
= [f gh ] = h f gh2 = gf h3
∂y ∂η x ∂y x


Consider the x-momentum equation

∂u ∂u dU ∂2u
u +v =U +ν 2
∂x ∂y dx ∂y

′ ′′ ′ ′ ′ ′ ′ ′ ′ ′ ′′
f gh(f ghh y + f gh + hf g ) − [f g + gf yh ](f gh2 ) =
′ ′ ′′′
gh(g h + gh ) + νgf h3 (2.115)

After simplification, we get

′ ′ ′
′′′[g h + gh ] ′ g ′′
f + 2
(1 − f 2 ) + ff = 0 (2.116)
νh νh
As f is a function of η alone, both coefficients should be constants

′ ′ ′
gh + hg g
= C1 and = C2
νh2 νh

1 dU
= C1
νh2 dx
32
or,
 
U 1 x dU
= C1
x νh2 U dx
or,
U
m = C1
xνh2
or,
r
U m
h=
νx C1
Now choosing C1 = m, the expression for small h reduces to that of flat plate, that is
r
U
h=
νx
Again,
′
g
= C2
νh
or,
r r
dg U νU √
= C2 ν = C2 = C2 ν C xm−1
dx νx x
or,
q √
dg m−1
= C22 C ν (x) 2
dx
After integrating, we get

C2 √
g= νUx
(m + 1)/2
√
Choosing C2 = m+1 , we obtain g = νUx
2 √
The result is compatible, because ψ = g(x)f (η) = νUxf (η)
The final resulting equation becomes
′′′ ′ ′′
f + C1 (1 − f 2 ) + C2 f f = 0
or,
′′′ m + 1 ′′ ′
f + m(1 − f 2 ) + ff = 0 (2.117)
2
This equation is called Falkner-Skan Equation. This equation can be solved as three
initial value problems for which the boundary conditions are as follows:
′
u(0) = 0 ⇒ f (0) = 0
′
u(∞) = U ⇒ f (∞) = 1
v(0) = 0 ⇒ f (0) = 0 (2.118)
The Eq. (2.117) can be solved using SHOOTING TECHNIQUE. The solutions for dif-
ferent values of m has been shown in Fig. 2.13.

33
2.8 More on Similarity Solution of Energy equation

m=1
m=0
f ’(η)

m<0

η
separation

Figure 2.13: Variation of velocity profile for different values of m

The two-dimensional energy equation is given as

 2
∂T ∂T k ∂2T µ ∂u
u +v = + (2.119)
∂x ∂y ρcp ∂y 2 ρcp ∂y
For similarity solution of this equation put
T − T∞
θ= = θ(η)
Tw − T∞
q
U
√
; ψ = νUxf (η); Uu = f (η), or, u = f gh
′ ′
It is known that, η = y νx
Also,
   #  
∂ψ  ∂ψ  ∂ψ ∂η √ 1 −1/2 dU √ 1 −1/2 √ ′ ∂η
v=− =− + =− νx U f + νU x f + νUx f
∂x y ∂x η ∂η ∂x 2 dx 2 ∂x

"r r #
νU f f νx dU √ ′ dη
=− + + νUx f
x 2 2 U dx dx

r r
1 νU ′ f νx dU
=− (f − ηf ) −
2 x 2 U dx
Finally, the energy equation becomes
" r # ( r r )
∂T ∂θ  y  U 1 νU f νx dU
′ w ′
Uf θ + (Tw − T∞ ) − + − (f − ηf ) −
∂x ∂η 2x νx 2 x 2 U dx

34
 ( r )
∂θ U k U ′′ µ U 3 ′′ 2
(Tw − T∞ ) = (Tw − T∞ ) θ + (f )
∂η νx ρcp νx ρcp νx

or,
 
k U ′′ ′ f U dU ′ dTw U3 ′′
(Tw − T∞ ) θ +θ (Tw − T∞ ) + − Uf θ+ (f )2 = 0
ρcp νx 2 x dx dx cp x

Finally it reduces to the form

′′
θ f ′ ′ (dTw /dx) ′′
+ (1 + m) θ − xf θ + Ec (f )2 = 0 (2.120)
Pr 2 (Tw − T∞ )

It can be solved by the Method of Separation of Variables, and so

 ′′ 
1 θ f ′ ′′ 2 x(dTw /dx)
′ + (1 + m) θ + Ec(f ) = =λ (2.121)
f θ Pr 2 (Tw − T∞ )

The equation
′′
θ f ′ ′′ ′
+ (1 + m) θ + Ec (f )2 − λf θ = 0 (2.122)
Pr 2
is required to be solved using Shooting Technique, for different values of Pr, Ec, m and λ.

The equation
′′
θ f ′ ′′ ′
+ (1 + m) θ + Ec (f )2 − λf θ = 0 (2.123)
Pr 2
is required to be solved using Shooting Technique, for different values of Pr, Ec, m and
λ.
For m = 0, we have flat plate problem; λ = 0 leads to constant wall temperature case
and Ec = 0 leads to the case without viscous dissipation. For this case, the simplified
version of equation becomes
′′
θ 1 ′
+ fθ = 0 (2.124)
Pr 2
with θ(0) = 1 and θ(∞) = 0. For this special case, λ = 0 ⇒ Tw = constant

Then the simplified equation becomes

′′ Pr ′
θ + fθ = 0 (2.125)
2
For P r = 1 the system becomes θ + 21 f θ = 0 and f + 12 f f = 0;
′′ ′ ′′′ ′′

′ ′
The boundary conditions are f (0) = 0, f (∞) = 1 and f (0) = 0
We can solve the energy equation using f + 21 f f = 0
′′′ ′′

35
Let us say B̃ = f , then B̃ + 12 f B̃ = 0 with the boundary conditions, B̃(0) =
′ ′′ ′

0 and B̃(∞) = 1
Now the energy equation is
′′ 1 ′
θ + fθ = 0 with θ(0) = 1, θ(∞) = 0
2
Put B = (1 − θ), the equation becomes
′′ 1 ′
−B − f B = 0 with B(0) = 0, B(∞) = 1
2
which gives B̃ = B or, θ = (1 − B). The solution is plotted in Fig. 2.14.

f’ = u/U
f’(η)
θ (η)
θ

Figure 2.14: Variation of velocity profile and temperature profile

2.9 Approximate Method for Boundary Layer Flows

over Non-zero Pressure Gradient Surfaces
Consider the case of a steady, two-dimensional and incompressible flow over a curved
surface. Upon integrating the dimensional form of boundary layer equation from y =
0(wall) to y = δ (where δ signifies the interface of the free stream and the boundary
layer), we obtain

Z δ   Z δ  
∂u ∂u 1 dp ∂2u
u +v dy = − + ν 2 dy (2.126)
0 ∂x ∂y 0 ρ dx ∂y
or
Z δ Z δ Z δ Z δ
∂u ∂u 1 dp ∂2u
u dy + v dy = − dy + ν dy (2.127)
0 ∂x 0 ∂y 0 ρ dx 0 ∂y 2
The second term of the left-hand side can be expanded as

Z δ Z δ
∂u ∂v
ν dy = [νu]δ0 − u dy (2.128)
0 ∂y 0 ∂y

36
or
Z δ Z δ  
∂u ∂u ∂u ∂v
ν dy = U vδ + u dy since =− (2.129)
0 ∂y 0 ∂x ∂x ∂y
or
Z δ Z δ Z δ
∂u ∂u ∂u
ν dy = −U dy + u dy (2.130)
0 ∂y 0 ∂x 0 ∂x
the equation can be re-constituted as

Z δ Z δ Z δ
∂u ∂u 1 dp ∂u
2u dy − U dy = − dy − ν |y=0 (2.131)
0 ∂x 0 ∂x 0 ρ dx ∂y
dp
Substituting the relation between dx
and the free stream velocity U for the inviscid zone
in Eq. (2.131) we get

!
µ ∂u
Z Z Z
δ
∂u δ
∂u δ
dU |
∂y y=0
2u dy − U dy − U dy = − (2.132)
0 ∂x 0 ∂x 0 dx ρ
or
Z δ 
∂u ∂u dU τw
2u −U −U dy = − (2.133)
0 ∂x ∂x dx ρ
which is reduced to
Z δ Z δ
∂ dU τw
[u(U − u)]dy + (U − u)dy = (2.134)
0 ∂x dx 0 ρ
Since the integrals vanish outside the boundary layer, we are allowed to put δ = ∞.
Z ∞ Z
∂ dU ∞ τw
[u(U − u)]dy + (U − u)dy = (2.135)
0 ∂x dx 0 ρ
or
Z ∞ Z ∞
d dU τw
[u(U − u)]dy + (U − u)dy = (2.136)
dx 0 dx 0 ρ
Substituting the definitions of displacement thickness and momentum thickness, we ob-
tain
d
dU τw
U 2 δ ∗∗ + δ ∗ U = (2.137)
dx dx ρ
Equation (2.136) is known as the momentum integral equation for two-dimensional in-
compressible laminar boundary layer over a non-zero pressure gradient surface. The same
remains valid for turbulent boundary layers as well. Needless to say, the wall shear stress

37


(τw ) will be different for laminar and turbulent flows. The term U dU
dx
signifies space
wise acceleration of the free stream. Existence of this term means the presence of free
stream pressure gradient in the flow direction.
Equation (2.136) can be further arranged as
dδ ∗∗ dU τw
U2 + (2δ ∗∗ + δ ∗ )U = (2.138)
dx dx ρ
The velocity profile of the boundary layer is considered to be a fourth-order polynomial
in terms of the dimensionless distance η = y/δ, and is expressed as

u/U = αη + bη 2 + cη 3 + dη 4 (2.139)

The boundary conditions are

ν ∂2u 1 dp
at η = 0 : u = 0, v = 0 and δ2 ∂η2
= ρ dx
= −U dU
dx
∂u ∂2u
at η = 1 : u = U, and ∂η
0,= ∂η2
=0
A dimensionless quantity, known as shape factor, is introduced as

δ 2 dU
Λ= (2.140)
ν dx
The following relations are obtained
Λ Λ Λ Λ
a=2+ , b = − , c = −2 + , d = 1 − (2.141)
6 2 2 6
Now, the velocity profile can be expressed as

u/U = F (η) + ΛG(η), (2.142)

where
1
F (η) = 2η − 2η 2 + η 4 , G(η) = η(1 − η)3 (2.143)
6
The shear stress τw = µ(∂u/∂y)y=0 is given by

τw δ Λ
=2+ (2.144)
µU 6
We use the following dimensionless parameters,
τw δ ∗∗
L= (2.145)
µU

(δ ∗∗ )2 dU
K= (2.146)
ν dx

H = δ ∗ /δ ∗∗ (2.147)

38
The integrated momentum Eq. (2.137) reduces to
∗∗
2 dδ dU τw
U + (2δ ∗∗ )U =
dx dx ρ
or
 
dδ ∗ ∗∗ δ ∗ dU τw
U +δ 2 + ∗∗ =
dx δ dx ρU
∗∗
The term τw /ρU can be written as τw /ρU = τµU
wδ
· ν
δ∗∗
= νL
δ∗∗
From the above two exprassions, one can write
dδ ∗∗ dU νL
U + δ ∗∗ (2 + H) = ∗∗ (2.148)
dx dx δ
or
U dδ ∗∗ (δ ∗∗ )2 dU
δ ∗∗ + (H + 2) = L
ν dx ν dx
or
U dδ ∗∗
2δ ∗∗ = 2[L − K(H + 2)]
ν dx
or
 
d (δ ∗∗ )2
U = 2[L − K(H + 2)] (2.149)
dx ν
The parameter L is related to the skin friction and K is linked to the pressure gradient.
If we take K as the independent variable, L and H can be shown to be the function of K
since
Z δ
∗
Uδ = (U − u)dy
0

or,
Z 1
∗
Uδ = δ (U − u)dη
0

or,
Z 1
δ∗  u
= 1− dη
δ 0 U

Z 1
δ∗ 3 Λ
= [1 − F (η) − ΛG(η)] dη = − (2.150)
δ 0 10 120

Z 1
δ ∗∗
= (F (η) + ΛG(η)(1 − F (η) − ΛG(η)) dη (2.151)
δ 0

39
 37 Λ Λ2
= − − (2.152)
315 945 9072

 2
[δ ∗∗ ]2 37 Λ Λ2
K= 2 Λ=Λ − − (2.153)
δ 315 945 9072

Therefore,
    
Λ δ ∗∗ Λ 37 Λ Λ2
L= 2+ = 2+ − − = f1 (K) (2.154)
6 δ 6 315 945 9072

δ∗ (3/10) − (Λ/120)
H= = = f2 (K) (2.155)
δ ∗∗ (37/315) − (Λ/945) − (Λ2 /9072)

The right-hand side of Eq. (2.149) is thus a function of K alone. Walz pointed out that
this function can be approximated with a good degree of accuracy by a linear function
of K so that

2[L − K(H + 2)] = a − bK

Equation (2.149) can now be written as

 
d U(δ ∗∗ )2 bU(δ ∗∗ )2 1 dU U(δ ∗∗ )2 1 dU
=a− + (2.156)
dx ν ν U dx ν U dx
or
 
d U(δ ∗∗ )2 U[δ ∗∗ ]2 1 dU
= a − (b − 1) (2.157)
dx ν ν U dx

Solution of this differential equation for the dependent variable (U[δ ∗∗ ]2 /ν) subject to
the boundary condition U = 0 when x = 0, gives
Z x
U(δ ∗∗ )2 a
= b−1 U b−1 dx (2.158)
ν U 0

With a = 0.47 and b = 6, the approximation is particularly close between the stagnation
point and the point of maximum velocity. Finally, the value of the dependent variable is
Z
∗∗ 2 0.47ν x 5
(δ ) = U dx (2.159)
U6 0
at any x. If U is a complicated function of x or known only point wise, the integral can
be evaluated numerically, say by Simpson’s rule. Once δ ∗∗ is known, other quantities can
be evaluated. Alternatively by taking the limit of Eq. (2.159), it can be shown that

(δ ∗∗ )2 |x=0 = 0.47ν/6U ′ (0) (2.160)

40
This corresponds to K = 0.0783. It may be mentioned that [δ ∗∗ ] is not equal to zero at
the stagnation point. If([δ ∗∗ ]2 /ν) is determined from Eq. (2.160), K(x) can be obtained
from Eq. (2.146). Table 2.3 gives the necessary parameters for obtaining results, such as
velocity profile and shear stress τw . The approximate method can be applied successfully
to a wide range of boundary layer flows over a non-zero pressure gradient surface.

As mentioned earlier, K and Λ are related to the pressure gradient and the shape factor.
Introduction of K and Λ in the integral analysis enables extension of Karman Pohlhausen
method for solving flows over curved geometry. However, the analysis is not valid for the
geometries, where Λ < −12 and Λ > +12.

41
 Table 2.3 Auxiliary functions after Holstein and Bohlen

————————————————————————————————————-
Λ K f1 (K) f2 (K)
————————————————————————————————————-
12 0.0948 2.250 0.356
10 0.0919 2.260 0.351
8 0.0831 2.289 0.340
7.6 0.0807 2.297 0.337
7.2 0.0781 2.305 0.333
7.0 0.0767 2.309 0.331
6.6 0.0737 2.318 0.338
6.2 0.0706 2.328 0.324
5.0 0.0599 2.361 0.310
3.0 0.0385 2.427 0.283
1.0 0.0135 2.508 0.252
0 0 2.554 0.235
-1 -0.0140 2.604 0.217
-3 -0.0429 2.716 0.179
-5 -0.0720 2.847 0.140
-7 -0.0999 2.999 0.100
-9 -0.1254 3.176 0.059
-11 -0.1474 3.383 0.019
-12 -0.1567 3.500 0
————————————————————————————————————-

2.10 Effect of Pressure Gradient on External Flows

For the case of a boundary layer on a flat plate, the pressure gradient of the external
stream is zero. Let us consider a body with curved surface (Fig. 2.15). Upstream of
the highest point the streamlines of the outer flow converge, resulting in an increase of
the free stream velocity U(x) and a consequent fall of pressure with x. Downstream of
the highest point the streamlines diverge, resulting in a decrease of U(x) and a rise in
pressure. In this section we shall investigate the effect of such a pressure gradient on the
shape of the boundary layer profile u(x, y). The boundary layer equation is
∂u ∂u 1 ∂p ∂2u
u +v =− +ν 2
∂x ∂y ρ ∂x ∂y
where the pressure gradient is found from the external velocity field as dp/dx = −ρU(dU/dx),
with x taken along the surface of the body. At the wall, the boundary layer equation
becomes
 2 
∂ u ∂p
µ 2
= (2.161)
∂y w ∂x

42
 dp dp
dx <0 dx >0
U(x)

Inflection
δ

U
8

∂2u ∂2u
( ) ∂y2
<0
wall
( )
∂y2
>0
wall

Figure 2.15: Velocity profile associated with separation in a cross flow over a circular
cylinder

In an accelerating stream dp/dx < 0 (see Fig. 2.15) and therefore

 2 
∂ u
< 0 (accelerating) (2.162)
∂y 2 wall

As the velocity profile has to merge smoothly with the external profile, the slope ∂u/∂y
slightly below the edge of the boundary layer decreases with y from a positive value to
zero; therefore, ∂ 2 u/∂y 2 slightly below the boundary layer edge is negative. Eq. (2.162)
then shows that ∂ 2 u/∂y 2 has the same sign at both the wall and the boundary layer
edge, and presumably throughout the boundary layer (Fig. 2.16). In contrast, for a
decelerating external stream, the curvature of the velocity profile at the wall is
 2 
∂ u
> 0 (decelerating) (2.163)
∂y 2 wall
so that the curvature changes sign somewhere within the boundary layer. In other words,
the boundary layer profile in a decelerating flow (dp/dx > 0) has a point of inflection
where ∂ 2 u/∂y 2 = 0 (Fig. 2.17).
The shape of the velocity profiles in the figures suggests that an adverse pressure gradient
tends to increase the thickness of the boundary layer. This can also be seen from the
continuity equation.

Z y
∂u
v(y) = − dy (2.164)
0 ∂x
Compared to a flat plate, a decelerating external stream causes a larger −∂u/∂x within
the boundary layer because the deceleration of the outer flow adds to the viscous decel-
eration within the boundary layer. From the above equation we observe that the v-field,

43
 y y y

∂u ∂ 2u
u
∂y ∂ y2
u (∂ u/∂ y) (∂2u/ ∂ y2)
dp
Figure 2.16: Profiles for dx
< 0 (favorable pressure gradient)

directed away from the surface, is larger for a decelerating flow. The boundary layer
therefore thickens not only by viscous diffusion but also by advection away from the sur-
face, resulting in a rapid increase in the boundary layer thickness with x. If pressure falls
along the direction of flow, dp/dx < 0 and we say that the pressure gradient is ”favor-
able”. If, on the other hand, the pressure rises along the directions of flow, dp/dx > 0 and
we say that the pressure gradient is ”adverse”. The rapid growth of the boundary layer
thickness in a decelerating stream, and the associated large v-field, causes the important
phenomena of separation, in which the external stream ceases to flow nearly parallel to
the boundary surface.

y y y

PI
u ∂u ∂2 u
∂y ∂ y2
u (∂ u / ∂ y) (∂2 u / ∂ y2)
dp
Figure 2.17: Profiles for dx
> 0 (adverse pressure gradient)

2.11 Description of Flow past a Circular Cylinder

Let us start with a consideration of the creeping flow around a circular cylinder, charac-
terized by Re < 1. (Here we shall define Re = U∞ d/ν, based on the upstream velocity
and the cylinder diameter.) Vorticity is generated close to the surface because of the no-
slip boundary condition. In the Stokes approximation this vorticity is simply diffused,
not advected, which results in a fore and aft symmetry.

44
As Re is increased beyond 1, the vorticity is increasingly confined behind the cylinder
because of advection. For Re > 4 two small attached or ”standing” eddies appear be-
hind the cylinder. The wake is completely laminar and the vortices act like ”rollers”
over which the main stream flows (Fig. 2.18). The eddies get larger as Re in increased.
A very interesting sequence of events begins to develop when the Reynolds number is

Re<4 4<Re<40

80<Re<200
Laminar boundary layer Turbulent boundary layer
Turbulent
81 ~ 125
wake Turbulent
Wake

5 5
Re<3×10 Re>3×10

Figure 2.18: Influence of Reynolds number on the wake of a circular cylinder

increased beyond 40, at which point the wake behind the cylinder becomes unstable.The
eddies start to oscillate in time and asymmetry is brought about in the wake structure
as such, the wake develops a slow oscillation in which the velocity is periodic in time,
with the amplitude of the oscillation increasing downstream. The oscillating wake starts
shedding two staggered rows of vortices with opposite sense of rotation. Karman inves-
tigated the phenomena and concluded that a nonstaggered row of vortices is unstable,
and a staggered row is stable only if the ratio of lateral distance between the vortices to
their longitudinal distance is 0.28. The staggered row of vortices behind a blunt body is
called a von Karman vortex street. The Karman vortex street is amazingly stable and it
is possible to see a large number of vortices downstream of the body. The vortices move
downstream at a speed smaller than the upstream velocity U. At this stage periodicity
is induced in the entire flow field due to the vortex shedding phenomena. During the
vortex shedding process, when an eddy on one side is shed, that on the other side forms,
resulting in an unsteady flow near the cylinder. As vortices of opposite circulations are
shed off alternately from the two sides, the circulation around the cylinder changes sign,
resulting in an oscillating ”lift” or lateral force. If the frequency of vortex shedding is
close to the natural frequency of some mode of vibration of the cylinder body, then an

45
appreciable lateral vibration takes place. Engineering structures such as chimney stacks,
suspension bridges and oil drilling platforms are designed so as to break up a coherent
shedding of vortices from the cylindrical structures. This is done by using helical strakes
on the cylinder surface, which break up the spanwise coherence of shedding, structure.

The passage of regular vortices causes velocity in the wake to have a dominant periodicity.
The frequency f is expressed as a nondimensional parameter known as the Strouhal
number, named after V. Strouhal, a German engineer. The Strouhal number is defined
as
fd
S≡
U
For a circular cylinder the value of S remains close to 0.21 for a large range of Reynolds
numbers. For small values of cylinder diameter and moderate values of U, the resulting
frequencies of the vortex shedding and oscillating lift lie in the acoustic range. For ex-
ample, at U = 10 m/ s and a wire diameter of 2 mm, the frequency corresponding to a
Strouhal number of 0.21 is f = 1050 cycles per second and the corresponding Reynilds
number is 1200.

Strouhal did his first measurement in 1878. About years later, Roshko in 1954, mea-
sured the frequencies using a hot-wire probe. For low Reynolds number laminar region,
Roshko consolidated his results to an equation of the form

S = 0.212(1 − 21.2/Re) (2.165)

At about Reynolds number of 500, multiple frequencies start showing up and the wake
tends to become turbulent.

2.12 Experimental Results for Circular Cylinder Flow

In order to understand the heat transfer results for the case of flow over a heated circular
cylinder, refer to Fig. 2.19. Consider the results for ReD < 105 (Figure 2.19). Starting
at the stagnation point, Nuθ decreases with increasing θ as a result of laminar boundary
layer development. However, a minimum is reached at θ = 80o . At this point separation
occurs, and Nuθ increases with θ because of the mixing associated with the vortex forma-
tion in the wake. In contrast, for ReD > 105 the variation of Nuθ with θ characterized by
two minima. The decline in Nuθ from the value at the stagnation point is again due to
laminar boundary layer development, but the sharp increase that occurs between 80o and
100o is now due to boundary layer transition to turbulence. With further development of
the turbulent boundary layer, Nuθ again begins to decline. Eventually turbulent bound-
ary layer separation occurs at θ ≈ 140o and than Nuθ increase as a result of considerable
mixing associated with wake region.
Over all average condition
hD
NuD = = CRem P r 1/3 (due to R. Hilpert) (2.166)
k
46
 800

700

600

500

Nuθ 400

300

200

100

0
0 30 60 90 120 150 180
θ

Figure 2.19: Local Nusselt number distribution for flow over a circular cylinder

Recently developed, more general correlation:

 
m P r∞
NuD = CRe (by A. Zhukauskas) (2.167)
P rw

Table 2.3
Re C m
1 - 40 0.75 0.4
40 - 1000 0.51 0.5
103 − 2 × 105 0.26 0.6
2 × 106 - 106 0.076 0.7

• all properties are evaluated at T∞ , except P rw , which is evaluated at Tw .

• C and m are given in Table 2.3

• for P r ≤ 10, n = 0.37 and for P r > 10 → n = 0.36

47
2.13 Some Other Important Correlations
For small Prandtl number fluids (liquid metals) the local heat transfer coefficient on a
flat plate is given by

Nux = 0.565 P e1/2

x , for P ex ≤ 0.01 (2.168)

From experiments (see Schlichting) the local friction coefficients for turbulent flow over
a flat plate is given by

Cf x = 0.0592 Rex−1/5 5 × 105 < Rex < 107 (2.169)

To a reasonable approximation, the velocity boundary layer thickness for turbulent flow
over a flat plate may be expressed as
δ 0.37
= (2.170)
x (Rex )−1/5

(Observe that δ varies as x4/5 in contrast to x1/2 for laminar flow)

The local Nusselt number for turbulent flow over a flat plate is:

Nux = 0.0296 Re4/5

x P r 1/3 (2.171)

In mixed boundary layer situation the average heat transfer coefficient for the flow over

Turbulent
Uα , Tα
Laminar

y xc
L
x

Figure 2.20: Transitional flow over a flat plate

a flat plate (Fig. 2.20) is given by

 Z xc Z L 
1
hL = hlam dx + hturb dx (2.172)
L 0 xc

or,
"  1/2 Z  4/5 Z #
xc L
k U∞ dx U∞ dx
hL = 0.332 1/2
+ 0.0296 P r 1/3
L ν 0 x ν xc x1/5

48
or,
h  i
4/5
NuL = 0.664 Re1/2
x,c + 0.037 ReL − Re4/5
x,c P r 1/3

or,
 
4/5
NuL = 0.037 ReL − A P r 1/3 (2.173)

If the typical, transition Reynolds number is Rex,c = 5 × 105

 
4/5
NuL = 0.037 ReL − 871 P r 1/3 (2.174)

Similarly,
0.074 A
C f,L = 1/5
− (2.175)
ReL ReL

where A = 1742

If L >> xc (ReL >> Rex,c )

4/5 −1/5
NuL = 0.037 ReL P r 1/3 and C f,L = 0.074 ReL

We know that the fluid properties vary with temperature across the boundary layer and
that this variation can certainly influence heat transfer. This influence may be handled
by one of the following two ways:
In one method all the properties evaluated at a mean boundary layer temperature Tf ,
termed the FILM TEMPERATURE
Tw + T∞
Tf =
2
In the alternate method, evaluate all the properties at T∞ and multiply the right hand
side of the expression for Nusselt number by an additional parameter. This parameter is
commonly of the form (P r∞ /P rw )r or (µ∞ /µw )r where w designates properties at surface
temperature.

49
 Excercise
1. Blasius Equation for flow over a flat plate must be iterated to find the correct value
′′
of f (0) which casus f (∞) to equal 1.0. Use Runge-Kutta method in combination with
any
proot finding method to find (u/U∞ ) as a function of η. The parameter
η = y/ νx/U∞ . Having solved velocity boundary layer equation, solve the thermal
boundary layer equation
Pr ′
θ′′ + f θ − λP rf ′θ = 0
2
The plate is at a constant temperature Tw and the free stream temperature is T∞ . The
non-dimensional temperature is given by θ = (T − T∞ )/(Tw − T∞ ). Find θ versus η for
different value of Pr. The available boundary condition are: at η = 0, θ = 1 and at
η = ∞, θ = 0. The correct value of θ′ (0) is to be obtained in order to find variation of θ
with η. Write a computer code that is capable of solving the problem for any given
value of Prandtl number of the flowing fluid.
2. For the case of flow over a flat heated plate, the thermal boundary layer equation is
 2 
∂T ∂T k ∂ T
u +ν =
∂x ∂y ρcp ∂y 2

with
r r
′ 1 νU∞ T − T∞ νx
u = U∞ f (η), ν = − [f (η) − ηf ′ (η)] , θ = , η = y/
2 x TW − T∞ U∞

In this problem wall temperature is a function of x while the non-dimensional

temperature field and the velocity field are the functions of similarity parameter η. On
substitution of the velocity and temperatures in terms of f and θ in the governing
equations, separate the variable groups that are functions of x and η. Show that the
thermal boundary layer equation can be written as
Pr ′
θ′′ + f θ − λP rf ′θ = 0
2
3. In the above mentioned problem (Problem 3), after having applied separation of
variables technique, consider the part that is function of x . We get (Tw − T∞ ) = Cxλ
Start with the definition of heat flux at the wall and find out the condition on λ that
makes the condition of constant wall heat flux on the wall. Use any relation that you
have used in Problem 3.
4. The thermal boundary layer equation for constant wall temperature, using similarity
variables, can be expressed as
Pr ′
θ′′ + fθ = 0
2
The boundary conditions are θ(0) = 1 and θ(∞) = 0. Find the solution as θ(η) using
analytical technique.

50
 5. For flow over a wedge-shaped body, the governing equations for mass, momentum
and energy are given by
∂u ∂ν
+ =0
∂x ∂y
∂u ∂u 1 ∂p ∂2µ
u +ν = +ν 2
∂x ∂y p ∂x ∂y
2
 2
∂u ∂T ∂ T µ ∂u
u +ν =α 2 +
∂x ∂y ∂y ρcp ∂y

The similarity variables are:

r
u′ T − T∞ νx
θ= andη = y/
U Tw − T∞ U
The solutions for the main stream velocity are given by
√
µ = U f ′ (η) and Ψ = νux f (n)
Z  r 
νx
W ith f (n) = F (η)dη and F (η) = F y/
U
The variables x and y indicate the flow and the normal direction respectively. The
temperatures Tw = Tw (x) and T∞ indicate the surface temperature and the free-stream
temperature. Show that the final form of the energy equation is
θ′′ 1
+ f θ′ (1 + m) − λf ′ θ + Ec(f ′′ )2 = 0
Pr 2
The parameter m is related to the wedge angle. Also find out the values of λ for the
constant wall temperature and constant wall heat flux conditions.
6. Explain the following:
(a) For a 2-D high speed flow over a flat plate, the viscous dissipation effect can be
 2
summarized as addition of the term µ ∂u ∂y
on the right hand side of the energy
equation,
 
∂T ∂T ∂2T
ρcp u +ν =k 2
∂x ∂y ∂y

(b)For the high speed flow with viscous heating over a flat plate, the wall heat flux is
given by
r  
U∞ Ec
qw = −k(Tw − T∞ ) 1− φ(0) θ′ (0)
νx 2
Where Ec is the Eckert number, k is is the thermal conductinvity of the fluid. Tw And
T∞ are the wall temperature and free stream temperature respectively,U∞ is the free
stream velocity, ν is the kinematic viscosity, x is the distance from the leading edge,

51
 q
U∞
η=y νx
and the nondimensional temperature is

 
Ec 1
θ = 1− φ(0) θ̃ + Ec φ (η)
2 2

The expression for the local Nusselt number, is similar to that of the flow without
viscous heating except the reference temperature for calculating the Nux is T aw where
T aw is the adiabatic wall temperature.
7. Wieghardt derived the Integral equation for Mechanical Energy for the laminar
boundary layers. The starting point of this equation was to multiply the equation of
motion by u and then integrate from y = 0 to y = h (h is greater than the local
boundary layer thickness). Can you obtain the equation by yourself with these hints?
Similar to momentum thickness, you may require defining another measure of the
boundary layer thickness, known as dissipation energy thickness given by
Z ∞
3


U δ3 = u U 2 − u2 dy
0

8. Consider the thermal boundary layer equation for flow over a flat plate as
   2 
∂T ∂T ∂ T
ρcp u +ν =k
∂x ∂y ∂y 2

The plate temperature is Tw (uniform) and the incoming fluid temperature is T∞ . Find
the expression for Energy Integral Equation (thermal). Explain how the partial
derivative of x is written as the ordinary derivative in the final form of Energy Integral
Equation.
9. For the flow over a non-zero pressure gradient surface, the boundary layer equation
retains the pressure gradient term. The pressure gradient can be expressed as the
function of the free stream acceleration (spatial). Show that the Momentum Integral
equation can be expressed in terms of momentum and displacement thicknesses as

d dU τw
(U 2 δ ∗∗ ) + δ ∗ U =
dx dx ρ

52