Introduction to Convection:

Flow and Thermal Considerations

Chapter Six and Appendix E

Sections 6.1 through 6.8
and E.1 through E.3
 Boundary Layer Similarity
Similarity Considerations

• As applied to the boundary layers, the principle of similarity is based on

determining similarity parameters that facilitate application of results obtained
for a surface experiencing one set of conditions to geometrically similar surfaces
experiencing different conditions. (Recall how introduction of the similarity
parameters Bi and Fo permitted generalization of results for transient, one-
dimensional conduction).
• Dependent boundary layer variables of interest are:
τ s and q′′ or h

• For a prescribed geometry, the corresponding independent variables are:

Geometrical: Size (L), Location (x,y)
Hydrodynamic: Velocity (V)
Fluid Properties:
Hydrodynamic: ρ , μ
Thermal : cp , k
Hence,
u = f ( x, y , L, V , ρ , μ )
τ s = f ( x, L,V , ρ , μ )
Similarity Considerations (cont.)

and
T = f ( x, y , L, V , ρ , μ , c p , k , Ts , T∞ )
h = f ( x, L, V , ρ , μ , c p , k , Ts , T∞ )

• Key similarity parameters may be inferred by non-dimensionalizing the momentum

and energy equations.
• Recast the boundary layer equations by introducing dimensionless forms of the
independent and dependent variables.
x y
x* ≡ y* ≡
L L
u v
u* ≡ v* ≡
V V
T − Ts
T* ≡
T∞ − Ts
• Neglecting viscous dissipation, the following normalized forms of the x-momentum
and energy equations are obtained:
* *
∂u ∂u 1 dp∞ ∂ 2u * ∂u * ∂u dp* 1 ∂ 2u*
u +v =− +ν 2 u *
+v *
=− * +
∂x ∂y ρ dx ∂y ∂x ∂y dx ReL ∂y *2
2
∂T ∂T ∂ 2T ν  ∂u  ∂T *
* * ∂T
*
1 ∂ 2T *
u +v =α 2 +   u +v =
∂x ∂y ∂y c p  ∂y  ∂x* ∂y * ReL Pr ∂y *2
Similarity Considerations (cont.)

ρVL VL
ReL ≡ = → the Reynolds Number
μ v
Inertia/viscous force
cpμ v Momentum/thermal diffusion
Pr ≡ = → the Prandtl Number
k α
How may the Reynolds and Prandtl numbers be interpreted physically?

• For a prescribed geometry,

(
u* = f x* , y * , ReL )
 μV
*
∂u  ∂u
τs = μ =  *
∂y y =0  L  ∂y y* =0

The dimensionless shear stress, or local friction coefficient, is then

τs
2 ∂u* ∂u*
Cf ≡ =
ρV / 2 ReL ∂y *
2
∂y *
(
= f x* , ReL )
y* = 0 y* = 0

2
Cf =
ReL
(
f x* , ReL )
What is the functional dependence of the average friction coefficient?
Only dependent on non-dimensional space coordinate and ReL, so, once you have a relation of the Cf,
the relation Cf could be universally applicable to different fluids, different V, different L.
Similarity Considerations (cont.)

• For a prescribed geometry,

(
T * = f x* , y * , ReL , Pr )
− k f ∂T / ∂y k f ( T∞ − Ts ) ∂T * k f ∂T *
y =0
h= =− =+
Ts − T∞ L ( Ts − T∞ ) ∂y * y* = 0
L ∂y * y* = 0

The dimensionless local convection coefficient is then

hL ∂T *
Nu ≡
kf
= *
∂y
(
= f x* , ReL , Pr )
y* = 0

hL convection
Nu → local Nusselt number Nu ≡ =
k f conduction

What is the functional dependence of the average Nusselt number?

How does the Nusselt number differ from the Biot number?
hL Rth ,cond
Bi ≡ =
ks Rth ,conv
Reynolds Analogy

The Reynolds Analogy

• Equivalence of dimensionless momentum and energy equations for
negligible pressure gradient (dp*/dx*~0) and Pr~1:

*∂u* * ∂u
*
1 ∂ 2u*
u +v = ∂u* * ∂u
*
dp* 1 ∂ 2u*
∂x* ∂y * Re ∂y *2 u*
+v =− * +
∂x* ∂y * dx ReL ∂y *2
Advection terms Diffusion
*∂T * * ∂T
*
1 ∂ 2T *
u *
+v *
=
∂T * *
1 ∂ 2T * ∂x ∂y ReL Pr ∂y *2
* * ∂T
u +v =
∂x* ∂y * Re ∂y *2

• Hence, for equivalent boundary conditions, the solutions are of the same form:

u* = T *
τs
2 ∂u* ∂u* ∂T *
Cf ≡ = = *
∂y * ∂y hL ∂T *
ρV 2 / 2 ReL ∂y * y* = 0 y* = 0 y* = 0 Nu ≡
kf
= *
∂y
(
= f x* , ReL , Pr )
Re y* =0
Cf = Nu
2
Reynolds Analogy (cont.)

or, with the Stanton number defined as,

h Nu
St ≡ =
ρVc p Re Pr

With Pr = 1, the Reynolds analogy, which relates important parameters of the velocity
and thermal boundary layers, is
Re Cf
Cf = Nu = St @ Pr = 1
2 2

• Modified Reynolds (Chilton-Colburn) Analogy:

– An empirical result that extends applicability of the Reynolds analogy:
Cf 2
= St Pr 3 ≡ jH 0.6 < Pr < 60
2
Colburn j factor for heat transfer

– Applicable to laminar flow if dp*/dx* ~ 0.

– Generally applicable to turbulent flow without restriction on dp*/dx*.

 Ex. 6-5

Q1: determine the heat flux to the balde at x* if it’s temperature is reduced to Ts=700 C.
Q2: determine the heat flux at the same x* with chord length L=80 mm, at V=80 m/s
Problem: Turbine Blade Scaling

Problem 6.28: Determination of heat transfer rate for prescribed

turbine blade operating conditions from wind tunnel data
obtained for a geometrically similar but smaller
blade. The blade surface area may be assumed to be
directly proportional to its characteristic length ( As ∝ L ) .

ASSUMPTIONS: (1) Steady-state conditions, (2) Constant properties, (3) Surface area A is
directly proportional to characteristic length L, (4) Negligible radiation, (5) Blade shapes are
geometrically similar.

ANALYSIS: For a prescribed geometry,

hL
Nu = = f ( ReL , Pr ) .
k
Problem: Turbine Blade Scaling (cont.)

The Reynolds numbers for the blades are

ReL,1 = (V1L1 / ν1 ) = 15 m2 /s ν1 ReL,2 = (V2 L2 / ν 2 ) = 15 m 2 /s ν 2 .

Hence, with constant properties ( v1 = v2 ) , ReL,1 = ReL,2. Also, Pr1 = Pr2 .

Therefore,
Nu 2 = Nu 1
( h2 L2 / k2 ) = ( h1L1 / k1 )
L L1 q1
h2 = 1 h1 =
L2 (
L2 A1 Ts,1 − T∞ )
The heat rate for the second blade is then
(
Ts,2 − T∞ )
( L A
q2 = h2 A2 Ts,2 − T∞ = 1 2 ) q1
L2 A1 Ts,1 − T∞( )
Ts,2 − T∞ ( 400 − 35) 1500 W
q2 = q1 = ( )
Ts,1 − T∞ ( 300 − 35)
q2 = 2066 W. <
COMMENTS: (i) The variation in ν from Case 1 to Case 2 would cause ReL,2 to differ from
ReL,1. However, for air and the prescribed temperatures, this non-constant property effect is
small. (ii) If the Reynolds numbers were not equal ( ReL,1 ≠ ReL2 ) , knowledge of the specific form of
( )
f ReL, Pr would be needed to determine h2.
End of lecture