Chapter 9: Convection in Turbulent Channel Flow

9.1 Introduction
• Laminar channel flow was discussed in Chapter 6; many features of turbulent flow are
similar
• Chapter begins with the criteria for fully developed velocity and temperature profiles
• Chapter Focus: Analysis of fully developed flows
• Analysis is limited to the following general boundary conditions:
• (i) uniform surface temperature
• (ii) uniform surface heat flux

9.2 Entry Length

• Criteria for entry length was discussed in Chapter 6
• As a rule of thumb, fully developed velocity and temperature profiles exist for
Lh L
≈ t ≈ 10 (6.7)
De De
where De is the hydraulic or equivalent diameter

4Af
De =
P
where A f is the flow area and P is the wetted perimeter

• Eq. (6.7) is recommended for Pr = 1 fluids

• More elaborate correlations exist, especially for hydrodymanic entry length; the
following approximations are recommended:
• From White
Lh
≈ 4.4 Re1D/e6 (9.1)
De
• From Latzko
Lh
≈ 0.623Re1D/e4 (9.2)
De
• Thermal entry length doesn’t lend itself to a simple, universally-applicable equation
since the flow is influenced by fluid properties and boundary conditions
• Hydrodynamic entry length is much shorter for turbulent flow than for laminar, so much
so that sometimes it’s neglected from analysis
• Thermal entry length is often important
• Analysis of heat transfer in the thermal entry length is complicated and is not covered in
the text

9.3 Governing Equations

• Figure 9.1 shows a circular pipe with the velocity in the x-direction is labeled as u
• Assumptions:
• Two-dimensional
• Axisymmetric
• Incompressible flow

9.3.1 Conservation Equations

• After Reynolds-averaging, conservation of mass reduces to:
∂u 1 ∂
+ ( rv r ) = 0 (9.3)
∂x r ∂r
• Using the same conditions, the Reynolds-averaged x-momentum equation reduces to:
∂u ∂v 1 ∂p 1 ∂ ⎡ ∂u ⎤
u + vr r = − + ⎢ r (ν + ε ) (9.4)
ρ ∂x r ∂r ∂r ⎥⎦
M
∂x ∂r ⎣
• Conservation of energy becomes:
∂T ∂T 1 ∂ ⎡ ∂T ⎤
u
∂x
+ vr =
∂r r ∂r ⎢r (α + ε H ) ∂r ⎥ (9.5)
⎣ ⎦

9.3.2 Apparent Shear Stress and Heat Flux

• The apparent shear stress and heat flux are defined similarly to that of the flat plate
development:
τ app ∂u
= (ν + ε M ) (9.6)
ρ ∂r
q ′app
′ ∂T
= −(α + ε H ) (9.7)
ρc p ∂r

9.3.3 Mean Velocity and Temperature

• Mean velocity and bulk, or mean, temperature are used in correlations for predicting
friction and heat transfer in duct flow
• Mean velocity is calculated by evaluating the mass flow rate in the duct
 ro

m = ρu m A =
∫ρ 0
u (2πr )dr

• Assuming constant density, this becomes

ro ro

∫ ∫
1 2
um = u (2π r )dr = u rdr (9.8)
π ro2 ro2
0 0

• The mean temperature in the duct is evaluated by integrating the total energy of the flow:
ro

mc p Tm =
∫
0
c p T u (2πr )dr

• Substituting (9.8) for the mass flow rate, and assuming constant specific heat,
ro

∫
0
T u rdr

Tm ≡
ro

∫ 0
u rdr

• This can be simplified by substituting the mean velocity, equation (9.8):

ro

∫
2
Tm = T u r dr (9.9)
u m ro2
0

9.4 Universal Velocity Profile

9.4.1 Results from Flat Plate Flow

• It was shown that universal velocity profile in a pipe is very similar to that over a flat
plate (see Fig. 8.13), especially when the plate is exposed to a zero or favorable pressure
gradient
• A pipe flow friction factor model was used to analyze flow over a flat plate using the
momentum integral method (see Section 8.4.3)
• The characteristics of the flow near the wall of a pipe are not influenced greatly by the
curvature of the wall of the radius of the pipe, so, invoking the two-layer model that we
used to model flow over a flat plate:
• Viscous sublayer
 u+ = y+ (8.54)
• Law of the Wall
1
u+ = ln y + + B (8.58)
κ
• Continuous wall law models by Spalding and Reichardt were applied to pipe flow
and discussed in Section 8.4.2
• For pipe flow, the wall coordinates are a little different than for flat plate flow, so the y-
coordinate is
y = ro − r (9.10)
• This yields the y+ wall coordinate:

+ (ro − r )u *
y = r0+ −r = +
(9.11)
ν
• The velocity wall coordinate is the same as before:
u
u+ ≡ (8.49)
u*
• The friction velocity is the same
u* ≡ τ o / ρ (8.46)

• The friction factor is based on the mean flow velocity instead of the free stream velocity:
τo
Cf = (9.12)
(1 / 2) ρ u m2
• Therefore, the friction velocity can be expressed as:
u* = um C f / 2

9.4.2 Development in Cylindrical Coordinates

• Since the velocity profile data for pipe flow matches that of flat plate flow, it allowed us
to develop expressions for universal velocity profiles solely from flat plate (Cartesian)
coordinates
• Developing expressions for universal velocity profiles using cylindrical coordinates is
developed after revealing important issues and insights
• Assuming fully-developed flow, the left side of the x-momentum equation (9.4) goes to
zero, leaving
1 ∂ ⎛ rτ ⎞ 1 ∂p
⎜ ⎟⎟ = (9.13)
r ∂r ⎜⎝ ρ ⎠ ρ ∂x
• Rearranging and integrating yields an expression for shear stress anywhere in the flow
r ∂p
τ (r ) = +C (9.14)
2 ∂x
• The constant C is zero since the velocity gradient (and hence the shear stress) is expected
to go to zero at r = 0
• Evaluating (9.14) at r and ro and taking the ratio of the two gives
τ (r ) r
= (9.15)
τo ro
• Equation (9.15) shows that the local shear is a linear function of radial location and
raises the following important issue:
• Equation (9.15) is a linear shear profile depicted in Fig. 9.2
• This expression contradicts how the fluid is expected to behave near the wall (recall
the Couette Flow assumption led us to the idea that τ is approximately constant in the
direction normal to the wall for the flat plate)
• Yet, in Section 8.4 that the universal velocity profile that resulted from this
assumption works well for flat plate flow as well as pipe flow
• This is reconciled as follows
• The near-wall region over which we make the Couette flow assumption covers a
very small distance
• Assume that, in that small region vary close to the wall of the pipe, the shear is
nearly constant: τ ≈ τ 0
• The Couette assumption approximates the behavior near the pipe wall as
∂u τ o
(ν + ε M ) = = constant (9.16)
∂r ρ
• Experimental data shows that near the wall, velocity profiles for flat plate and pipe flow
are essentially the same and suggests that the near-wall behavior is not influenced by the
outer flow, or even the curvature of the wall (see Fig. 8.13)

9.4.3 Velocity Profile for the Entire Pipe

• From the previous development the velocity gradient (and the shear stress) is expected to
be zero at the centerline of the pipe; none of the universal velocity profiles developed
thus far behave this way
• Reichardt attempted to account for the entire region of the pipe by suggesting the
following model for eddy viscosity:
 r ⎞⎡
2⎤
ε M κy + ⎛ ⎛r ⎞
= ⎜⎜1 + ⎟⎟ ⎢1 + 2⎜⎜ ⎟⎟ ⎥ (9.17)
ν 6 ⎝ ro ⎠ ⎢ ⎝ ro ⎠ ⎥
⎣ ⎦
• This leads to the following expression for the velocity profile

1 ⎡ 1.5(1 + r / ro ) ⎤
u+ = ln ⎢ y + ⎥+B (9.18)
κ ⎣⎢ 1 + 2(r / ro ) 2 ⎦⎥

where Reichardt used κ = 0.40 and B = 5.5

• The slope of this equation is zero at r = 0 , so the behavior matches that at the core
of the pipe
• The viscous sublayer is not accounted for
• As r → r0 , equation (9.18) reduces to the Law of the Wall form, eq. (8.56)

9.5 Friction Factor for Pipe Flow

9.5.1 Blasius Correlation for Smooth Pipe

• Blasius developed a purely empirical correlation for flow through a smooth pipe using
dimensional analysis and experimental data:
C f ≈ 0.0791 Re D−1 / 4 , (4000 < Re D < 10 5 ) (9.19)

where the friction factor is based on the mean flow velocity C f ≡ τ o /(1 / 2) ρu m2

• Though less accurate and versatile than later correlations, this lead to the 1/7th Power
Law velocity profile

9.5.2 The 1/7th Power Law Velocity Profile

• A crude but simple approximation for the velocity profile in a circular pipe was
discovered by Prandtl and von Kármán, leading from the Blasius correlation
• Formulation
• Recasting the Blasius correlation in terms of wall shear stress
−1 / 4
τo ⎛ 2r u ⎞
= 0.0791⎜ o m ⎟
1
2
ρu m2 ⎝ ν ⎠

• Then rearranging:
τ o = 0.03326 ρ u m7 / 4 ro−1 / 4ν 1 / 4 (a)
• Assume a power law can be used to approximate the velocity profile:
 q
u ⎛ y ⎞
= ⎜⎜ ⎟⎟ (b)
u CL ⎝ ro ⎠
with u CL representing the centerline velocity
• Assume that the mean velocity in the flow can be related to the centerline velocity as
u CL = (const )u m (c)
• Substituting (b) and (c) for the mean velocity in (a) yields

⎡ ⎛y −1 / q ⎤ 7 / 4
⎞
τ o = (const ) ρ ⎢ u ⎜⎜ ⎟⎟ ⎥ ro−1/ 4ν 1 / 4
⎢⎣ ⎝ ro ⎠ ⎥⎦

• This is simplified:
τ o = (const ) ρu 7 / 4 y ( −7 / 4 q ) ro( 7 / 4 q −1 / 4)ν 1 / 4 (d)
• Prandtl and von Kármán argued that the wall shear stress is not a function of the size
of the pipe, so the exponent on r0 should be zero
• Setting the exponent to zero, the value of q must be equal to 1/7, leading to the
classic 1/7th power law velocity profile:
1/ 7
u ⎛ y⎞
= ⎜⎜ ⎟⎟ (9.20)
uCL ⎝ ro ⎠

• Experimental data shows that this profile adequately models velocity profile through
a large portion of the pipe and is frequently used in models for momentum and heat
transfer (recall Section 8.4.3)
• Limitations
• Accurate for a narrow range of Reynolds numbers: (104 to 106)
• Yields an infinite velocity gradient at the wall
• Does not yield a gradient of zero at the centerline
• Nikuradse (a student of Prandtl’s) measured velocity profiles in smooth pipe over a wide
range of Reynolds numbers, and reported that the exponent varied with Reynolds
number:
n
u ⎛ y ⎞
= ⎜⎜ ⎟⎟ (9.21)
u CL ⎝ ro ⎠
• Table 9.1 lists Nikuradse’s measurements, including measurement of pipe friction factor
of the form
C
Cf = (9.22)
Re1D/ m
• Nikuradse’s results show that the velocity profile becomes fuller as the mean velocity
increases

9.5.3 Prandtl’s Law for Smooth Pipe

• A more theoretical model for friction factor is developed by employing the universal
velocity profile
• Formulation
• Beginning with the Law of the Wall, equation (8.58), substituting the wall
coordinates u + and y + , as well as the friction velocity u * = τ o / ρ = u m C f / 2
yields

2 1 ⎛ yu Cf ⎞
= ln⎜ m ⎟+B
u
(9.23)
um Cf κ ⎜ ν 2 ⎟
⎝ ⎠
• Assume that the equation holds at any value of y; evaluate the expression at the
centerline of the duct, y = ro = D / 2 , where u = u CL , by substituting these and the
Reynolds number:

u CL 2 1 ⎛ Re Cf ⎞
= ln⎜ D ⎟+B (9.24)
um Cf κ ⎜⎝ 2 2 ⎟
⎠
• By looking at the Law of the Wall, one can obtain a functional relationship for the
friction factor, though unfortunately a relationship for u CL / u m is unknown
• To evaluate the mean velocity, u m , substitute the velocity profile (8.58) into an
expression for the mean velocity, equation (9.10):
ro ro

∫ ∫
1 2
um = u (2π r )dr = u (ro − y )dy (9.25)
π ro2 ro2
0 0

where y = ro − r
• Performing the integration, the mean velocity becomes:
⎡ 1 ⎛ r u* ⎞ 3 ⎤
u m = u ⎢ ln⎜ o ⎟ + B −
*
⎥ (9.26)
⎢⎣ κ ⎜⎝ ν ⎟⎠ 2κ ⎥
⎦
• Making substitutions again

C f ⎡ 1 ⎛ Re D Cf ⎞ ⎤
um = um ⎢ ln⎜ ⎟+B− 3 ⎥ (9.27)
2 ⎢κ ⎜ 2 2 ⎟ 2κ ⎥
⎣ ⎝ ⎠ ⎦
 • u m cancels out of the equation; however u CL does not appear either, so the above
expression can be used directly to find an expression for C f by rearranging and
substituting κ = 0.40 and B = 5.0
1
Cf /2
(
= 2.44 ln Re D C f / 2 − 0.349)
• The expression is not yet complete, as it was assumed that the Law of the Wall is
accurate everywhere and ignores the presence of a viscous sublayer or wake region
• The constants are adjusted to fit the experimental data, yielding
1
Cf /2
(
= 2.46 ln C f / 2 D C f / 2 + 0.29,) (Re D > 4000 ) (9.28)

• This is Prandtl’s universal law of friction for smooth pipes; it is also known as the
Kármán-Nikuradse equation
• Despite the empiricism of using a curve fit to obtain the constants in (9.28), using a
more theoretical basis to develop the function has given the result a wider range of
applicability than Blasius’s correlation
• Equation (9.28) must be solved iteratively for C f

• A simpler, empirical relation that closely matches Prandtl’s is

Cf
2
≈ 0.023Re D−1 / 5 , (3 × 10 4
< Re D < 10 6 ) (9.29)

• This equation is also suitable for non-circular ducts with the Reynolds number
calculated using the hydraulic diameter

9.5.4 Effect of Surface Roughness

• Roughness shifts the universal velocity profile downward (see Fig. 8.16 and Section
8.4.4)
• The velocity profile in the logarithmic layer can be written as
1
u+ = ln y + + B − ΔB
κ
where ΔB is the shift in the curve, which increases with wall roughness k +
• The behavior also depends on the type of roughness, which ranges from uniform
geometries like rivets to random structures like sandblasted metal
• Equivalent sand-based roughness model:

1 ⎡ Re D f 1 / 2 ⎤
≈ 2.0 log10 ⎢ 1/ 2 ⎥
− 0 .8 (9.30)
f 1/ 2 ⎣1 + 0 . 1( k / D ) Re D f ⎦
 where it is common to use the Darcy friction factor
f = 4C f (9.31)

• If the relative roughness k / D is low enough, it doesn’t have much of an effect on

the equation
• Scaling shows that roughness is not important if (k / D ) Re D < 10
• If (k / D) Re D > 1000 , the roughness term dominates in the denominator, and the
Reynolds number cancels
• Friction is no longer dependent on the Re D
• Colebrook and White developed the following formula for commercial pipes:

1 ⎛k/D 2.51 ⎞
1/ 2
= −2.0 log10 ⎜⎜ + 1/ 2
⎟
⎟ (9.32)
f ⎝ 3.7 Re D f ⎠
• Representative roughness values presented in Table 9.2
• This function appears in the classic Moody chart (Fig. 9.3)

9.6 Momentum-Heat Transfer Analogies

• Analogy method is applied to pipe flow to the case of constant heat flux boundary
condition
• Though an analogy cannot be made for the case of a constant surface temperature,
resulting models approximately hold for this case as well
• Formulation
• For hydrodynamically fully developed flow the x-momentum equation (9.4) becomes
1 dp 1 ∂ ⎡ ∂u ⎤
=
ρ dx r ∂r ⎢r (ν + ε M ) ∂r ⎥ (9.33a)
⎣ ⎦
• The energy equation reduces to:
∂T 1 ∂ ⎡ ∂T ⎤
u = ⎢ r (α + ε H ) (9.33b)
∂x r ∂r ⎣ ∂r ⎥⎦
• Recall that an analogy is possible if the momentum and energy equations are
identical, so
• Note that in pipe flow the pressure gradient is non-zero, although constant with
respect to x.
• To ensure an analogy, the left side of (9.33b) must then be constant
• For thermally fully developed flow and a constant heat flux at the wall, the shape
of the temperature profile is constant with respect to x, leading to
 ∂T
= constant
∂x
• Analogy appears to be possible; however, boundary conditions must also match
• Boundary conditions are:
At r = 0 :
du ( 0) ∂T (0)
= 0, =0 (9.34a)
dr ∂r
At r = r0 :

u (ro ) = 0, T (ro ) = Ts ( x) (9.34b)

du (ro ) ∂T (ro )
μ =τo, k = q o′′ (9.34c)
dr ∂r
where qo′′ is assumed to be into the flow (in the negative r-direction)
• By normalizing the variables as:
u T − Ts x r
U= ,θ= , X = , and R = ,
um Tm − T s L ro
it can be shown that both governing equations and boundary conditions are identical
in form

9.6.1 Reynolds Analogy for Pipe Flow

• Assuming that ν = α (Pr = 1) and ε H = ε M ( Prt = 1), the same assumptions used to
develop Reynold’s analogy for a flat plate, then the governing equations (9.33a) and
(9.33b) are identical
• Following the same procedure as in the original derivation, the Reynolds analogy is
essentially identical for pipe flow:
q o′′ Cf
St D ≡ = , Pr = 1
ρ u m c p (Ts − Tm ) 2
or
Nu D Cf
St D = = (9.35)
Re D Pr 2
• Note that in this case the Stanton number is defined in terms of the mean velocity and
bulk temperature, as is the wall shear stress: τ o = 12 C f ρ u m2
9.6.2 Adapting Flat Plate Analogies to Pipe Flow
• Other flat plate analogies can be adapted to pipe flow, with modifications
• Example: von Kármán analogy
• For pipe flow, conditions at the edge of the boundary layer are approximated by
conditions at the centerline of the pipe:
V∞ ≈ u CL and T∞ ≈ TCL
• These substitutions affect the friction factor, which translates to:
τo
Cf ≈ 2
1
2
ρ uCL

• Following the development exactly as before, the result is almost identical:

q o′′ Cf /2
= (9.36)
ρ u CL c p (Ts − TCL ) Cf ⎧ ⎡ 5 Pr + 1⎤ ⎫
1+ 5 ⎨( Pr − 1) + ln ⎢ ⎬
2 ⎩ ⎣ 6 ⎥⎦ ⎭
• The left side of (9.36) and the friction factor are expressed in terms of centerline
variables instead of the more common and convenient mean quantities u m and Tm
• This is corrected:

q o′′ ⎛ u m (Ts − Tm ) ⎞
⎜ ⎟=
(C f / 2)(u m / uCL )2
ρ u m c p (Ts − Tm ) ⎜⎝ u CL (Ts − TCL ) ⎟⎠ u Cf ⎧ ⎡ 5Pr + 1⎤ ⎫
1+ 5 m ⎨( Pr − 1) + ln ⎢ ⎥⎬
u CL 2 ⎩ ⎣ 6 ⎦⎭
where
2
C f is defined in terms of the mean velocity C f = τ o / 12 ρ u m

The terms q o′′ / ρ u m c p (Ts − Tm ) collectively are the Stanton number for
pipe flow
• Simplifying yields von Kármán Analogy for pipe flow:
⎛ T − Tm ⎞ (C f / 2)(u m / u CL )
St D ⎜⎜ s ⎟⎟ = (9.37)
⎝ Ts − TCL ⎠ ⎛u ⎞ Cf ⎧ ⎡ 5 Pr + 1 ⎤ ⎫
1 + 5⎜⎜ m ⎟⎟ ⎨( Pr − 1) + ln ⎢ ⎥⎬
⎝ u CL ⎠ 2 ⎩ ⎣ 6 ⎦⎭
• Estimates for the ratios (u m / u CL ) and (Ts − Tm ) / (Ts − TCL ) are found using the
definition of mean temperature, equation (9.9):
th
• u m and Tm are estimated using the 1/7 Law profiles, which for a circular pipe are:
 1/ 7
u ⎛ y ⎞
= ⎜⎜ ⎟⎟ (9.20)
u CL ⎝ ro ⎠
• Similar to (8.111) for a flat plate:
1/ 7
T − Ts ⎛ y⎞
= ⎜⎜ ⎟⎟ (9.38)
TCL − Ts ⎝ ro ⎠

• Substituting these models into (9.8) and (9.9), it can be shown that
um
= 0.817 (9.39)
u CL
Tm − T s
= 0.833 (9.40)
TCL − Ts

9.6.3 Other Analogy-Based Correlations

• A simple correlation for turbulent flow in a duct is based on the Colburn analogy:
• Starting equation (8.96), and using equation (9.27) for the friction factor, one finds
St D = 0.023Re D−1 / 5 Pr −2 / 3

Nu D = 0.023Re D4 / 5 Pr 1 / 3 (9.41)
• The Dittus-Boelter correlation is an empirical correlation based on the Colburn analogy:
Nu D = 0.023Re D4 / 5 Pr n (9.42)
where n = 0.4 for heating ( Ts > Tm ) and n = 0.3 for cooling
• Simplicity and the fact that this analogy compares well with experimental data make
it popular
• In recent years its accuracy, along with that of the Colburn analogy, have been
challenged
• Models by Petukhov and the Gnielinski correlation (see Section 9.8) are preferred for
their improved accuracy and range of applicability
• Analogies remain a common way to model heat transfer in pipes; there are models
developed specifically for pipe flows

9.7 Algebraic Method Using Uniform Temperature Profile

• As was done for the flat plate, the universal temperature and velocity profiles can be
used to estimate the heat transfer in a circular duct
• Formulation
• The Nusselt number for flow in a duct is defined as
 hD q o′′ D
Nu D ≡ = (9.43)
k (Ts − Tm )k
• To invoke the universal temperature profile, the definition of T + , equation 8.102, is
used to define the mean temperature as
ρc p u* ρc pum C f / 2
Tm+ = (Ts − Tm ) = (Ts − Tm ) (9.44)
q o′′ q o′′

• For duct flow, the friction velocity u * is defined in terms of the mean velocity, so
substituting the above into (9.43) for q o′′ and invoking the definitions of the Reynolds
and Prandtl numbers:
Re D Pr C f / 2
Nu D = (9.45)
Tm+
• There are several ways to proceed from here:
• One approach: evaluate Tm+ using a dimensionless version of (9.33):
ro+

∫
2
Tm+ = T + u + (ro+ − y + ) dy + (9.46)
u m+ ro+ 2
0

• Substituting appropriate universal temperature and velocity profiles into

(9.46) and integrating must be done numerically
• A second approach yields a simpler closed-form solution
• Rewrite the original Nusselt number relation (9.43) as follows:
q o′′ D (Ts − TCL )
Nu D =
(Ts − Tm )k (Ts − TCL )
where TCL is the centerline temperature

• Substituting the definition of T + for the centerline temperature in the

denominator:
Re D Pr C f / 2 (Ts − TCL )
Nu D = +
(9.47)
TCL (Ts − Tm )
+
• The universal temperature profile, equation (8.118), is used to evaluate TCL :

+ Prt +
TCL = ln y CL + 13Pr 2 / 3 − 7 (9.48)
κ
 • As was done for the flat plate, substitute the Law of the Wall velocity profile
+
(8.59) for ln y CL :

+ 1 +
u CL = ln y CL +B (9.49)
κ
• Substituting these into the Nusselt number relation:
Re D Pr C f / 2 (Ts − TCL )
Nu D =
[ +
Prt (u CL − B ) + 13Pr 2/3
]
− 7 (Ts − Tm )
(9.50)

+
• Expressions are needed for u CL and (Ts − TCL ) /(Ts − Tm ) :

• For the centerline velocity, use the definition of u + for pipe flow

+ u CL u CL 2
u CL = = (9.51)
u* um Cf

• Mean velocity and temperature are needed, so, the 1/7th power law is used to
avoid the complexity of the logarithmic velocity and temperature profiles
• From the previous section, the 1/7th power law yields:
um T − Ts
= 0.817 and m = 0.833
u CL TCL − Ts
• Using the definition of Stanton number, St D = Nu D /( Re D Pr ) , and selecting
Prt = 0.9 and B = 5.0, equation (9.50) is rearranged:

Cf /2
St D = (9.52)
(
0.92 + 10.8 Pr 2/3 − 0.89 C f / 2 )
• A first approximation suggests this model be limited to ReD < 1 x 105
• The ultimate test is to compare this model to experimental data

9.8 Other Correlations for Smooth Pipes

• Petukhov evoked Reichardt’s model for eddy diffusivity and velocity profile (9.15, 9.16)
to obtain:
Cf /2 ⎛ 0.5 ≤ Pr ≤ 2000 ⎞
, ⎜⎜ 4 6⎟
St D =
(
1.07 + 12.7 Pr 2/3 ) ⎟
− 1 C f / 2 ⎝10 < Re D < 5 × 10 ⎠
(9.53)

• This compares well to experimental data over a wide range of Prandtl and Reynolds
numbers
• The following model was used for the friction factor:
 Cf
= (2.236 ln Re D − 4.639) −2 (9.54)
2
• Note the similarity between Petukhov’s relation (9.53) and the algebraic result, equation
(9.52)
• In 1976, Gnielinski modified Petukhov’s model slightly, extending the model to include
lower Reynolds numbers:
( Re D − 1000) PrC f / 2 ⎛ 0.5 ≤ Pr ≤ 2000 ⎞
Nu D = , ⎜⎜ 6⎟
(
1 + 12.7 Pr 2/3 ) 3 ⎟
− 1 C f / 2 ⎝ 3 × 10 < Re D < 5 × 10 ⎠
(9.55)

• Pethkhov’s friction model can be used in (9.55) for the friction factor
• For the above models, properties should be evaluated at the film temperature
• These correlations are reasonable for channels with constant surface temperature as well
as constant heat flux; the flows are relatively insensitive to boundary conditions

9.9 Heat Transfer in Rough Pipes

• The effects of roughness on the heat transfer from flat plates was discussed in Section
8.5.6, and much of the same physical intuition applies to flow in channels
• Norris presents the following empirical correlation for flow through circular tubes:
n
Nu ⎛ Cf ⎞ ⎛ Cf ⎞
=⎜ ⎟ , ⎜ < 4⎟ (9.56)
Nu smooth ⎜⎝ C f , smooth ⎟
⎠
⎜C
⎝ f , smooth
⎟
⎠
where n = 0.68 Pr 0.215
• A correlation like Colebrook’s (9.30) could be used to determine the rough-pipe friction
factor
• The behavior of this relation reflects what is expected physically:
• The Prandtl number influences the effect of roughness
• For very low Prandtl fluids the roughness plays little role in heat transfer
• The influence of roughness size is limited
• The effect of increasing roughness vanishes beyond (C f / C f ,smooth ) ≈ 4 , hence a
maximum is reached
• Though roughness enhances heat transfer, it increases friction, which increases pumping
costs, though the increase of friction due to roughness also reaches a limiting value
• The application of roughness to increase heat transfer requires benefits to be weighed
against increasing costs