C H A P T E R 19

Turbulent Boundary Layers

The one uncontroversial fact about turbulence is that it is the most complicated
kind of fluid motion.
Peter Bradshaw
Imperial College of Science and
Technology, London 1978
Turbulence was, and still is, one of the great unsolved mysteries of science, and
it intrigued some of the best scientific minds of the day. Arnold Sommerfeld,
the noted German theoretical physicist of the 1920s, once told me, for instance,
that before he died he would like to understand two phenomena—quantum
mechanics and turbulence. Sommerfeld died in 1924. I believe he was somewhat
nearer to an understanding of the quantum, the discovery that led to modern
physics, but no closer to the meaning of turbulence.
Theodore von Karman, 1967

PREVIEW BOX

Nature, when left to herself, always goes to the state because they remain attached to a surface for much
of maximum disorder. This is particularly true for larger distances downstream than a laminar boundary
boundary-layer flows under real conditions. For most layer under the same external flow conditions. Hence,
practical applications in aerodynamics, the flow in pressure drag due to flow separation is usually smaller
boundary layers is predominantly turbulent—nature for bodies with turbulent boundary layers compared
going to the state of maximum disorder. Turbulent to those with laminar boundary layers.
boundary layers can be bad news; for the same ex- Turbulent boundary layers are the subject of this
ternal flow conditions, the turbulent skin friction and chapter. There are no stand-alone theoretical results
aerodynamic heating is larger, frequently much larger, for turbulent boundary layers—any analysis must
than laminar skin friction and aerodynamic heating. incorporate empirical data in some form. So we will
But turbulent boundary layers can also be good news not beat around the bush. We go directly to empirically

1051
1052 PA RT 4 Viscous Flow

based formulas that allow the estimation (and it is books have been written on the subject of turbulent
truly only an estimation) of turbulent boundary-layer flow. The objective of this very short chapter is sim-
thickness and skin friction. There are many different ply to give you some ability to compute turbulent
approaches to the calculation of turbulent flows, all boundary-layer results, albeit imprecisely. Hang on,
requiring some input from experimental data. Whole and enjoy this short ride.

19.1 INTRODUCTION
The subject of turbulent flow is deep, extensively studied, but at the time of
writing still imprecise. The basic nature of turbulence, and therefore our ability to
predict its characteristics, is still an unsolved problem in classical physics. Many
books have been written on turbulent flows, and many people have spent their
professional lives working on the subject. As a result, it is presumptuous for us to
try to carry out a thorough discussion of turbulent boundary layers in this chapter.
Instead, the purpose of this chapter is simply to provide a contrast with our study
of laminar boundary layers in Chapter 18. Here, we will only be able to provide a
flavor of turbulent boundary layers, but this is all that is necessary in the present
book. Turbulence is a subject that we leave for you to study more extensively as
a subject on its own.
Before proceeding further, return to Section 15.2 and review the basic dis-
cussion of the nature of turbulence that is given there. In the present chapter, we
will pick up where Section 15.2 leaves off.
Also, we note that no pure theory of turbulent flow exists. Every analysis of
turbulent flow requires some type of empirical data in order to obtain a practi-
cal answer. As we examine the calculation of turbulent boundary layers in the
following sections, the impact of this statement will become blatantly obvious.
Finally, because this chapter is short, there is no need for a road map to act as a
guide.

19.2 RESULTS FOR TURBULENT BOUNDARY

LAYERS ON A FLAT PLATE
In this section, we discuss a few results for the turbulent boundary layer on
a flat plate, both incompressible and compressible, simply to provide a basis
of comparison with the laminar results described in the previous section. For
considerably more detail on the subject of turbulent boundary layers, consult
References 40 to 42.
For incompressible flow over a flat plate, the boundary-layer thickness is
given approximately by

0.37x
δ= (19.1)
Re1/5
x
 C H AP T E R 1 9 Turbulent Boundary Layers 1053

Note from Equation (19.1) that the turbulent boundary-layer thickness varies
approximately as Re−1/5
x in contrast to Re−1/2
x for a laminar boundary layer. Also,
turbulent values of δ grow more rapidly with distance along the surface; δ ∝ x 4/5
for a turbulent flow in contrast to δ ∝ x 1/2 for a laminar flow. With regard to
skin-friction drag, for incompressible turbulent flow over a flat plate, we have

0.074
Cf = (19.2)
Re1/5
c

Note that for turbulent flow, C f varies as Re−1/5

c in comparison with the Re−1/2
c
variation for laminar flow. Hence, Equation (19.2) yields larger friction drag coef-
ficients for turbulent flow in comparison with Equation (18.22) for laminar flow.
The effects of compressibility on Equation (19.2) are shown in Figure 19.1,
where C f is plotted versus Re∞ with M∞ as a parameter. The turbulent flow re-
sults are shown toward the right of Figure 19.1, at the higher values of Reynolds
numbers where turbulent conditions are expected to occur, and laminar flow re-
sults are shown toward the left of the figure, at lower values of Reynolds numbers.
This type of figure—friction drag coefficient for both laminar and turbulent flow
as a function of Re on a log-log plot—is a classic picture, and it allows a ready
contrast of the two types of flow. From this figure, we can see that, for the same

Figure 19.1 Turbulent friction drag coefficient for a flat plate as a function of Reynolds and Mach
numbers. Adiabatic wall Pr = 0.75. For contrast, some laminar results are shown. (Data are from the
calculations of Van Driest, E. R.: “Turbulent Boundary Layer in Compressible Fluids,” J. Aeronaut. Sci.,
vol. 18, no. 3, March 1951, p. 145).
1054 PA RT 4 Viscous Flow

Re∞ , turbulent skin friction is higher than laminar; also, the slopes of the turbulent
curves are smaller than the slopes of the laminar curves—a graphic comparison of
the Re−1/5 variation in contrast to the laminar Re−1/2 variation. Note that the effect
of increasing M∞ is to reduce C f at constant Re and that this effect is stronger on
the turbulent flow results. Indeed, C f for the turbulent results decreases by almost
an order of magnitude (at the higher values of Re∞ ) when M∞ is increased from
0 to 10. For the laminar flow, the decrease in C f as M∞ is increased though the
same Mach number range is far less pronounced.

19.2.1 Reference Temperature Method for Turbulent Flow

The reference temperature method discussed in Section 18.4 for laminar boundary
layers can be applied to turbulent boundary layers as well. With the reference
temperature T ∗ given by Equation (18.53), the incompressible turbulent flat plate
result for C f given by Equation (19.2) can be modified for compressible turbulent
flow as
0.074
C ∗f = (19.3)
(Re∗c )1/5
Df
where C ∗f = (19.4)
2
ρ
1 ∗ 2
ue S

E X A M PL E 19.1
Consider the same flat plate under the same external flow conditions given in Example 18.1.
Calculate the friction drag on the plate assuming a turbulent boundary layer for a freestream
velocity of (a) 100 m/s, and (b) 1000 m/s.
■ Solution
(a) From Example 18.1a, Rec = 1.36 × 107 . Hence, from Equation (19.2)
0.074 0.074 0.074
Cf = = = = 2.77 × 10−3
(Rec )1/5 (1.36 × 107 )1/5 26.71
Also from Example 18.1, we have ρ∞ = 1.22 kg/m3 and S = 40 m2 . Hence, for one side
of the plate,
2
D f = 12 ρ∞ V∞ SC f = 12 (1.22)(100)2 (40)(2.77 × 10−3 ) = 675.9 N

The total friction drag taking into account both sides of the plate is

D = 2D f = 2(675.9) = 1352 N

Comparing this result for turbulent flow with the laminar result in Example 1.81a, we
have
Dturbulent 1352
= = 7.7
Dlaminar 175.6
 C H AP T E R 1 9 Turbulent Boundary Layers 1055

Turbulent flow causes a factor of 7.7 increase in friction drag compared to the laminar flow.
You can easily see why the understanding of, and prediction of, turbulent flow, especially
the prediction of when the flow will transist from laminar to turbulent flow, is so important.
(b) For V∞ = 1000 m/s, from Example 18.1b, Rec = 1.36 × 108 , and M∞ = 2.94.
From Figure 19.1, we have
C f = 1.34 × 10−3

Hence, 2
D f = 12 ρ∞ V∞ SC f = 12 (1.22)(1000)2 (40)(1.34 × 10−3 ) = 32,700 N.
The total friction drag is

D = 2(32,700) = 65,400 N

Again, comparing this result with that from Example 18.1b, we have
Dturbulent 65,400
= = 13
Dlaminar 5026
Note that, at the higher Mach number of 2.92, turbulence increased the drag by a factor of
13, whereas for the incompressible case, the increase was 7.7, a smaller amount. The differ-
ence between the drag for laminar and turbulent flow is more pronounced at higher speeds.

EXAM PLE 19.2

Repeat Example 19.1b, except using the reference temperature method. Assume the plate
has an adiabatic wall.
■ Solution
We draw on the results calculated in Example 18.2. The recovery factor for a turbulent
flow is slightly different than that for a laminar flow. However, we will not account for
that difference, and we will assume that the reference temperature for this case is the same
as given in Example 18.2. Hence, from Example 18.2, we have
Re∗c = 3.754 × 107 and ρ ∗ = 0.574 kg/m3
From Equation (19.3), we have
0.074 0.074
C ∗f = = = 2.26 × 10−3
(Re∗c )1/5 (3.754 × 107 )1/5
From Equation (19.4),
D f = 12 ρ ∗ u 2e SC ∗f = 12 (0.574)(1000)2 (40)(2.26 × 10−3 ) = 25,945 N

Hence, D = 2(25,945) = 51,890 N

Comparing this answer with that obtained in Example 19.1b, we find a 20 percent dis-
crepancy between the two methods of calculations. This is not surprising. It simply points
out the great uncertainty in making calculations of turbulent skin friction.
1056 PA RT 4 Viscous Flow

19.2.2 The Meador-Smart Reference Temperature

Method for Turbulent Flow
The method developed recently by Meador and Smart, discussed in Section 18.4.1,
gives a reference temperature equation for turbulent flow slightly different than
that for laminar flow. For a turbulent flow, their equation is
   
T∗ Tw γ −1
= 0.5 1 + + 0.16 r Me2
Te Te 2
They also give a local turbulent skin-friction coefficient for incompressible flow as
τw 0.02296
cf = =
1
ρ u2
2 e e
(Rex )0.139
When integrated over the entire plate of length c, this gives for the net skin-friction
drag coefficient (prove it to yourself)
Df 0.02667
Cf = =
1
ρ V2 S
2 ∞ ∞
(Rec )0.139

E X A M PL E 19.3
Repeat Example 9.2 using the Meador-Smart reference temperature method.
■ Solution
From the above equation,
   
T∗ Tw γ −1
= 0.5 1 + + 0.16r Me2
Te Te 2
For turbulent flow, the recovery factor is approximately

r = Pr1/3 = (0.71)1/3 = 0.892

Taw − Te = r (T0 − Te )
 
Taw T0
or = 1+r −1
Te Te
For Me = 2.94,
T0
= 2.74
Te
Taw
= 1 + 0.892(1.74) = 2.55
Tr
Since the flat plate has an adiabatic wall, Tw = Taw . The Meador-Smith equation then
becomes
 
T∗ Tw
= 0.5 1 + + 0.16(0.892)(0.2)(2.94)2 = 0.5(1 + 2.55) + 0.2467 = 2.02
Te Te
 C H AP T E R 1 9 Turbulent Boundary Layers 1057

T ∗ = 2.02 Te = 2.02 (288) = 581.8 K

p 1.01 × 105
ρ∗ = ∗
= = 0.605 kg/m3
RT (287)(581.8)
From Sutherland’s law (note that T0 in Sutherland’s law is a reference temperature, not
the total temperature)
 3/2  3/2
μ∗ T∗ T0 + 110 581.8 398
= = = 1.651
μ0 T0 T ∗ + 110 288 691.8

μ∗ = 1.651 μ0 = 1.651(1.7894 × 10−5 ) = 2.05 × 10−5 kg/m · s

ρ∗uec (0.605)(1000)(2)
Re∗c = = = 4.1 × 107
μ∗ 2.95 × 10−5
From the Meador-Smith choice of the turbulent skin-friction coefficient equation,
0.02667 0.02667
C ∗f = = = 2.32 × 10−3
(Re∗c )0.139 (4.1 × 107 )0.139

D f = 12 ρ ∗ V∞
2
SC ∗f = 12 (0.605)(1000)2 (40)(2.32 × 10−3 ) = 28,070 N

Total drag = D = 2D f = 2(28,070) = 56,140 N

Note: This result is more accurate than that obtained in Example 9.2; it shows a 14 percent
discrepancy compared with the result obtained in Example 19.1b.

19.2.3 Prediction of Airfoil Drag

The flat-plate results obtained in Chapter 18 for laminar flow, and in the present
chapter for turbulent flow, can be used for engineering prediction of skin-friction
drag on thin airfoils. Using results from Chapters 18 and 19, airfoil drag in low-
speed incompressible flow is treated in Section 4.12, and supersonic airfoil drag
is discussed in Section 12.4. If you have not read Sections 4.12 and 12.4, do so
now. They give an important practical application of the boundary-layer results
we have just covered. Indeed, Sections 4.12 and 12.4 provide a vital continuation
of our discussion of viscous flow, and for all practical purposes they can be
considered integral sections of Part 4 of this book, although they were inserted in
the earlier chapters to provide some viscous flow reality to our otherwise inviscid
flow presentations. The prediction of airfoil drag is one of the most important
aspects of aerodynamics. Take it seriously, and make certain that you read, or
have read, Sections 4.12 and 12.4.

19.3 TURBULENCE MODELING

The simple equations given in Section 19.2 for boundary-layer thickness and
skin-friction coefficient for a turbulent flow over a flat plate are simplified results
that are heavily empirically based. Modern calculations of turbulent flows over
1058 PA RT 4 Viscous Flow

arbitrarily shaped bodies involve the solution of the continuity, momentum, and
energy equations along with some model of the turbulence. The calculations
are carried out by means of computational fluid dynamic techniques. Here we
will discuss only one model of tubulence, the Baldwin-Lomax turbulence model,
which has become popular over the past two decades. We emphasize that the
following discussion is intended only to give you the flavor of what is meant by a
turbulence model.

19.3.1 The Baldwin-Lomax Model

In order to include the effects of turbulence in any analysis or computation, it
is first necessary to have a model for the turbulence itself. Turbulence modeling
is a state-of-the-art subject, and a recent survey of such modeling as applied to
computations is given in Reference 81. Again, it is beyond the scope of the present
book to give a detailed presentation of various turbulence models; the reader is
referred to the literature for such matters. Instead, we choose to discuss only one
such model here, because: (a) it is a typical example of an engineering-oriented
turbulence model, (b) it is the model used in the majority of modern applications
in turbulent, subsonic, supersonic, and hypersonic flows, and (c) we will discuss
in the next chapter several applications which use this model. The model is called
the Baldwin-Lomax turbulence model, first proposed in Reference 82. It is in the
class of what is called an “eddy viscosity” model, where the effects of turbulence
in the governing viscous flow equations (such as the boundary-layer equations
or the Navier-Stokes equations) are included simply by adding an additional
term to the transport coefficients. For example, in all our previous viscous flow
equations, μ is replaced by (μ + μT ) and k by (k + k T ), where μT and k T
are the eddy viscosity and eddy thermal conductivity, respectively—both due to
turbulence. In these expressions, μ and k are denoted as the “molecular” viscosity
and thermal conductivity, respectively. For example, the x momentum boundary-
layer equation for turbulent flow is written as
 
∂u ∂u ∂p ∂ ∂u
ρu + ρv =− + (μ + μT ) (19.5)
∂x ∂y ∂x ∂y ∂y
Moreover, the Baldwin-Lomax model is also in the class of “algebraic,” or “zero-
equation,” models meaning that the formulation of the turbulence model utilizes
just algebraic relations involving the flow properties. This is in contrast to one-
and two-equation models which involve partial differential equations for the con-
vection, creation, and dissipation of the turbulent kinetic energy and (frequently)
the local vorticity. (See Reference 83 for a concise description of such one- and
two-equation turbulence models.)
The Baldwin-Lomax turbulence model is described below. We give just a
“cookbook” prescription for the model; the motivation and justification for the
model are described at length in Reference 82. This, like all other turbulence mod-
els, is highly empirical. The final justification for its use is that it yields reasonable
results across a wide range of Mach numbers, from subsonic to hypersonic. The
 C H AP T E R 1 9 Turbulent Boundary Layers 1059

model assumes that the turbulent-boundary layer is split into two layers, an inner
and an outer layer, with different expressions for μT in each layer:

(μT )inner y ≤ ycrossover
μT = (19.6)
(μT )outer y ≥ ycrossover
where y is the local normal distance from the wall, and the crossover point from
the inner to the outer layer is denoted by ycrossover . By definition, ycrossover is that
point in the turbulent boundary where (μT )outer becomes less than (μT )inner . For
the inner region:
(μT )inner = ρl 2 |ω| (19.7)
  
−y +
where l = ky 1 − exp (19.8)
A+
√
ρw τw y
y+ = (19.9)
μw
and k and A+ are two dimensionless constants, specified later. In Equation (19.7),
ω is the local vorticity, defined for a two dimensional flow as
∂u ∂v
ω= − (19.10)
∂y ∂x
For the outer region:
(μT )outer = ρ K Ccp Fwake FKleb (19.11)
where K and Ccp are two additional constants, and Fwake and FKleb are related to
the function
  
−y +
f (y) = y|ω| 1 − exp (19.12)
A+
Equation (19.12) will have a maximum value along a given normal distance y;
this maximum value and the location where it occurs are denoted by Fmax and
ymax , respectively. In Equation (19.11), Fwake is taken to be either ymax Fmax or
2
Cwk ymax Udif /Fmax , whichever is smaller, where Cwk is constant, and

Udif = u 2 + v 2 (19.13)
Also, in Equation (19.11), FKleb is the Klebanoff intermittency factor, given by
   −1
y 6
FKleb (y) = 1 + 5.5 CKleb (19.14)
ymax
The six dimensionless constants that appear in the above equations are: A+ =
26.0, Ccp = 1.6, CKleb = 0.3, Cwk = 0.25, k = 0.4, and K = 0.0168. These con-
stants are taken directly from Reference 82 with the understanding that, while they
are not precisely the correct constants for most flows in general, they have been
1060 PA RT 4 Viscous Flow

used successfully for a number of different applications. Note that, unlike many
algebraic eddy viscosity models that are based on a characteristic length, the
Baldwin-Lomax model is based on the local vorticity ω. This is a distinct advan-
tage for the analysis of flows without an obvious mixing length, such as separated
flows. Note that, like all eddy-viscosity turbulent models, the value of μT ob-
tained above is dependent on the flow-field properties themselves (for example ω
and ρ); this is in contrast to the molecular viscosity μ, which is solely a property
of the gas itself.
The molecular values of viscosity coefficient and thermal conductivity are
related through the Prandtl number
μc p
k= (19.15)
Pr
In lieu of developing a detailed turbulence model for the turbulent thermal con-
ductivity k T , the usual procedure is to define a “turbulent” Prandtl number as
PrT = μT c p /k T . Thus, analogous to Equation (19.15), we have
μT c p
kT = (19.16)
PrT
where the usual assumption is that PrT = 1. Therefore, μT is obtained from a given
eddy-viscosity model (such as the Baldwin-Lomax model), and the corresponding
k T is obtained from Equation (19.16).
Turbulence itself is a flow field; it is not a simple property of the gas. This
is why, as mentioned above, in an algebraic eddy viscosity model the values of
μT and k T depend on the solution of the flow field—they are not pure properties
of the gas as are μ and k. This is clearly seen in the Baldwin-Lomax model via
Equation (19.7), where μT is a function of the local vorticity in the flow, ω—a
flow-field variable which comes out as part of the solution for the particular case
at hand.

19.4 FINAL COMMENTS

This chapter and the previous two have dealt with boundary layers, especially
those on a flat plate. We end with the presentation of an artist’s rendering a
photograph in Figure 19.2 showing the development of velocity profiles in the
boundary layer over a flat plate. The fluid is water, which flows from left to right.
The profiles in the original photograph are made visible by the hydrogen bubble
technique, the same used for Figure 16.13. The Reynolds number is low (the
freestream velocity is only 0.6 m/s); hence, the boundary-layer thickness is large.
However, the thickness of the plate is only 0.5 mm, which means that the boundary
layer shown here is on the order of 1 mm thick—still small on an absolute scale.
In any event, if you need any further proof of the existence of boundary layers,
Figure 19.2 is it.
 C H AP T E R 1 9 Turbulent Boundary Layers 1061

Figure 19.2 Rendering of velocity profiles for the laminar flow over a flat
plate. Flow is from left to right.

19.5 SUMMARY

Approximations for the turbulent, incompressible flow over a flat plate are
0.37x
δ= (19.1)
Re1/5
x

0.074
Cf = (19.2)
Re1/5
c
To account for compressibility effects, the data shown in Figure 19.1 can be
used, or alternatively the reference temperature method can be employed.

When the continuity, momentum, and energy equations are used to solve a
turbulent flow, some type of turbulence model must be used. In the eddy
viscosity concept, the viscosity coefficient and thermal conductivity in these
equations must be the sum of the molecular and turbulent values.
1062 PA RT 4 Viscous Flow

19.6 PROBLEMS
Note: The standard sea level value of viscosity coefficient for air is μ = 1.7894 ×
10−5 kg/(m · s) = 3.7373 × 10−7 slug/(ft · s).
19.1 The wing on a Piper Cherokee general aviation aircraft is rectangular, with
a span of 9.75 m and a chord of 1.6 m. The aircraft is flying at cruising
speed (141 mi/h) at sea level. Assume that the skin-friction drag on the
wing can be approximated by the drag on a flat plate of the same
dimensions. Calculate the skin-friction drag:
a. If the flow were completely laminar (which is not the case in real life)
b. If the flow were completely turbulent (which is more realistic)
Compare the two results.
19.2 For the case in Problem 19.1, calculate the boundary-layer thickness at the
trailing edge for
a. Completely laminar flow
b. Completely turbulent flow
19.3 For the case in Problem 19.1, calculate the skin-friction drag accounting
for transition. Assume the transition Reynolds number = 5 × 105 .
19.4 Consider Mach 4 flow at standard sea level conditions over a flat plate
of chord 5 in. Assuming all laminar flow and adiabatic wall conditions,
calculate the skin-friction drag on the plate per unit span.
19.5 Repeat Problem 19.4 for the case of all turbulent flow.
19.6 Consider a compressible, laminar boundary layer over a flat plate.
Assuming Pr = 1 and a calorically perfect gas, show that the profile of
total temperature through the boundary layer is a function of the velocity
profile via
u
T0 = Tw + (T0,e − Tw )
ue
where Tw = wall temperature and T0,e and u e are the total temperature
and velocity, respectively, at the outer edge of the boundary layer. [Hint:
Compare Equations (18.32) and (18.41).]
19.7 Consider a high-speed vehicle flying at a standard altitude of 35 km,
where the ambient pressure and temperature are 583.59 N/m2 and
246.1 K, respectively. The radius of the spherical nose of the vehicle is
2.54 cm. Assume the Prandtl number for air at these conditions is 0.72,
that c p is 1008 joules/(kg K), and that the viscosity coefficient is given by
Sutherland’s law. The wall temperature at the nose is 400 K. Assume the
recovery factor at the nose is 1.0. Calculate the aerodynamic heat transfer
to the stagnation point for flight velocities of (a) 1500 m/s, and
(b) 4500 m/s. From these results, make a comment about how the heat
transfer varies with flight velocity.