Vous êtes sur la page 1sur 31

ME 219

Chapter 7
328
Internal Flows – Fluid flow in Pipes
Chapter 6 Viscous Flow in Ducts
Ref: F M White, Fluid Mechanics
Flow
(a)

(b)

(c)

Fig. 6.3 Formation of a turbulent


puff in pipe flow: (a) and (b) near (d)
the entrance; (c) somewhat down-
stream; (d) far downstream. (From
Ref. 45, courtesy of P. R. Bandy-
opadhyay.) 1

and may be visualized by experimenting with glass tube flow. Figure 6.3 shows a puff
as photographed by Bandyopadhyay [45]. Near the entrance (Fig. 6.3a and b) there is
an irregular laminar-turbulent interface, and vortex roll-up is visible. Further down-
stream (Fig. 6.3c) the puff becomes fully turbulent and very active, with helical mo-
tions visible. Far downstream (Fig. 6.3d), the puff is cone-shaped and less active, with
a fuzzy ill-defined interface, sometimes called the “relaminarization” region.
A complete description of the statistical aspects of turbulence is given in Ref. 1, while
theory and data on transition effects are given in Refs. 2 and 3. At this introductory level
we merely point out that the primary parameter affecting transition is the Reynolds num-
ber. If Re ! UL/", where U is the average stream velocity and L is the “width,” or trans-
verse thickness, of the shear layer, the following approximate ranges occur:
VISCOUS FLOW IN PIPES
0 # Re # 1: highly viscous laminar “creeping” motion
1 # Re # 100: laminar, strong Reynolds-number dependence
• Applications: pipeline to carry crude
3
oil
100 # Re # 10 : laminar, boundary-layer theory useful
• Natural systems of “pipes” that
10 : carry blood throughout our body and air
3 4
10 # Re # transition to turbulence
104 # Re # 106: turbulent, moderate Reynolds-number dependence
into and out of our
106 # lungs
Re # $: turbulent, slight Reynolds-number dependence

• water pipes in our These


homes and theranges
are representative distribution systemwiththat
which vary somewhat delivers
flow geometry, the
surface
roughness, and the level of fluctuations in the inlet stream. The great majority of our
water from theanalyses
city are
well to the
concerned withhouse
laminar flow or with turbulent flow, and one should not
normally design a flow operation in the transition region.
•Numerous hoses and pipes carry hydraulic fluid or other fluids to various
Historical OutlineSince turbulent flow is more prevalent than laminar flow, experimenters have observed
components of vehicles and machines
turbulence for centuries without being aware of the details. Before 1930 flow instru-
ments were too insensitive to record rapid fluctuations, and workers simply reported
• The air quality within our buildings is maintained at comfortable levels
by the distribution of conditioned heated, cooled air through a maze
of pipes and ducts

• Although all of these systems are different, the fluid-mechanics principles


governing the fluid motions are common. The purpose of this
chapter is to understand the basic processes involved in such
2
cases

TYPICAL PIPE SYSTEM COMPONENTS 3

1
Pipe Flow Open Channel Flow
Pipe Flow: driving force is the pressure gradient alone, gravity
may/may not be important (inclined and large heights)
Open Channel Flow: driving force is the gravity alone – water flows
down the hill. If the pipe is not full, it is not possible tomaintain this
pressure difference P1 - P2

We are considering those pipes in which the fluid is running full


4
6.1 Reynolds-Number Regimes 329

mean values of velocity, pressure, force, etc. But turbulence can change the mean val-
ues dramatically, e.g., the sharp drop in drag coefficient in Fig. 5.3. A German engineer
named G. H. L. Hagen first reported in 1839 that there might be two regimes of vis-
cous flow. He measured water flow in long brass pipes and deduced a pressure-drop law
LQ
"p # (const) ! $ entrance effect (6.1)
R4
This is exactly our laminar-flow scaling law from Example 5.4, but Hagen did not re-
alize that the constant was proportional to the fluid viscosity.
6.1 Reynolds-Number
The formula broke down as Hagen increased Q beyond a certain limit, i.e., 329
Regimes past the
critical Reynolds number, and he stated in his paper that there must be a second mode
of flow characterized by “strong movements of water for which "p varies as the sec-
ond power of the discharge. . . .” He admitted that he could not clarify the reasons for
mean values of velocity, pressure, force, etc. But turbulence can change the mean val-
the change.
ues dramatically, e.g., the sharp drop in example
A typical drag coefficient
of Hagen’s datainis Fig.
shown5.3. A6.4.
in Fig. German engineer
The pressure drop varies
named G. H. L. linearly with
Hagen firstpressure
reported # Q/A up
inVdrop
1839 to about
that 1.1 ft/s,
there mightwhere be
theretwo
is a sharp change. Above
regimes of vis- about
Hagen measured V # 2.2 ft/s the pressureand
drop isdeduced
nearly quadratic a withpressure-drop
V. The actual power "p % V law
1.75

cous flow. He measured water flowseems in long brass


impossible pipes and
on dimensional groundsdeduced a pressure-drop
but is easily explained when thelaw dimen-
sionless pipe-flow data (Fig. 5.10) are displayed.
LQ Reynolds, a British engineering professor, showed that the change
In 1883 Osborne
"p # (const) ! $ entrance effect (6.1)
depended upon R4the parameter &Vd/', now named in his honor. By introducing a dye
This is exactly our laminar-flow120scaling law from Example 5.4, but Hagen did not re-
alize that the constant was proportional to the fluid viscosity.
The formula broke down as Hagen increased Q beyond a Turbulent
certain flowlimit, i.e., past the
∆p α V 1.75
critical Reynolds number, and he100stated in his paper that there must be a second mode
of flow characterized by “strong movements of water for which "p varies as the sec-
ond power of the discharge. . . .”80He admitted that he could not clarify the reasons for
the change.
Pressure drop ∆p, lbf/ft2

A typical example of Hagen’s data is shown in Fig. 6.4. The pressure drop varies
linearly with V # Q/A up to about 60 1.1 ft/s, where there is a sharp change. Above about

V # 2.2 ft/s the pressure drop is nearly quadratic with V. The actual power "p % V1.75
seems impossible on dimensional grounds but is easily explained when the dimen-
sionless pipe-flow data (Fig. 5.10)
40
are displayed.
In 1883 Osborne Reynolds, a British engineering
Laminar flow professor, showed that the change
∆p α V
depended upon the parameter &Vd/ 20
', now named in his honor. By introducing a dye

120 Transition

Fig. 6.4 Experimental evidence of 0


transition for water flow in a !14!-in 0 0.5 1.0 1.5 2.0 2.5 5
smooth pipe 10 ft long. Average velocity V, ft/s
Turbulent flow
100 ∆p α V 1.75

80
Pressure drop ∆p, lbf/ft2

60
ENTRANCE REGION AND FULLY DEVELOPED FLOW
6.2 Internal versus External Viscous Flows 331

le
= 4.4 [Re ]1 / 6
Growing Boundary Developed
boundary Inviscid layers velocity
40 layers core flow merge profile u (r)
D
Turbulent
Laminar flow
Flow r

∆p α V x
le
= 0.06 Re
20 u (r, x)

D Transition
Laminar Flow
Entrance length Le Fully developed
(developing profile region) flow region
perimental evidence of 0
or water flow in a !14!-in 0 0.5 1.0 1.5 2.0 2.5
e 10 ft long. Average velocity V, ft/s Pressure

Entrance
Linear
pressure
pressure
drop
drop in
fully developed
flow region

Fig. 6.6 Developing velocity pro-


files and pressure changes in the x
entrance of a duct flow. 0 Le

6
shear is constant, and the pressure drops linearly with x, for either laminar or turbu-
lent flow. All these details are shown in Fig. 6.6.
Dimensional analysis shows that the Reynolds number is the only parameter af-
fecting entrance length. If
Q
Le ! f(d, V, ", #) V!$
A
L "Vd
then
d # ! "
$e ! g $ ! g(Re) (6.4)

For laminar flow [2, 3], the accepted correlation is


L
$e # 0.06 Re laminar (6.5)
d
The maximum laminar entrance length, at Red,crit ! 2300, is Le ! 138d, which is the
longest development length possible.
2
In turbulent flow the boundary layers grow faster, and Le is relatively shorter, ac-
cording to the approximation for smooth walls
ENTRANCE REGION AND FULLY DEVELOPED FLOW

le
= 0.06 Re Laminar Flow
D
le
= 4.4 [Re ]1 / 6 Turbulent Flow
D 7

Typical values of L/D

Re D l e /D
4000 13
1e4 16
1e5 28
1e6 51
1e7 90

Pressure and Shear Stress


Fully developed flow
P Entrance flow
∂p
= Cons tan t
∂x
Entrance pres s ure
drop

X1= 0 X2 X3
•For ho rizo ntal pipe f low, grav ity has no eff ect except fo r a hyd ro stat ic
pressure variation across the pipe, that is usually negligible
• Pre ss ure diffe rence, be tween one sect ion of t he h orizo nta l p ipe a nd
another which forces the fluid through the pipe
• Visco us effect s prov id e the re stra in ing force th at ex actly ba lance s the
pressure force, thereby allowing the fluid to flow through the pipe with
9
no acceleration

3
•If viscous effects were absent in such flows, the pressure would be consta nt
throughout the pipe, except for the hydrostatic variation
• In non-f ully deve lo ped f lo w region s, such as t he entra nce region of a pipe, the
flu id accelerat es o r decelerate s a s it flo ws (th e ve loc ity pr ofile c hange s from
a u nif orm pr ofile at th e e ntra nce of the p ipe t o it s f ully deve lo ped p rof ile at
the end of the entrance region).
• In th e en trance reg ion the re is a ba la nce bet we en pre ss ure, v isco us, a nd ine rtial
(accelerat ion) force s. Th e re su lt is a pre ss ure distr ib ution along the
hor iz onta l p ipe a s s ho wn in F ig. The ma gn itu de of t he pr es sur e grad ie nt, is
larger in the en trance reg io n t han in t he fu lly deve lo pe d re gion, whe re it is a
constant,
• The fact that there is a nonzero pressure gradient along the horizontal pipe is a
result of viscous effects.
• The need for the pressure drop can be viewed from two different standpoints.
• In t erm s of a f orce balance, t he pr es sur e f orce is n eed ed to ove rcome th e
viscous forces generated.
• In term s of an e nergy ba la nce, th e wo rk do ne by th e pre ss ure force is ne ede d
to overcome the viscous dissipation of energy throughout the fluid
10

The nature of the pipe flow is strongly dependent on whether the flow
is laminar or turbulent. This is a direct consequence of the differences
in the nature of the shear stress in laminar andturbulent flows
The shear stress in laminar flow is a direct result of momentum
transfer among the randomly moving molecules (a microscopic
phenomenon)
The shear stress in turbulent flow is largely a result of momentum
transfer among the randomly moving, finite-sized bundles of fluid
particles (a macroscopic phenomenon).
THE NET RESULT IS THAT THE PHYSICAL PROPERTIES OF THE SHEAR
STRESS ARE QUITE DIFFERENT FOR LAMINAR FLOW THAN FOR
TURBULENT FLOW.

11

Laminar and Turbulent flow

Laminar Re < 2100

Turbulent
Re < 2100 - 4000
Trans itional

Re > 4000
x uA
Turbulent

Tra nsitio na l

La mina r

Q
12
t

4
Famous Reynolds experiment

Os borne Reynolds
(wik ipedia)

13

Representative ranges of flow based on Re


ReD Le/D
0 < Re < 1 Highly viscous laminar creeping flow
1 < Re < 100 Laminar, strong Re dependence
100 < Re < 1e3 Laminar, boundary layer theory useful
1e3 < Re < 1e4 Transition to turbulence
1e4 < Re < 1e6 Turbulent, moderate Re dependence
1e6 < Re < ∞ Turbulent, slight Re dependence
Ranges depend on
flow geometry, surface
roughness, level of
fluctuation within inlet
stream
326 Chapter 6 Viscous Flow in Ducts

u u u

Small natural Continuous


disturbances Intermittent turbulence
damp quickly bursts of
Fig. 6.1 The three regimes of vis- turbulence
cous flow: (a) laminar flow at low 14
Re; (b) transition at intermediate t t t
Re; (c) turbulent flow at high Re. (a) (b) (c)

flow will appear steady on average but will reveal rapid, random fluctuations if turbu-
lence is present, as sketched in Fig. 6.1. If the flow is laminar, there may be occasional
natural disturbances which damp out quickly (Fig. 6.1a). If transition is occurring, there
will be sharp bursts of turbulent fluctuation (Fig. 6.1b) as the increasing Reynolds num-
ber causes a breakdown or instability of laminar motion. At sufficiently large Re, the
flow will fluctuate continually (Fig. 6.1c) and is termed fully turbulent. The fluctua-
tions, typically ranging from 1 to 20 percent of the average velocity, are not strictly
periodic but are random and encompass a continuous range, or spectrum, of frequen-
cies. In a typical wind-tunnel flow at high Re, the turbulent frequency ranges from 1
DEFINITION OF TURBULENCE:
to 10,000 Hz, and the wavelength ranges from about 0.01 to 400 cm.

Taylor and Von KarmanEXAMPLE


(1937)6.1
The accepted transition Reynolds number for flow in a circular pipe is Red,crit ! 2300. For flow
Turbulence is an irregular motion which in general makes its
through a 5-cm-diameter pipe, at what velocity will this occur at 20°C for (a) airflow and (b) wa-
ter flow?
appearance in fluids, gaseous
Solution
or liquids, when they flow past solid
surfaces or even whenAlmost
neighbouring streams of same fluid past over
all pipe-flow formulas are based on the average velocity V ! Q/A, not centerline or any
one another other point velocity. Thus transition is specified at "Vd/# ! 2300. With d known, we introduce
the appropriate fluid properties at 20°C from Tables A.3 and A.4:

Turbulent fluid motion(a) Air:


is an $ irregular condition ! 2300 ofor flow V ! 0.7in
$ which
3
"Vd (1.205 kg/m )V(0.05 m) m
! $$$
# 1.80 E-5 kg/(m % s) s
various quantities show a random "Vd variation
(998 kg/m )V(0.05 m) with time and
3
m space
(b) Water: $ ! $$$ ! 2300 or V ! 0.046 $
coordinates, so that statistically #
distinct average values scan be
0.001 kg/(m % s)

discerned These are very low velocities, so most engineering air and water pipe flows are turbulent, not
laminar. We might expect laminar duct flow with more viscous fluids such as lubricating oils or
glycerin.

In free surface flows, turbulence can be observed directly. Figure 6.2 shows liquid
flow issuing from the open end of a tube. The low-Reynolds-number jet (Fig. 6.2a) is
smooth and laminar, with the fast center motion and slower wall flow forming different
trajectories joined by a liquid sheet. The higher-Reynolds-number turbulent flow (Fig.
6.2b) is unsteady and irregular but, when averaged over time, is steady and predictable.

15

5
DESIRABLE SITUATIONS OF TURBULENT FLOW
•To tra nsf er th e re qu ired heat bet wee n a s olid a nd a n adjace nt fluid s uch as in the
cooling coils of an air conditioner or a boiler of a power plant would requ ire an
enormously large heat exchanger if the flow were laminar
• Tur bu le nce is also of imp orta nce in the m ixing of f lu id s. Sm oke f rom a stack wou ld
continue f or m iles a s a r ib bo n of po llutant with out rap id dispe rs io n wit hin the
surrounding air if the flow were laminar rather than turbulent
• Alth oug h the re is m ix ing o n a m olecula r scale ( lam inar f lo w), it is seve ral o rder s of
magnitud e s lo wer an d le s s e ffective t han t he mix ing o n a macro scop ic scale
(turb ulent flow). It is con sidera bly ea sier t o mix cream in to a cu p of coff ee (t ur bu lent
flow) than to thoroughly mix two colors of a viscous paint (laminar flow)

DESIRABLE SITUATIONS OF LAMINAR FLOW


•Pres su re dr op in p ip es - p o wer re qu ire ment s for pum ping can b e con sidera bly
lower if the flow is laminar rather than turbulent
• B lo od flow thr oug h a per so n’s art eries is n orma lly la minar, except in t he larg est
arteries with high blood flowrates
•Aerody nam ic drag o n an a irp lan e wing can be con sidera bly sm aller wit h laminar
flow past it than with turbulent

16

FULLY DEVELOPED TURBULENT FLOW: MEAN MOTION


AND FLUCTUATIONS
Random turbulent Turbulent
fluctuations

u
4000
Turbulent burs ts ReD
Trans itional

2000

Laminar

334 Chapter 6 Viscous Flow in Ducts


t
u p

p = p + p′

u = u + u′
u′
p

u
p′

Fig. 6.7 Definition of mean and


t t
fluctuating turbulent variables: 17
(a) velocity; (b) pressure. (a) (b)

However, the mean square of a fluctuation is not zero and is a measure of the inten-
sity of the turbulence
1
!!2! " *
u
T
"T

0
u! 2 dt # 0 (6.11)

!!!$!! and u
Nor in general are the mean fluctuation products such as u !!p
!! zero in a typi-
cal turbulent flow.
Reynolds’ idea was to split each property into mean plus fluctuating variables
u"!
u % u! $ " $! % $! w " !
w % w! p " !
p % p! (6.12)
Substitute these into Eqs. (6.7), and take the time mean of each equation. The conti-
nuity relation reduces to

MEAN MOTION AND


)u
*
)x
)$!
! %*
)y
FLUCTUATIONS
%
)w
*! "0
)z
(6.13)

which is no different from a laminar continuity relation.


However, each component of the momentum equation (6.7b), after time averaging,
will contain mean values plus three mean products, or correlations, of fluctuating ve-
locities. The most important of these is the momentum relation in the mainstream, or

t o +T
x, direction, which takes the form

1 &
du!
*
)p
" '* &
)
! % g %*
#( ' & $
)u
*! !!2!
u

∫ u (x( , 'y&, $z , t ) (dt' &


x

u = dt )x )x )x
(6.14)

T # $ # $
) )u
! ) )u
!
%* * !!!!!! % *
u * !!w
u !!
)y )y )z )z
t
The three correlation terms '&u
o
because they have the same
!!!, '&u!!!$!!!, and '&u
2
!!w
!! are called turbulent stresses
dimensions and occur right alongside the newtonian (lam-
inar) stress terms (()u
!/)x), etc. Actually, they are convective acceleration terms (which
is why the density appears), not stresses, but they have the mathematical effect of stress

u = u + uʹ ; uʹ = u − u
and are so termed almost universally in the literature.
The turbulent stresses are unknown a priori and must be related by experiment to

18

6
Fluctuation u’

t o +T ⎡ t +T t o +T ⎤ ⎡ t o +T t o +T ⎤
1 1⎢o ⎥ 1 ⎢ ⎥
uʹ =
T ∫ (u − u ) dt = T ⎢ ∫ u dt − ∫ u dt ⎥ = T ⎢ ∫ u dt − u ∫ dt ⎥
to ⎢⎣ t o to ⎥⎦ ⎢⎣ t o to ⎥⎦

⎡ t +T ⎤
1⎢o ⎥ = 1 [T u − u T ] = 0
uʹ = u dt − u T
T⎢ ∫ ⎥ T
⎢⎣ t o ⎥⎦

uʹ = 0
19

Mean square of fluctuation is measure of intensity of turbulence

_______ to + T
1
( u ʹ )2 = =
T ∫ ( uʹ )
to
2
dt ≠ 0

1
⎡ t +T ⎤2
⎢1 o ⎥
______ ⎢ ∫ (uʹ)2 dt ⎥
2 T t

(uʹ) = ⎣ o ⎥⎦
Turbulence Intensity =
u u
20

u2 = (uʹ + u )2 = (uʹ)2 + u 2 + 2u uʹ
Taking mean on both sides

______ _______ ________ ________

u2 = (uʹ)2 +
u 2 + 2u uʹ

________ t o +T ⎡ t +T ⎤ ________
1 u⎢o ⎥ u
u uʹ = ∫ (u uʹ) dt = T ⎢ ∫ uʹ dt ⎥ = T [zero ] = 0 u uʹ = 0
T
to ⎢⎣ t o ⎥⎦

______ _______

u2 = (uʹ)2 + u .u
21

7
____ ___________________ ______ ______ ______
uv = (u + uʹ)(v + vʹ) = u v + uʹv + vʹu + uʹvʹ

____ ______
uv = u v + + uʹvʹ
RULES

___
f = f ∂f ∂f
____ =
f+ g = f + g ∂s ∂s
______
____
f .g = f .g ∫ f .ds = ∫ f ds
22

REYNOLDS EQUATIONS AND REYNOLDS STRESSES

∂u ∂v ∂w
+ + =0
∂x ∂ y ∂z
⎛ ∂u ∂u ∂u ∂u ⎞ ∂P ⎡ ∂ 2u ∂ 2u ∂ 2u ⎤
ρ ⎜⎜ + u + v + w ⎟⎟ = − + µ⎢ + + ⎥
⎝ ∂t ∂x ∂y ∂z ⎠ ∂x ⎢⎣ ∂x 2 ∂y 2 ∂z 2 ⎥⎦

⎛ ∂v ∂v ∂v ∂v ⎞ ∂P ⎡ ∂ 2v ∂ 2v ∂ 2v ⎤
ρ ⎜⎜ + u + v + w ⎟⎟ = − + µ⎢ + + ⎥
⎝ ∂t ∂x ∂y ∂z ⎠ ∂y ⎢⎣ ∂x 2 ∂y 2 ∂z 2 ⎥⎦

⎛ ∂w ∂w ∂w ∂w ⎞ ∂P ⎡ ∂ 2w ∂ 2w ∂ 2w ⎤
ρ ⎜⎜ + u +v +w ⎟⎟ = − + µ⎢ + + ⎥
⎝ ∂t ∂x ∂y ∂z ⎠ ∂z ⎢⎣ ∂x 2 ∂y 2 ∂z 2 ⎥⎦

23

u = u + uʹ ; v = v + v ʹ ; w = w + wʹ ; p = p + pʹ
∂u ∂v ∂w
+ + =0
∂x ∂ y ∂z
Taking mean on both sides
_____ _____ _____
∂ u + uʹ ∂ v + vʹ ∂ w + wʹ
+ + =0
∂x ∂y ∂z

_____ __
∂ u + uʹ ∂ u ∂ uʹ
= +
∂x ∂x ∂x

∂u ∂v ∂w
+ + =0
∂x ∂ y ∂z
24

8
⎛ ∂u ∂u ∂u ∂u ⎞ ∂P ⎡ ∂ 2u ∂ 2u ∂ 2u ⎤
ρ ⎜⎜ +u +v +w ⎟⎟ = − + µ⎢ + + ⎥
⎝ ∂t ∂x ∂y ∂z ⎠ ∂x ⎢⎣ ∂x 2 ∂y 2 ∂z 2 ⎥⎦
⎛ ∂u ∂ (u 2 ) ∂ ( uv ) ∂ ( uw ) ⎛ ∂u ∂v ∂w ⎞ ⎞
ρ⎜ + + + − ⎜u + u + u ⎟⎟
⎜ ∂t ∂x ∂y ∂z ⎝ ∂x ∂y ∂z ⎠ ⎟
⎝ ⎠
∂P ⎡ ∂ 2u ∂ 2u ∂ 2u ⎤ Continuity
=− +µ⎢ 2 + 2 + 2⎥
∂x ⎣ ∂x ∂y ∂z ⎦
_____ __
∂ u + uʹ ∂ u ∂ uʹ ∂u
= + =
∂t ∂t ∂t ∂t

∂u
= zero for Mean Steady Flows
∂t
25

___ __________ ____ ____


∂ u2 ∂ (u + uʹ)2 ∂ uʹ 2 ∂u 2 ∂ uʹ 2 ∂u
= = + = + 2u
∂x ∂x ∂x ∂x ∂x ∂x
___ ____
∂ u2 ∂ uʹ 2 ∂u
= + 2u
∂x ∂x ∂x
⎛ ____ ____ ___⎞
⎜ ⎟
___ _____________ ∂ ⎜ u v + uʹv ʹ + u v ʹ + v uʹ ⎟
∂ uv ∂ (u + uʹ)(v + v ʹ) ⎝ ⎠
= =
∂y ∂y ∂y
___ ____
∂ uv ∂v ∂u ∂ uʹv ʹ
=u +v +
∂y ∂y ∂y ∂y
___ ____
∂ uw ∂w ∂u ∂ uʹwʹ
=u +w +
∂z ∂z ∂z ∂z 26

_____ __
∂ p + pʹ ∂ p ∂ pʹ
= +
∂x ∂x ∂x
_____
∂ p + pʹ ∂ p
=
∂x ∂x
⎡ ____ ____ ____ ⎤
∂ uʹ 2
⎢ ⎥
∂u ∂v ∂u ∂ uʹv ʹ ∂w ∂u ∂ uʹw ʹ ⎥
ρ⎢ + 2u +u +v + +u +w +
⎢ ∂x ∂x ∂y ∂y ∂y ∂z ∂z ∂z ⎥
⎢ ⎥
⎣ ⎦

∂p ⎡ ∂ 2u ∂ 2u ∂ 2u ⎤
=− + µ⎢ + + ⎥
∂x 2 ∂y 2 ∂z 2 ⎦⎥
⎣⎢ ∂x

27

9
Continuity
⎡ ⎛ ∂ u ∂ v ∂w ⎞ ⎤ ⎡ ∂u ∂u ∂u ⎤
ρ ⎢ u ⎜⎜ + + ⎟⎟ ⎥ + ρ ⎢ u +v +w
⎣ ⎝ ∂x ∂ y ∂z ⎠ ⎦ ⎣ ∂x ∂y ∂z ⎥⎦
⎡ ____ ____ ____ ⎤
∂p ⎡∂ 2u ∂ 2u ∂ 2u ⎤

∂ uʹ 2 ∂ uʹv ʹ

∂ u ʹw ʹ ⎥
=− + µ ⎢ + + ⎥ −ρ ⎢ + +
∂x ⎢⎣ ∂x 2 ∂y 2 ∂z 2 ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎣ ⎦
⎡ ____ ____ ____ ⎤
⎡ ∂u ∂u ∂u ⎤ ∂p ⎡ ∂ 2u ∂ 2u ∂ 2u ⎤

∂ uʹ 2 ∂ uʹv ʹ

∂ uʹw ʹ ⎥
ρ ⎢u +v +w ⎥ =− + µ⎢ + + ⎥ − ρ⎢ + +
⎣ ∂x ∂y ∂z ⎦ ∂x 2 ∂y 2 ∂z 2 ⎦⎥ ⎢ ∂x ∂y ∂z ⎥
⎣⎢ ∂x ⎢ ⎥
⎣ ⎦

28

⎡ ____ ____ ____ ⎤


⎡ ∂u ∂u ∂u ⎤ ∂p
⎢ 2 ⎥
ρ ⎢u +v +w =− + µ ⎡∇ 2 u ⎤ − ρ ⎢ ∂ uʹ +
∂ uʹv ʹ ∂ uʹwʹ ⎥
+
⎣ ∂x ∂y ∂z ⎥⎦ ∂x ⎣⎢ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎣ ⎦
⎡ ____ ____ ____ ⎤
⎡ ∂v ∂v ∂w⎤ ∂p

∂ uʹv ʹ ∂ v ʹ 2 ∂ v ʹwʹ ⎥

ρ ⎢u +v +w =− + µ ⎡∇ 2 v ⎤ − ρ ⎢ + +
⎣ ∂x ∂y ∂z ⎥⎦ ∂y ⎢⎣ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎣ ⎦
⎡ ____ ____ ____ ⎤
⎡ ∂w ∂w ∂w⎤ ∂p

∂ uʹw ʹ ∂ v ʹw ʹ ∂ wʹ ⎥ 2 ⎥
⎡ 2 ⎤
ρ ⎢u
∂ x
+v
∂ y
+w
∂ z ⎥ = − ∂ z + µ ⎢⎣∇ w ⎦⎥ − ρ ⎢⎢ ∂x + ∂y +
∂ z ⎥
⎣ ⎦
⎢ ⎥
⎣ ⎦

29

⎡ ____ ____ ____ ⎤


⎡ ∂u ∂u ∂u ⎤ ∂p

∂ uʹ2 ∂ uʹv ʹ ∂ uʹwʹ ⎥

ρ ⎢u +v +w =− + µ ⎡∇ 2 u ⎤ − ρ ⎢ + +
⎣ ∂x ∂y ∂z ⎥⎦ ∂x ⎣⎢ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎣ ⎦
⎡ ____ ____ ____ ⎤
⎡ ∂v ∂v ∂w⎤ ∂p

uʹ v ʹ v ʹ 2 v ʹw ʹ ⎥
ρ ⎢u +v +w =− ⎡ 2 ⎤
+ µ ∇ v −ρ ⎢

+

+
∂ ⎥
⎣ ∂x ∂y ∂z ⎥⎦ ∂y ⎢⎣ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎣ ⎦
⎡ ____ ____ ____ ⎤
∂ wʹ 2 ⎥
⎡ ∂w ⎢ ʹ ʹ ⎥
∂w ∂w⎤ ∂p ∂ u w ∂ v ʹw ʹ
ρ ⎢u +v +w =− + µ ⎡∇ 2 w ⎤ − ρ ⎢ + +
⎣ ∂x ∂y ∂z ⎥⎦ ∂z ⎣⎢ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎢ ⎥
⎣ ⎦

⎡ ∂u ∂u ∂u ⎤ ∂p
ρ ⎢u +v + w ⎥ =− + µ ⎡∇ 2 u ⎤
⎣ ∂x ∂y ∂z ⎦ ∂x ⎢⎣ ⎥⎦ Steady state Navie r Stok es
equat io ns, if t he v elocity
⎡ ∂v
ρ ⎢u +v
∂v
+w
∂w⎤
=−
∂p
+ µ ⎡∇ 2 v ⎤
compo nent s an d pre ss ure
⎣ ∂x ∂y ∂z ⎥⎦ ∂y ⎣⎢ ⎥⎦
compo nent s are r ep laced by m ean
components or time averages
⎡ ∂w ∂w ∂w⎤ ∂p
ρ ⎢u +v +w =− + µ ⎡∇ 2 w ⎤
⎣ ∂x ∂y ∂z ⎥⎦ ∂z ⎣⎢ ⎥⎦

30

10
Resultant surface force per unit area due to the additional terms
⎡ ' ∂τ 'xy ∂τ 'xz ⎤ ⎡ ∂τ ' ∂σ 'yy ∂τ 'yz ⎤ ⎡ ' ∂τ ' ' ⎤
∂σ
P = i ⎢ xx + + ⎥ + j ⎢ xy + + ⎥ + k ⎢ ∂τ xz + yz + ∂σ zz ⎥
⎢ ∂x ∂y ∂z ⎥ ⎢ ∂x ∂y ∂z ⎥ ⎢ ∂x ∂y ∂z ⎥
⎣⎢ ⎦⎥ ⎢⎣ ⎦⎥ ⎣⎢ ⎦⎥

Turbulent stresses
___ ___ ___
σ 'xx = − ρ uʹ 2 σ 'yy = − ρ v ʹ 2 σ 'zz = − ρ w ʹ 2

___ ___ ___


τ xy' = − ρ u ʹvʹ τ xz' = − ρ uʹwʹ τ yz' = − ρ vʹwʹ

31

⎡ ∂u ∂u ∂u ⎤ ∂p
⎡ '
∂σ ∂τ 'xy ∂τ 'xz ⎤
ρ ⎢u +v +w =− + µ ⎡∇ 2 u ⎤ +− ρ ⎢ xx + + ⎥
⎣ ∂x ∂y ∂z ⎥⎦ ∂x ⎣⎢ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎣⎢ ⎦⎥
⎡ ∂τ ' ∂σ 'yy ∂τ 'yz ⎤
⎡ ∂v ∂v ∂w⎤ ∂p ⎡ 2 ⎤ ⎢ xy ⎥
ρ ⎢u +v +w ⎥ = − ∂ y + µ ⎢⎣∇ v ⎥⎦ −+ ρ ⎢ ∂x + ∂y + ∂z ⎥
⎣ ∂x ∂y ∂z ⎦ ⎢⎣ ⎥⎦

⎡ ∂w ∂w ∂w⎤ ∂p
⎡ '
∂τ ∂τ 'yz ∂σ 'zz ⎥

ρ ⎢u +v +w =− + µ ⎡∇ 2 w ⎤ +− ρ ⎢ xz + +
⎣ ∂x ∂y ∂z ⎥⎦ ∂z ⎢⎣ ⎥⎦ ⎢ ∂x ∂y ∂z ⎥
⎣⎢ ⎦⎥

___ ___ ___


σ ʹxx τ ʹxy τ ʹxz − ρ uʹ 2 − ρ u ʹv ʹ − ρ u ʹw ʹ
= ___ ___ ___
τ ʹxy σ ʹyy τ ʹyz − ρ uʹv ʹ − ρ v ʹ 2 − ρ v ʹw ʹ
___ ___ ___
τ ʹxz τ ʹyz σ zz − ρ uʹw ʹ − ρ v ʹw ʹ − ρ w ʹ 2

In tu rb ule nt flow, lam ina r stre ss es m ust b e increa se d by ad dit io nal stre se s –
32
REYNOLDS STRESSES (1895 - Reynolds)

Laminar flow shear stress


Turbulent flow as a series of
caused by random motion
random three dimensional eddies
of molecules

33

11
du _____
τ = τ lam + τ turb = µ − ρ uʹv ʹ
dy
_____
Laminar flow ρ uʹvʹ = 0 uʹ = 0 vʹ = 0

34

6.3 Semiempirical Turbulent Shear Correlations 335

y y

y = δ (x)
U(x)

Outer
turbulent
τ (x, y) layer

τ turb u (x, y) Overlap layer

τ lam Viscous wall layer


Fig. 6.8 Typical velocity and shear
τ w(x) 0
distributions in turbulent flow near
a wall: (a) shear; (b) velocity. (a) (b)

geometry and flow conditions, as detailed in Refs. 1 to 3. Fortunately, in duct and


!#!$!#! associated with direction y normal to the wall
boundary-layer flow, the stress !"u
is dominant, and we can approximate with excellent accuracy a simpler streamwise
momentum equation
du *p
! " !+ *&
! % "g % +
"+ x (6.15)
dt *x *y
• VISCOUS SUBLAYER : 𝞽lam >>> 𝞽 turb *u
where & ' (+! ! "u
!#!$!#! ' &lam % &turb (6.16)
*y
• OVERLAP LAYER: 𝞽lam ~ 𝞽6.8turb
Figure shows the distribution of & and &turb from typical measurements across
lam
a turbulent-shear layer near a wall. Laminar shear is dominant near the wall (the wall
• TURBULENT LAYER: 𝞽 <<< 𝞽
layer), and turbulent shear dominates in the outer layer. There is an intermediate re-
gion,lam turb
called the overlap layer, where both laminar and turbulent shear35 are important.
These three regions are labeled in Fig. 6.8.
In the outer layer &turb is two or three orders of magnitude greater than &lam, and
vice versa in the wall layer. These experimental facts enable us to use a crude but very
u(y) across a turbulent wall layer.
effective model for the velocity distribution !

The Logarithmic-Overlap Law We have seen in Fig. 6.8 that there are three regions in turbulent flow near a wall:
1. Wall layer: Viscous shear dominates.
2. Outer layer: Turbulent shear dominates.
3. Overlap layer: Both types of shear are important.

Drop overbar now From now on let us agree to drop the overbar from velocity u
onwards !. Let &w be the wall shear
stress, and let ) and U represent the thickness and velocity at the edge of the outer
layer, y ' ).
For the wall
U is independent of shear layer,thickness
layer Prandtl deduced in 1930 that u must be independent of the shear-
layer thickness
u ' f((, &w, ", y) (6.17)
u =Byfdimensional
( µ ,τ wanalysis,
, ρ , this
y)is equivalent to

u ⎛ yu * ⎞
u+ = *
=F⎜ ⎟ Law of wall
u ⎝ v ⎠
1/2
⎛τ ⎞
u* = ⎜ w ⎟ Friction velocity
⎝ρ ⎠

36

12
Subsequently, Kármán in 1933 deduced that u in the outer layer is independent of
molecular viscosity, but its deviation from the stream velocity U must depend on the
layer thickness $ and the other properties
u yu* %w 1/2
u! " ( " F (
(Uu*
u* " (
) g($, %w, &, y) &
# u)outer " ! " (6.18)
(6.19) ! "
Equation
Again, (6.18) is called
by dimensional the law
analysis weofrewrite thisand
the wall, as the quantity u* is termed the friction
velocity because it has dimensions {LT#1}, although it is not actually a flow velocity.
# u thatyu in the outer layer is independent of
Udeduced
Subsequently, Kármán in 1933(
molecular viscosity, but its deviation
"G ( !" (6.20)
u*from the $stream velocity U must depend on the
layer thickness $ and the other properties
where u* has the same meaning as in Eq. (6.18). Equation (6.20) is called the
velocity-defect law for the outer (Ulayer.
# u)outer " g($, %w, &, y) (6.19)
Both the wall law (6.18) and the defect law (6.20) are found to be accurate for a
Again, by dimensional analysis we rewrite this as
wide variety of experimental turbulent duct and boundary-layer flows [1 to 3]. They
U # u smoothly y
are different in form, yet they must overlap
! "
G ( in the intermediate layer. In (6.20)
1937
*
( yu"
C. B. Millikan showed that this can y + be ; $ if the overlap-layer velocity varies
=u*true only
logarithmically
where u* haswith
ν
the y:same meaning as in Eq. (6.18). Equation (6.20) is called the
velocity-defect law for uthe outer +yu* u
1 ulayer.
Both the wall law ( "(
(6.18) ln ( = !*B; law overlap layer
found to be accurate(6.21)
u* 'and the) defect
u (6.20) are for a
wide variety of experimental turbulent duct and boundary-layer flows [1 to 3]. They
Over the full range
are different of turbulent
in form, smooth
yet they must wall flows, theindimensionless
τ wsmoothly
* overlap the intermediateconstants and
layer. In'1937
areB.found
B C. to have
Millikan showed that thisucan=
the approximate values
be true'; # 0.41
only if and B # 5.0. Equation
the overlap-layer (6.21)
velocity is
varies
called the logarithmic-overlap
logarithmically with y: layer.
ρ
Thus by dimensional reasoning and physical insight we infer that a plot of u versus
u 1 yu*
ln y in a turbulent-shear ( layer" will
( lnshow
( a!curved
B wall region,
overlap layer a curved outer region,
(6.21)
u* ' )
and a straight-line logarithmic overlap. Figure 6.9 shows that this is exactly the case.
The
Overfourtheouter-law
full rangeprofiles shownsmooth
of turbulent all merge smoothly
wall with
flows, the the logarithmic-overlap
dimensionless constants ' law
and
are found
butB have to have
different the approximate
magnitudes because theyvalues
vary'# in 0.41
external B # 5.0.gradient.
and pressure EquationThe(6.21)
wallis
called
law the logarithmic-overlap
is unique and follows the linear layer.
viscous relation
Thus by dimensional reasoning and physical insight we infer that a plot of u versus
ln y in a turbulent-shear layer uwill u yu*
!
"show
( "a ( curved" ywall
! region, a curved outer region,
(6.22)
and a straight-line logarithmic overlap. u* Figure ) 6.9 shows that this is exactly the case.
Thethe
from fourwall
outer-law
to aboutprofiles
y! "shown all merge
5, thereafter smoothly
curving overwith the logarithmic-overlap
to merge with the logarithmiclaw
butathave
law aboutdifferent magnitudes because they vary in external pressure gradient. The wall
y! " 30.
law is unique and follows the linear viscous relation
Believe it or not, Fig. 6.9, which is nothing more than a shrewd correlation of 37 ve-
locity profiles, is the basis for most existing
u “theory”
yu* of turbulent-shear flows. Notice
u! " ( " ( " y! (6.22)
that we have not solved any equations atu*all but)have merely expressed the streamwise
velocity in a neat form. !
from the wall to about y " 5, thereafter curving over to merge with the logarithmic
There is serendipity in Fig. 6.9: The logarithmic law (6.21), instead of just being a
law at about y! " 30.
short overlapping link, actually approximates nearly the entire velocity profile, except
Believe it or not, Fig. 6.9, which is nothing more than a shrewd correlation of ve-
for the outer law when the pressure is increasing strongly downstream (as in a dif-
locity profiles, is the basis for most existing “theory” of turbulent-shear flows. Notice
fuser). Thehave
that we inner-wall lawany
not solved typically extends
equations at allover less than
but have 2 percent
merely expressedof the
the profile and
streamwise
can be neglected. 6.3 Semiempirical Turbulent Shear Correlations 337
velocity in a neatThus
form.we can use Eq. (6.21) as an excellent approximation to solve
There is serendipity in Fig. 6.9:30The logarithmic law (6.21), instead of just being a
short overlapping link, actually approximates nearly the entire velocity profile, except
Outer law profiles:
for the outer law when the pressure is increasing Strong
strongly downstream
increasing pressure (as in a dif-
Flat plate flow
fuser). The inner-wall law typically25 extends over less
Pipethan
flow 2 percent of the profile and
Strong decreasing pressure
can be neglected. Thus we can use Eq. (6.21) as an excellent approximation to solve

20 Linear u+ = y +
viscous
sublayer,
er
Eq. (6.22) lay
ap
erl
u*
u

15 Ov
u+ =

Logarithmic
overlap
Eq. (6.21)
10

Experimental data

5
er
ay
rl
ne
In

Fig. 6.9 Experimental verification


of the inner-, outer-, and overlap- 0
1 10 10 2 10 3 10 4
layer laws relating velocity profiles
y + = yu*
in turbulent wall flow. ν

nearly every turbulent-flow problem presented in this and the next chapter. Many ad-
ditional applications are given in Refs. 2 and 3.

EXAMPLE 6.3
Air at 20°C flows through a 14-cm-diameter tube under fully developed conditions. The cen-
terline velocity is u0 ! 5 m/s. Estimate from Fig. 6.9 (a) the friction velocity u*, (b) the wall
shear stress "w, and (c) the average velocity V ! Q/A.

Solution
Part (a) For pipe flow Fig. 6.9 shows that the logarithmic law, Eq. (6.21), is accurate all the way to the
Head loss – friction factor center of the tube. From Fig. E6.3 y ! R # r should go from the wall to the centerline as shown.
u ( y) At the center u ! u0, y ! R, and Eq. (6.21) becomes
u0 = 5 m/s
y=R u0 1 Ru*
' ! ' ln ' $ 5.0 (1)
u* 0.41 %
y
4τ L
Since we know that u0 ! 5 m/s w and R ! 0.07 m, u* is the only unknown in Eq. (1). Find the
r
Using steady flow energy equation
h =
solution by trialf and error or by EES
r = R = 7 cm ρ g du* ! 0.228 m/s ! 22.8 cm/s 8τ w Ans. (a)
f =
where we have taken % ! 1.51 & 10 m /s for air from Table 1.4. 2
ρV
#5 2
E6.3
2
LV
Empirical correlation by Weisbach h = f
f
d 2g
f = Darcy friction factor

64
flam =
Re

39

13
Friction Factor
Pressure force = Viscous forces

π L
ΔP × D 2 = τ w × π D L ⇒ ΔP = τ w × 4 ×
4 D

ΔP τw L
= × 4×
1 2 1 2 D
ρuavg ρuavg
2 2
ΔP L
= 4C f 4Cf = f
1 2 D
ρuavg
2
ΔP L ρghL L
= f ⇒ 1 2 = f D
1 2 D ρu
ρu 2 avg
2 avg
Darcy–Weisbach equation 2
f Luavg
hL =
2 gD 40

FRICTION FACTOR FOR TURBULENT FLOW IN A


HYDRAULICALLY SMOOTH PIPE
R
Q = π R 2 uavg = ∫ 2π r u dr r y
0
0 r= R–y
Q= ∫ 2π (R − y ) u( −dy ) dr = -dy
R
R
u*
πR 2 uavg = ∫ 2π (R − y )u( dy ) Multiply by
ν2
0
R
Chapter 6 Viscous Flow in Ducts
2 u* u*
R uavg = ∫ 2( R − y)uν (dy )
ν2 0
2

1/2
u yu*
! " ! "
! %w
2
u "("F ( u* " ( (6.18)
⎛ Ru * ⎞ uavg R
Ru ⎛u* *
yu ⎞ u ⎛ yu&* ⎞
)*
⎜ (6.18)
Equation
= the
⎟ is called ∫ 2 law − wall,yand
⎜ of the ⎟ d⎜ ⎟
⎝ ν ⎠ u
*
0 ⎝
ν ν ⎠ u *the ⎝quantity
ν ⎠ u* is termed the friction
velocity because it has dimensions {LT#1}, although it is not actually a flow velocity.
41

Subsequently, Kármán in 1933 deduced that u in the outer layer is independent of


molecular viscosity, but its deviation from the stream velocity U must depend on the
layer thickness $ and the other properties
(U # u)outer " g($, %w, &, y) (6.19)
Again, by dimensional analysis we rewrite this as
U#u y
("G (
u* $ !" (6.20)
2 R
where u* has ⎛the ⎞ umeaning
Rusame *
avg as⎛ Ru
*
yu ⎞Equation
in Eq. (6.18).
*
u ⎛ yu(6.20) *
⎞ is called the
velocity-defect ⎜law for⎟the outer = 2⎜ ∫ − y⎟ * d ⎜ ⎟
⎝ ν ⎠ u * layer. ν ν u ν
0 defect law (6.20) are found to ⎠be accurate for a
⎝ ⎠ ⎝
Both the wall law (6.18) and the
wide variety of experimental turbulent duct and boundary-layer flows [1 to 3]. They
are different in form, yet they
u mustRoverlap+smoothly
+
in the intermediate layer. In 1937
C. B. Millikan showed R + 2thatavg
*
this ∫ (
= can 2be Rtrue−only
y + ifu +the )
dyoverlap-layer
+
velocity varies
logarithmically with y: u 0

u 1 yu*
( " ( ln ( ! B overlap layer (6.21)
u* ' )
Over the full range of turbulent smooth wall flows, the dimensionless constants ' and
B are found to have the approximate values ' # 0.41 and B # 5.0. Equation (6.21) is
called the logarithmic-overlap
u layer.
R+
Thus by dimensionalR + 2 avg reasoning
*
= ∫ (2andR +physical )
− y + ⎡⎣insight
A ln ywe+
B ⎤⎦ that
+infer dy +a plot of u versus
ln y in a turbulent-shearu layer will 0 show a curved wall region, a curved outer region,
and a straight-line logarithmic overlap. Figure 6.9 shows that this is exactly the case.
The four outer-law profiles shown all merge +smoothly withR the logarithmic-overlap
+ +
R law
uavg R +
⎧y 2 y+2 ⎫ R +
y+2
but have
R + 2 different
= 2 Rmagnitudes
+
A { y + ln y + −because
y + } − 2they
A ⎨ vary
ln in external
y+ − ⎬ pressure
+ 2 BR + y + gradient.
− 2 B The wall
u* 0 2 4 ⎭0 0 2 0
law is unique and follows the linear viscous relation ⎩

u yu*
u! " ( " ( " y! (6.22)
42
u* )
!
from the wall to about y " 5, thereafter curving over to merge with the logarithmic
law at about y! " 30.
Believe it or not, Fig. 6.9, which is nothing more than a shrewd correlation of ve-
locity profiles, is the basis for most existing “theory” of turbulent-shear flows. Notice
that we have not solved any equations at all but have merely expressed the streamwise
velocity in a neat form.
There is serendipity in Fig. 6.9: The logarithmic law (6.21), instead of just being a
short overlapping link, actually approximates nearly the entire velocity profile, except
for the outer law when the pressure is increasing strongly downstream (as in a dif-
fuser). The inner-wall law typically extends over less than 2 percent of the profile and
14
can be neglected. Thus we can use Eq. (6.21) as an excellent approximation to solve
Integration by parts

∫ udv = uv − ∫ vdu
1
∫ ln xdx = ∫ ln x dx = (ln x )x − ∫ x x dx = x ln x − x + C
u v

⎛ x2 ⎞ x2 1 x2 x2
∫ x ln xdx = ln x⎜⎜⎝ ⎟−
2 ⎟⎠ ∫ 2 x
dx =
2
ln x −
4
+C

x2
xdx = dv ⇒ v = ; u = ln x
2

43

R+ R+
uavg R+ ⎧ y+2 y+2 ⎫ R+ y+2
R+2 *
= 2 R + A { y + ln y + − y + } − 2A⎨ ln y + − ⎬ + 2 BR y
+ +
− 2B
u 0
⎩ 2 4 ⎭0 0 2 0

uavg ⎧ R+2 R+2 ⎫ R+2


R+2 = 2 R + A {R + ln R + − R + } − 2 A ⎨ ln R + − +2
⎬ + 2 BR − 2 B
u* ⎩ 2 4 ⎭ 2

uavg ⎡ 1 1 1⎤
R+2 = R + 2 ⎢ 2 A ln R + − 2 A − 2 A ln R + + 2 A + 2 B − 2 B ⎥
u* ⎣ 2 4 2⎦

uavg ⎡ 1 ⎤
= ⎢ 2 A ln R + − 2 A − A ln R + + A + 2 B − B ⎥
u* ⎣ 2 ⎦
uavg ⎡ 3 ⎤
= ⎢ A ln R + − A + B ⎥ A = 2.5 B = 5.5
u* ⎣ 2 ⎦
uavg
= ⎡⎣ 2.5ln R + + 1.75⎤⎦
u* 44

2
uavg uavg ρ uavg 2 8
*
= = = = uavg
u τw τw cf f = ⎡⎣ 2.5ln R + + 1.75⎤⎦
ρ u*

uavg 8
= = 2.5ln R + + 1.75
u* f

8 ⎛ Ru * ⎞
= 2.5ln ⎜ ⎟ + 1.75
f ⎝ ν ⎠
8 ⎛ 2 Ruavg u * ⎞
= 2.5ln ⎜ + 1.75
f ⎜ ν 2uavg ⎟⎟
⎝ ⎠

1 2.5 ⎛ 1 f ⎞⎟ 1.75
= ln⎜ Re +
f 8 ⎜ 2 8 ⎟⎠ 8

45

15
1 2.5 2.5 ⎛ 1 ⎞ 1.75
f
=
8
(
ln Re )
f +
8
ln⎜⎜ ⎟⎟ +
⎝2 8⎠ 8

1
f
= 0.8838 ln Re ( )
f − 0.9129

Converting to base-10 log


log b = ln b . log e
log Re f log b
ln ⎡⎣ Re f ⎤⎦ = = 2.30258log Re f ln b =
log e log e
1
f
= 2.035log Re ( )
f − 0.9129

PRANDTL’S UNIVERSAL LAW OF FRICTION, adjusted constants


slightly to fit friction data better

1
f
= 2.0 log Re ( )
f − 0.8 – verified Nikuradse’s expmts
upto Re = 3.4 x 106
46

Blasius Equation

Since f appears on both sides of the Prandtl law,


Blasius proposed the following:

⎛ ⎞ −1 / 4
⎜ ΔP ⎟ D ⎡ ρVD ⎤
⎜ ⎟ = f = 0.3164 ⎢ ⎥
⎜ 1 ρV 2 ⎟ L ⎣ µ ⎦
⎜ ⎟
⎝2 ⎠

f = 0.316 Re −1/4 4000 < Re < 105

47

Using Blasius equation

f = 0.316 Re −1/4

64 L V2
Instead of f = in hf = f
Re d 2g

µ LQ 7/4
h f = 0.241
ρ 1/4 gD19/4

Turbulent Laminar

h f αV 1.75 h f αV
hf α D −4.75 h f α D −4
Verified Hagen’s
48
experiments

16
1
f
= 2.0 log Re ( )
f − 0.8 f = 0.316 Re −1/4

Re Prandtl Equation Blasius Equation


1000 0.0622 0.0567
2000 0.0494 0.0473
5000 0.0374 0.0376

10000 0.0309 0.0316


20000 0.0259 0.0266
50000 0.0209 0.0212

100000 0.018 0.0178


200000 0.0156 0.015
500000 0.0131 0.0119
1000000 0.0116 0.01
2000000 0.0104 0.008414

5000000 0.009 0.0069 49


10000000 0.0081 0.0056

Pran dtl’s u niv ersa l eq uat io n ag ree s we ll with Bla sius e quat io n up to R e = 1 0000 0 but
at h igh er va lu es B la sius eq uat ion d eviat es pr ogre ss ive ly m ore fr om t he re su lts of
measurements whereas Prandtl’s equation maintains good agreement

Paul Richard Heinrich Blasius (1883 – 1970)

Johann Nikuradse (1894 –1979)

Hager, Blasius: A life in research and education, Experiments in Fluids 34 (2003) 566–571
Hager, Johann N ikuradse – Hydraulic experimenter, Journal of Hydraulic Research, 50 V ol. 46,
No. 4 (2008), pp. 435–444

Zagarola and Smits (1998)

Reynolds numbers from 31000 – 35 x 106

1 7.04
= 1.92 log Re ( )
f − 0.475 −
f (Re D f )
0.55

Predicted 1.4% of the Princeton’s experimental data

Mark V. Zagarola. and Alexander J. Smits, Mean flow scaling of turbulent pipe flow, J.
Fluid Mech. (1998), Vol. 373, pp. 33-79
B.J. McKeon, M.V. Zagarola and A.J.Smits, A new friction factor relationship for
fully developed pipe flow, J. Fluid Mech. (2005), Vol. 538, pp. 429-443. 51

17
Example
D = 50 mm µ = 0.001 Pa Thickness of viscous sublayer?
L = 10 m ρ = 1000 kg / m 3

Re = 100000
1
Re =
ρuavg D
µ f
= 2.0 log Re ( )
f − 0.8
1000 × uavg × 50 × 10 −3
100000 =
0.001 Re = 100000 f = 0.018

uavg = 2 m / s

τw τw τ w = 9 Pa τw 9
f = 4C f = 4 ⇒ 0.018 = 4 u* = = = 0.09487m / s
1 2 1 ρ 1000
ρu × 1000 × 2 2
2 avg 2
uτ = 0.09487 m / s
9
* y
yu 1000
y+ = ⇒5=
ν 0.001 y = 5.27 × 10 −5 m = 0.0527mm 52
1000

EMPIRICAL POWER LAW VELOCITY DISTRIBUTION

1 1
u u ⎛ r⎞ ⎛ y⎞
= = ⎜1 − ⎟n = ⎜ ⎟n r y
umax uc ⎝ R⎠ ⎝ R⎠
r= R–y
uavg ∫ u dA
= dr = -dy
uc Auc
R
2πuc 1/ n 1/ n
R R ∫ udA = ∫ (Ry ) − y( y ) dy
∫ udA = 2π ∫ r u dr = 2π ∫ u ( R − y ) dy R1 / n 0
0 0 R
⎡ n+ 1 n+ 1 ⎤
1 2πuc ⎢ R[Y ] n [Y ] n + 1 ⎥
R
⎛ y⎞
∫ udA = R 1 / n

n + 1

n + 1

∫ udA = 2πuc ∫ ⎜⎝ R ⎟⎠ n ( R − y ) dy
⎢ +1 ⎥
⎣⎢ n n ⎦⎥
0 0

R
2πuc 1/ n
∫ udA = ∫ ( y ) ( R − y ) dy
R1 / n 0 53

2n + 1
⎡ 2n + 1 n + 1 ⎤
R 2πuc R n ⎢ n − n ⎥
⎡ n+ 1 n+ 1 ⎤ ∫ udA = ⎢ ⎥
2πu ⎢ R[Y ] [Y ] + 1 ⎥ R n
1 / ⎢ n + 1 . 2n + 1 ⎥
∫ udA = R1 / cn ⎢⎢ n+ 1n − n + n1 ⎥⎥ ⎣⎢ n n ⎦⎥
+1
⎢⎣ n n ⎥⎦
0 ⎡ 2n + 1 − n − 1 ⎤
⎢ ⎥
∫ udA = 2πuc R 2 ⎢ n
⎡ n+ 1 2n + 1 ⎤ n + 1 2n + 1 ⎥
2πuc ⎢ R[R ] n [R] n ⎥⎥ ⎢ . ⎥
⎣ n n ⎦
∫ udA = 1 / n

⎢ n + 1

2 n +1 ⎥
R

⎣ n n ⎥
⎦ ⎡ n ⎤
⎢ ⎥
∫ udA = 2πuc R 2 ⎢ n

⎡ 2n + 1

2n + 1 ⎤ ⎢ n + 1 . 2n + 1 ⎥
2πuc ⎢ R n [R] n ⎥⎥ ⎣⎢ n n ⎦⎥
∫ udA = 1 / n ⎢ n + 1

2 n +1 ⎥
R ⎢ ⎥
⎣⎢ n n ⎦⎥ 2
⎡ 2n 2 ⎤
∫ udA = πuc R ⎢⎢ [n + 1][2n + 1] ⎥⎥
2n + 1 ⎣ ⎦
⎡ ⎤
2πuc R n ⎢ 1 1 ⎥
∫ udA = ⎢ − ⎥
R1 / n ⎢ n+ 1 2n + 1 ⎥
⎢ n n ⎥⎦

54

18
2n 2
πuc R 2
uavg
=
∫ u dA =
(n + 1)(2n + 1) =
2n 2
2
uc Auc πR uc (n + 1)(2n + 1)

uavg 2n 2
=
uc (n + 1)(2n + 1)

55

1
n = 10
n=6

n=8
Laminar

r 0.5
R

Tur bulent

0
0.5 1
u
Vc

56

11

10

n 8

5
104 105 106

ρVD
Re =
µ

57

19
Comparison of laminar and turbulent Velocity profiles for turbulent flow in
velocity profile smooth circular tubes

58

1 1
u u ⎛ r⎞ ⎛ y⎞
= = ⎜1 − ⎟n = ⎜ ⎟n
umax uc ⎝ R⎠ ⎝ R⎠

59

FRICTION FACTOR FOR TURBULENT FLOW IN A HYDRAULICALLY


ROUGH PIPE
• Roughness disturbs the flow in the wall region
• Roughness elements are large enough, they break the viscous
sublayer
ε u*
ε+ = <5 Pipe is hydraulically smooth
ν

ε u*
ε+ = >5 Pipe is hydraulically rough
ν

60

20
61

⎛ ⎞
⎜ ΔP ⎟ L ⎛ ε ρVD ⎞
⎜ ⎟ = φ⎜ ,
⎜1 ⎜ ⎟⎟
2⎟ D ⎝D µ ⎠
⎜ ρV ⎟
⎝2 ⎠

⎛ ⎞
⎜ ΔP ⎟ D ⎛ ε ρVD ⎞
⎜ ⎟ = φ ⎜⎜ , ⎟⎟
⎜ 1 ρV 2 ⎟ L ⎝D µ ⎠
⎜ ⎟
⎝2 ⎠

⎛ ε ρVD ⎞
f = φ ⎜⎜ , ⎟⎟
⎝D µ ⎠

ε uτ ε uτ
ε+ = >5 ε+ = <5
ν ν
HYDRAULICALLY ROUGH HYDRAULICALLY SMOOTH 62

⎛ y⎞
u + = 2.44 ln ⎜ ⎟ + 8.5
⎝ε ⎠

63

21
R
Q = π R 2 uavg = ∫ 2π r u dr r y
0
0 r= R–y
Q= ∫ 2π (R − y ) u( −dy ) dr = -dy
R
R
πR 2 uavg = ∫ 2π (R − y )u( dy )
0
uavg R
u
R2 = ∫ 2 (R − y ) ( dy )
uτ uτ
0

uavg R
R2 = ∫ 2 (R − y ) u + ( dy )

0
64

uavg R ⎛ y⎞
u + = A ln ⎜ ⎟ + C
R2 = ∫ 2 (R − y ) u + ( dy ) ⎝ε ⎠

0
uavg R
R2 = ∫ 2 (R − y )( A ln y − A ln ε + C )( dy )

0
uavg R
R2 = 2 ∫ RA ln y − Ay ln y − AR ln ε + Ay ln ε + CR − Cy )dy

0
⎡ R R R⎤
uavg 2 y2 2 y2 ⎥
2 ⎢ R Ay R Ay R
R = 2 ⎢ ARy ln y − ARy − ln y + A − ARy ln ε + ln ε + CRy − C
uτ o 2 4 0 2 0 2 ⎥
⎢ 0 0 0 ⎥⎦

⎡ ⎤
uavg AR 2 AR 2 AR 2 CR 2
R2 = 2 ⎢⎢ AR 2 ln R − AR 2 − ln R + − AR 2 ln ε + ln ε + CR 2 − ⎥

uτ 2 4 2 2
⎣⎢ ⎦⎥

uavg ⎡ A A A C ⎤
= 2 ⎢ A ln R − A − ln R + − A ln ε + ln ε + C − ⎥
uτ 2 4 2 2 ⎥
⎣⎢ ⎦ 65

uavg ⎡A 3A A C ⎤
= 2 ⎢ ln R − − A ln ε − ln ε + ⎥
uτ ⎢⎣ 2 4 2 2 ⎥

uavg R 3A A = 2.44; C = 8.5
= A ln − +C
uτ ε 2

uavg uavg 2
ρuavg 2 8
= = = =
uτ τw τw cf f
ρ

8 R 1 R
= 1.98637 log + 1.7112
= 2.44 ln + 4.84 f ε
f ε

1 R
= 0.86267 ln + 1.7112
f ε 66

22
1 R 1.1132 = − 1.9863 log x
= 1.98637 log + 1.7112
f ε 1.1132
log x = = − 0.560460679
− 1.9863
1 D 1
= 1.98637 log + 1.98637 log + 1.7112 x = 10 − 0. 560460679 = 0.275130869
f ε 2
1
= 3.6346
1 ε x
= − 1.98637 log + 1.1132
f D
⎡ ε ⎤
1 ⎢ ⎥
= − 1.98637 log ⎢ D ⎥
f ⎢ 3.6346 ⎥
⎣⎢ ⎦⎥

⎡ ε ⎤
1 ⎢ ⎥
= − 2.0 log ⎢ D ⎥
VonKarman’s Relation
f ⎢ 3 .7 ⎥
⎣ ⎦ 67

1
f
= 2.0 log Re ( )
f − 0.8 ; Smooth Pipe

1 ⎛ 2.5 ⎞
1 ⎛ 1 ⎞
is equivalent to = − 2.0 log⎜ ⎟
= − 2.0 log⎜ ⎟ − 0.8
f ⎜ Re f ⎟
f ⎜ Re f ⎟ ⎝ ⎠
⎝ ⎠

Combining smooth and rough pipe friction factors


⎡ ε ⎤
1 ⎛ 2.5 ⎞ 1 ⎢ ⎥
= − 2.0 log⎜ ⎟ = − 2.0 log ⎢ D ⎥
f ⎜ Re f ⎟ f ⎢ 3 .7 ⎥
⎝ ⎠ ⎣ ⎦

⎛ ε ⎞ COLEBROOK’S
1 ⎜ 2.5 ⎟ FRICTION FACTOR
= − 2.0 log⎜ D + ⎟
CORRELATION
f ⎜ 3.7 Re f ⎟
⎜ ⎟
⎝ ⎠ 68

⎡ εuτ ν ⎤
⎢ ν . Du
ε+
⎥ ⎡ ⎤
1
= − 2.0 log ⎢ τ ⎥ 1
f ⎢ 3.7 ⎥ = − 2.0 log ⎢ ⎥
⎢ ⎥ f ⎢⎣ 1.31 Re f ⎥⎦
⎣ ⎦
⎡ + ν uavg ⎤
⎢ε . . ⎥
1 ⎢ Duτ uavg ⎥
= − 2.0 log ⎢ ⎥
f 3.7
⎢ ⎥
⎢⎣ ⎥⎦
⎡ + ν uavg ⎤
⎢ε . . ⎥
1 ⎢ Duavg uτ ⎥
= − 2.0 log ⎢ ⎥
f 3.7
⎢ ⎥
⎢⎣ ⎥⎦
uavg 8
=
uτ f
⎡ + ν 8 ⎤
⎢ε . . ⎥
1 ⎢ Duavg f ⎥
= − 2.0 log ⎢ ⎥
f 3.7 69
⎢ ⎥
⎣⎢ ⎥⎦

23
MOODY CHART
The author does not claim to offer anything particular ly new or or iginal, h is aim merely
70 be ing
to embody the now accepted conclusions in convenient form for engineering use.

Equivalent Roughness for New Pipes [From Moody and Colebrook ]

Pipe Equivalent Roughness (mm)


Riveted steel 0.9–9.0
Concrete 0.3–3.0
Wood stave 0.18–0.9
Cast iron 0.26
Galvanized iron 0.15

Commercial steel or wrought iron 0.045


Drawn tubing 0.0015

Plastic, glass 0.0 (smooth)

71

Air u nder sta ndar d con dition s flo ws th roug h a 4.0-mm- diamete r dra wn tub ing with
an average ve loc ity of 5 0 m/s. F or s uch con dit io ns the f lo w wou ld nor mally be
turb ulent. H owev er, if precaut io ns are taken to elim inate d ist urb ances to the flow
(the entra nce t o the tu be is ve ry smo oth, the a ir is dus t fr ee, t he tub e doe s not
vibrat e, etc.), it may be p os s ib le to ma inta in lam inar f lo w. a Dete rm ine t he pre s sure
dro p in a 0.1-m sect io n of the t ube if t he flow is la mina r. b Repe at the calcu lat io ns if
the flow is turbulent

ρVD 1.225 × 50 × 4 × 10 −3
Re = = = 13215
µ 18.54 × 10 −6
If flow is laminar
64 64
f = = = 4.843 × 10 − 3
Re 13215

fLV 2 ρfLV 2 1.225 × 4.843 × 10 −3 × 0.1 × 50 2


hf = ⇒ ΔP = = = 185 Pa
2 gD 2D 2 × 4 × 10 − 3

π
P = Q! ΔP =
2

4
( )
× 4 × 10 − 3 × 50 × 185 = 0.1162 W
72

24
ε 0.0015
ε = 0.0015mm ⇒ = = 3.75 × 10 − 4
D 4
ρVD 1.225 × 50 × 4 × 10 −3
Re = = = 13215
µ 18.54 × 10 − 6

Moody Chart f = 0.028


fLV 2 ρfLV 2 1.225 × 0.028 × 0.1 × 50 2
hf = ⇒ ΔP = = = 3062.5 Pa
2 gD 2D 2 × 4 × 10 − 3
⎛ ε ⎞
1 ⎜ 2.5 ⎟ 1 ⎛ 3.75 × 10 − 4 2.5 ⎞⎟
= − 2.0 log⎜ D + ⎟ = − 2.0 log⎜ +
⎜ 3.7 Re f ⎜
f


⎟ f ⎝ 3.7 13215 f ⎟⎠
⎝ ⎠
f = 0.028 LHS = 5.976 RHS = 5.8188
f = 0.029 LHS = 5.872 RHS = 5.833 f = 0.0293
f = 0.0291 LHS = 5.862 RHS = 5.834
f = 0.0292 LHS = 5.852 RHS = 5.836
f = 0.0293 LHS = 5.842 RHS = 5.837 73

fLV 2 ρfLV 2 1.225 × 0.0293 × 0.1 × 50 2


hf = ⇒ ΔP = = = 3204.69 Pa
2 gD 2D 2 × 4 × 10 − 3

π
P = Q! ΔP = × 4 × 10 − 3 × 50 × 3204.69 = 2.014 W
2

4
( )

74

Non-circular pipes
Circular pipe

Circular pipes can withstand


large pressure differences
between the inside and the
outside without undergoing
Water
any distortion, but the
50 atm
noncircular pipes cannot.
Rectangular
duct

Air
1.2 atm 75

25
MEAN VELOCITY

V=0 m& = ρ Vm Ac =
∫ ρ V ( r,x ) dA
Ac
c

R
Vmax
∫ ρ V ( r,x ) dA ∫ ρ V ( r,x ) 2π rdr
Ac
c
2
R
0
Vm =
ρ Ac
=
ρ π R2
=
R2 ∫V ( r,x ) rdr
0
(a) Actual

Vm

(b) Idealized

Actual an d idea lized ve locity pr ofile s f or


flo w in a t ube (t he ma ss f lo w rate of t he
fluid is the same for both cases).
76

NON-CIRCULAR CONDUITS

•Calculations for fully developed turbulent flow in ducts in non-circular


cross section are usually carried out byusing the Moody chart data
for round pipes with the diameter replaced by the hydraulic
diameter and Reynolds number based on the hydraulic diameter.

•Such calculations are usually accurate to within about 15%. If greater


accuracy is needed, a more detailed analysis based on the
77
specific geometry of interest is needed.

78

26
HYDRAULIC DIAMETER
4 Ac
Circular tube:
Dh =
Dh =
(
4 π D2 4 )=D P
D πD

Square duct:

4a 2
Dh = =a
4a
a
a

Rectangular duct:

4ab 2ab
Dh = =
2 (a + b) a + b
a
b 79

MINOR LOSSES- losses caused by valves, bends, tees etc


MAJOR LOSSES- losses caused by the friction of pipe
ΔP
KL =
1
ρV 2
2
ΔP ρghL gh L
KL = = =
1 1 1 2
ρV 2 ρV 2 V
2 2 2

V2
hL = K L
2g

Flow through a valve 80

ENTRANCE FLOW CONDITIONS AND LOSS COEFFICIENTS

KL = 0.8 Re-entrant KL = 0.5 Sharp Edged

KL = 0.2 Slightly rounded KL = 0.04 well rounded


81

27
FLOW PATTERN AND PRESSURE DISTRIBUTION FOR A
SHARP EDGE ORIFICE

Lost
because of
viscous
dissipation

82
Venacontracta Is Often Developed At The Entrace Of The Pipe

ENTRANCE LOSS COEFFICIENT AS A FUNCTION OF


ROUNDING OF THE INLET EDGE

83

EXIT FLOW CONDITIONS AND LOSS COEFFICIENTS

KL = 1.0 Re-entrant
KL = 1.0 Sharp Edged

84
KL = 1.0 Slightly rounded KL = 1.0 well rounded

28
LOSS COEFFICIENT FOR A SUDDEN CONTRACTION
85

LOSS COEFFICIENT FOR A SUDDEN EXPANSION

86

Loss coefficient of
pipe components

87

29
Water at 1 0 d eg C (ν= 1.3 07 x 10-6 m2/s) is to f lo w f rom re serv oir A to re serv oir B
thro ugh a ca st-ir on p ipe (𝞮 = 0.26 mm) of lengt h 2 0 m at a rate of 0.002 m3/s as
sho wn in F ig. The syst em conta in s a shar p-edg ed e ntra nce a nd six r egu lar t hrea ded
90° elbows. Determine the pipe diameter needed.

88

p1 V12 p V2
+ + z1 = 2 + 2 + z2 + hL
ρg 2g ρg 2 g

p1 = p2 = V1 = V2 = z2 = 0
V2 ⎛ L ⎞
z1 = ⎜f + ∑K L ⎟
2g ⎝ D ⎠

Q 4Q! 4 × 2 × 10 −3 2.55 × 10 −3
V= = = =
A πD 2 πD 2 D2

K Lent = 0.5; K Lelbow = 1.5; K Lexit = 1


2
V ⎛ 20 ⎞
2= ⎜f
2 × 9.81 ⎝ D
+ ∑ (0.5 + 6(1.5) + 1.0)⎟⎠
2
1 ⎛ 2.55 × 10 − 3 ⎞ ⎛ 20 ⎞
2= ⎜
2 × 9.81 ⎜⎝ D2
⎟⎟ ⎜ f
⎠ ⎝ D
+ ∑ (0.5 + 6(1.5) + 1.0)⎟⎠
89

6.03 × 106 D 5 − 10.5 D − 20 f = 0

2.55 × 10 −3
VD D 1.95 × 10 3
Re = = D2 =
ν 1.307 × 10 − 6 D

ε 2.6 × 10 −4
=
D D

Assu ming f rict ion facto r wo uld lea d us to so lving f ifth o rde r e qua tion for gett ing
diameter of the pipe . Hence, assume diameter

D = 0.05 m 6.03 × 106 D 5 − 10.5 D − 20 f = 0


−4 −4
ε 2.6 × 10 2.6 × 10 5
= = = 5.2 × 10 − 3 6.03 × 10 6 (0.05 ) − 10.5(0.05 ) − 20 f = 0
D D 0.05
1.95 × 10 3 f = 0.068
Re = = 3.9 × 10 4
0.05
Based on 𝞮/D and Re, from Moody chart, f = 0.033
90

30
D = 0.045 m 6.03 × 106 D 5 − 10.5 D − 20 f = 0
−4 −4
ε 2.6 × 10 2.6 × 10
= = = 5.78 × 10 − 3 6.03 × 10 6 (0.045 )5 − 10.5(0.045 ) − 20 f = 0
D D 0.045
1.95 × 10 3 f = 0.032
Re = = 4.33 × 10 4
0.045

Based on 𝞮/D and Re, from Moody chart, f = 0.032

91

31

Vous aimerez peut-être aussi