FLOW IN PIPES
F
luid flow in circular and noncircular pipes is commonly encountered in
practice. The hot and cold water that we use in our homes is pumped
through pipes. Water in a city is distributed by extensive piping net-
works. Oil and natural gas are transported hundreds of miles by large
pipelines. Blood is carried throughout our bodies by arteries and veins. The
cooling water in an engine is transported by hoses to the pipes in the radia-
tor where it is cooled as it flows. Thermal energy in a hydronic space heat-
ing system is transferred to the circulating water in the boiler, and then it is
transported to the desired locations through pipes.
Fluid flow is classified as external and internal, depending on whether the
fluid is forced to flow over a surface or in a conduit. Internal and external
flows exhibit very different characteristics. In this chapter we consider inter-
nal flow where the conduit is completely filled with the fluid, and flow is
driven primarily by a pressure difference. This should not be confused with
open-channel flow where the conduit is partially filled by the fluid and thus
the flow is partially bounded by solid surfaces, as in an irrigation ditch, and
flow is driven by gravity alone.
We start this chapter with a general physical description of internal flow
and the velocity boundary layer. We continue with a discussion of the
dimensionless Reynolds number and its physical significance. We then dis-
cuss the characteristics of flow inside pipes and introduce the pressure drop
correlations associated with it for both laminar and turbulent flows. Then
we present the minor losses and determine the pressure drop and pumping
power requirements for real-world piping systems. Finally, we present an
overview of flow measurement devices.
321
CHAPTER
8
OBJECTIVES
When you finish reading this chapter, you
should be able to
■ Have a deeper understanding of
laminar and turbulent flow in
pipes and the analysis of fully
developed flow
■ Calculate the major and minor
losses associated with pipe
flow in piping networks and
determine the pumping power
requirements
■ Understand the different velocity
and flow rate measurement
techniques and learn their
advantages and disadvantages
cen72367_ch08.qxd 11/4/04 7:13 PM Page 321
8–1
■
INTRODUCTION
Liquid or gas flow through pipes or ducts is commonly used in heating and
cooling applications and fluid distribution networks. The fluid in such appli-
cations is usually forced to flow by a fan or pump through a flow section.
We pay particular attention to friction, which is directly related to the pres-
sure drop and head loss during flow through pipes and ducts. The pressure
drop is then used to determine the pumping power requirement. A typical
piping system involves pipes of different diameters connected to each other
by various fittings or elbows to route the fluid, valves to control the flow
rate, and pumps to pressurize the fluid.
The terms pipe, duct, and conduit are usually used interchangeably for
flow sections. In general, flow sections of circular cross section are referred
to as pipes (especially when the fluid is a liquid), and flow sections of non-
circular cross section as ducts (especially when the fluid is a gas). Small-
diameter pipes are usually referred to as tubes. Given this uncertainty, we
will use more descriptive phrases (such as a circular pipe or a rectangular
duct) whenever necessary to avoid any misunderstandings.
You have probably noticed that most fluids, especially liquids, are trans-
ported in circular pipes. This is because pipes with a circular cross section
can withstand large pressure differences between the inside and the outside
without undergoing significant distortion. Noncircular pipes are usually
used in applications such as the heating and cooling systems of buildings
where the pressure difference is relatively small, the manufacturing and
installation costs are lower, and the available space is limited for ductwork
(Fig. 8–1).
Although the theory of fluid flow is reasonably well understood, theoreti-
cal solutions are obtained only for a few simple cases such as fully devel-
oped laminar flow in a circular pipe. Therefore, we must rely on experimen-
tal results and empirical relations for most fluid flow problems rather than
closed-form analytical solutions. Noting that the experimental results are
obtained under carefully controlled laboratory conditions and that no two
systems are exactly alike, we must not be so naive as to view the results
obtained as “exact.” An error of 10 percent (or more) in friction factors cal-
culated using the relations in this chapter is the “norm” rather than the
“exception.”
The fluid velocity in a pipe changes from zero at the surface because of
the no-slip condition to a maximum at the pipe center. In fluid flow, it is
convenient to work with an average velocity V
avg
, which remains constant in
incompressible flow when the cross-sectional area of the pipe is constant
(Fig. 8–2). The average velocity in heating and cooling applications may
change somewhat because of changes in density with temperature. But, in
practice, we evaluate the fluid properties at some average temperature and
treat them as constants. The convenience of working with constant proper-
ties usually more than justifies the slight loss in accuracy.
Also, the friction between the fluid particles in a pipe does cause a slight
rise in fluid temperature as a result of the mechanical energy being con-
verted to sensible thermal energy. But this temperature rise due to frictional
heating is usually too small to warrant any consideration in calculations and
thus is disregarded. For example, in the absence of any heat transfer, no
322
FLUID MECHANICS
Circular pipe
Rectangular
duct
Water
50 atm
Air
1.2 atm
FIGURE 8–1
Circular pipes can withstand large
pressure differences between the
inside and the outside without
undergoing any significant distortion,
but noncircular pipes cannot.
V
avg
FIGURE 8–2
Average velocity V
avg
is defined as the
average speed through a cross section.
For fully developed laminar pipe flow,
V
avg
is half of maximum velocity.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 322
noticeable difference can be detected between the inlet and outlet tempera-
tures of water flowing in a pipe. The primary consequence of friction in
fluid flow is pressure drop, and thus any significant temperature change in
the fluid is due to heat transfer.
The value of the average velocity V
avg
at some streamwise cross-section is
determined from the requirement that the conservation of mass principle be
satisfied (Fig. 8–2). That is,
(8–1)
where m
.
is the mass flow rate, r is the density, A
c
is the cross-sectional area,
and u(r) is the velocity profile. Then the average velocity for incompressible
flow in a circular pipe of radius R can be expressed as
(8–2)
Therefore, when we know the flow rate or the velocity profile, the average
velocity can be determined easily.
8–2
■
LAMINAR AND TURBULENT FLOWS
If you have been around smokers, you probably noticed that the cigarette
smoke rises in a smooth plume for the first few centimeters and then starts
fluctuating randomly in all directions as it continues its rise. Other plumes
behave similarly (Fig. 8–3). Likewise, a careful inspection of flow in a pipe
reveals that the fluid flow is streamlined at low velocities but turns chaotic
as the velocity is increased above a critical value, as shown in Fig. 8–4. The
flow regime in the first case is said to be
laminar, characterized by smooth
streamlines and highly ordered motion, and
turbulent in the second case,
where it is characterized by velocity fluctuations and highly disordered
motion. The
transition from laminar to turbulent flow does not occur sud-
denly; rather, it occurs over some region in which the flow fluctuates
between laminar and turbulent flows before it becomes fully turbulent. Most
flows encountered in practice are turbulent. Laminar flow is encountered
when highly viscous fluids such as oils flow in small pipes or narrow
passages.
We can verify the existence of these laminar, transitional, and turbulent
flow regimes by injecting some dye streaks into the flow in a glass pipe, as
the British engineer Osborne Reynolds (1842–1912) did over a century ago.
We observe that the dye streak forms a straight and smooth line at low
velocities when the flow is laminar (we may see some blurring because of
molecular diffusion), has bursts of fluctuations in the transitional regime, and
zigzags rapidly and randomly when the flow becomes fully turbulent. These
zigzags and the dispersion of the dye are indicative of the fluctuations in the
main flow and the rapid mixing of fluid particles from adjacent layers.
The intense mixing of the fluid in turbulent flow as a result of rapid fluctu-
ations enhances momentum transfer between fluid particles, which increases
the friction force on the surface and thus the required pumping power. The
friction factor reaches a maximum when the flow becomes fully turbulent.
V
avg
冮
A
c
ru(r) dA
c
rA
c
冮
R
0
ru(r)2pr dr
rpR
2
2
R
2
冮
R
0
u(r)r dr
m
#
rV
avg
A
c
冮
A
c
ru(r) dA
c
323
CHAPTER 8
Laminar
flow
Turbulen
t
flow
FIGURE 8–3
Laminar and turbulent flow regimes
of candle smoke.
(a) Laminar flow
Dye trace
Dye injection
(b) Turbulent flow
Dye trace
Dye injection
V
avg
V
avg
FIGURE 8–4
The behavior of colored fluid injected
into the flow in laminar and turbulent
flows in a pipe.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 323
Reynolds Number
The transition from laminar to turbulent flow depends on the geometry, sur-
face roughness, flow velocity, surface temperature, and type of fluid, among
other things. After exhaustive experiments in the 1880s, Osborne Reynolds
discovered that the flow regime depends mainly on the ratio of inertial
forces to viscous forces in the fluid. This ratio is called the
Reynolds num-
ber
and is expressed for internal flow in a circular pipe as (Fig. 8–5)
(8–3)
where V
avg
average flow velocity (m/s), D characteristic length of the
geometry (diameter in this case, in m), and n m/r kinematic viscosity
of the fluid (m
2
/s). Note that the Reynolds number is a dimensionless quan-
tity (Chap. 7). Also, kinematic viscosity has the unit m
2
/s, and can be
viewed as viscous diffusivity or diffusivity for momentum.
At large Reynolds numbers, the inertial forces, which are proportional to
the fluid density and the square of the fluid velocity, are large relative to the
viscous forces, and thus the viscous forces cannot prevent the random and
rapid fluctuations of the fluid. At small or moderate Reynolds numbers,
however, the viscous forces are large enough to suppress these fluctuations
and to keep the fluid “in line.” Thus the flow is turbulent in the first case
and laminar in the second.
The Reynolds number at which the flow becomes turbulent is called the
critical Reynolds number, Re
cr
. The value of the critical Reynolds number
is different for different geometries and flow conditions. For internal flow in
a circular pipe, the generally accepted value of the critical Reynolds number
is Re
cr
2300.
For flow through noncircular pipes, the Reynolds number is based on the
hydraulic diameter D
h
defined as (Fig. 8–6)
Hydraulic diameter: (8–4)
where A
c
is the cross-sectional area of the pipe and p is its wetted perimeter.
The hydraulic diameter is defined such that it reduces to ordinary diameter
D for circular pipes,
Circular pipes:
It certainly is desirable to have precise values of Reynolds numbers for
laminar, transitional, and turbulent flows, but this is not the case in practice.
It turns out that the transition from laminar to turbulent flow also depends
on the degree of disturbance of the flow by surface roughness, pipe vibra-
tions, and fluctuations in the flow. Under most practical conditions, the flow
in a circular pipe is laminar for Re 2300, turbulent for Re 4000, and
transitional in between. That is,
Re 4000 turbulent flow
2300 Re 4000 transitional flow
Re 2300 laminar flow
D
h
4A
c
p
4(pD
2
/4)
pD
D
D
h
4A
c
p
Re
Inertial forces
Viscous forces
V
avg
D
n
rV
avg
D
m
324
FLUID MECHANICS
avg
Inertial forces
––––––––––––
Viscous forces
Re =
avg
avg
avg
L
V
avg
FIGURE 8–5
The Reynolds number can be viewed
as the ratio of inertial forces to viscous
forces acting on a fluid element.
D
h
== D
4(pD
2
/4)
pD
D
h
== a
4a
2
4a
D
h
==
4ab
2(a + b)
2ab
a + b
Circular tube:
Rectangular duct:
Square duct:
a
b
D
a
a
FIGURE 8–6
The hydraulic diameter D
h
4A
c
/p is
defined such that it reduces to ordinary
diameter for circular tubes.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 324
In transitional flow, the flow switches between laminar and turbulent ran-
domly (Fig. 8–7). It should be kept in mind that laminar flow can be main-
tained at much higher Reynolds numbers in very smooth pipes by avoiding
flow disturbances and pipe vibrations. In such carefully controlled experi-
ments, laminar flow has been maintained at Reynolds numbers of up to
100,000.
8–3
■
THE ENTRANCE REGION
Consider a fluid entering a circular pipe at a uniform velocity. Because of
the no-slip condition, the fluid particles in the layer in contact with the sur-
face of the pipe come to a complete stop. This layer also causes the fluid
particles in the adjacent layers to slow down gradually as a result of friction.
To make up for this velocity reduction, the velocity of the fluid at the mid-
section of the pipe has to increase to keep the mass flow rate through the
pipe constant. As a result, a velocity gradient develops along the pipe.
The region of the flow in which the effects of the viscous shearing forces
caused by fluid viscosity are felt is called the velocity boundary layer or
just the
boundary layer. The hypothetical boundary surface divides the
flow in a pipe into two regions: the boundary layer region, in which the
viscous effects and the velocity changes are significant, and the
irrotational
(core) flow region,
in which the frictional effects are negligible and the
velocity remains essentially constant in the radial direction.
The thickness of this boundary layer increases in the flow direction until
the boundary layer reaches the pipe center and thus fills the entire pipe, as
shown in Fig. 8–8. The region from the pipe inlet to the point at which the
boundary layer merges at the centerline is called the hydrodynamic
entrance region, and the length of this region is called the
hydrodynamic
entry length
L
h
. Flow in the entrance region is called hydrodynamically
developing flow since this is the region where the velocity profile develops.
The region beyond the entrance region in which the velocity profile is fully
developed and remains unchanged is called the
hydrodynamically fully
developed region.
The flow is said to be fully developed when the normal-
ized temperature profile remains unchanged as well. Hydrodynamically
developed flow is equivalent to fully developed flow when the fluid in the
pipe is not heated or cooled since the fluid temperature in this case remains
325
CHAPTER 8
Laminar Turbulent
V
avg
Dye trace
Dye injection
FIGURE 8–7
In the transitional flow region
of 2300 Re 4000, the flow
switches between laminar and
turbulent randomly.
x
r
Hydrodynamic entrance region
Hydrodynamically fully developed region
Velocity boundary
layer
Developing velocity
profile
Fully developed
velocity profile
Irrotational (core)
flow region
V
avg
V
avg
V
avg
V
avg
V
avg
FIGURE 8–8
The development of the velocity
boundary layer in a pipe. (The
developed average velocity profile is
parabolic in laminar flow, as shown,
but somewhat flatter or fuller in
turbulent flow.)
cen72367_ch08.qxd 11/4/04 7:13 PM Page 325
essentially constant throughout. The velocity profile in the fully developed
region is parabolic in laminar flow and somewhat flatter (or fuller) in turbu-
lent flow due to eddy motion and more vigorous mixing in the radial direc-
tion. The time-averaged velocity profile remains unchanged when the flow
is fully developed, and thus
Hydrodynamically fully developed: (8–5)
The shear stress at the pipe wall t
w
is related to the slope of the velocity
profile at the surface. Noting that the velocity profile remains unchanged in
the hydrodynamically fully developed region, the wall shear stress also
remains constant in that region (Fig. 8–9).
Consider fluid flow in the hydrodynamic entrance region of a pipe. The
wall shear stress is the highest at the pipe inlet where the thickness of the
boundary layer is smallest, and decreases gradually to the fully developed
value, as shown in Fig. 8–10. Therefore, the pressure drop is higher in the
entrance regions of a pipe, and the effect of the entrance region is always to
increase the average friction factor for the entire pipe. This increase may be
significant for short pipes but is negligible for long ones.
Entry Lengths
The hydrodynamic entry length is usually taken to be the distance from the
pipe entrance to where the wall shear stress (and thus the friction factor)
reaches within about 2 percent of the fully developed value. In laminar flow,
the hydrodynamic entry length is given approximately as [see Kays and
Crawford (1993) and Shah and Bhatti (1987)]
(8–6)L
h, laminar
⬵ 0.05ReD
u(r, x)
x
0 → u u(r)
326
FLUID MECHANICS
t
w
t
w
t
w
t
w
FIGURE 8–9
In the fully developed flow region of
a pipe, the velocity profile does not
change downstream, and thus the wall
shear stress remains constant as well.
t
w
t
w
t
w
t
w
t
w
t
w
L
h
x
r
x
Fully
developed
region
Velocity boundary layer
Fully developed
region
Entrance region
Entrance region
t
w w
t
ww
V
avg
FIGURE 8–10
The variation of wall shear stress in
the flow direction for flow in a pipe
from the entrance region into the fully
developed region.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 326
For Re 20, the hydrodynamic entry length is about the size of the diame-
ter, but increases linearly with velocity. In the limiting laminar case of Re
2300, the hydrodynamic entry length is 115D.
In turbulent flow, the intense mixing during random fluctuations usually
overshadows the effects of molecular diffusion. The hydrodynamic entry
length for turbulent flow can be approximated as [see Bhatti and Shah
(1987) and Zhi-qing (1982)]
(8–7)
The entry length is much shorter in turbulent flow, as expected, and its depen-
dence on the Reynolds number is weaker. In many pipe flows of practical
engineering interest, the entrance effects become insignificant beyond a pipe
length of 10 diameters, and the hydrodynamic entry length is approximated as
(8–8)
Precise correlations for calculating the frictional head losses in entrance
regions are available in the literature. However, the pipes used in practice
are usually several times the length of the entrance region, and thus the flow
through the pipes is often assumed to be fully developed for the entire
length of the pipe. This simplistic approach gives reasonable results for
long pipes but sometimes poor results for short ones since it underpredicts
the wall shear stress and thus the friction factor.
8–4
■
LAMINAR FLOW IN PIPES
We mentioned in Section 8–2 that flow in pipes is laminar for Re 2300,
and that the flow is fully developed if the pipe is sufficiently long (relative
to the entry length) so that the entrance effects are negligible. In this section
we consider the steady laminar flow of an incompressible fluid with con-
stant properties in the fully developed region of a straight circular pipe. We
obtain the momentum equation by applying a momentum balance to a dif-
ferential volume element, and obtain the velocity profile by solving it. Then
we use it to obtain a relation for the friction factor. An important aspect of
the analysis here is that it is one of the few available for viscous flow.
In fully developed laminar flow, each fluid particle moves at a constant
axial velocity along a streamline and the velocity profile u(r) remains
unchanged in the flow direction. There is no motion in the radial direction,
and thus the velocity component in the direction normal to flow is every-
where zero. There is no acceleration since the flow is steady and fully
developed.
Now consider a ring-shaped differential volume element of radius r, thick-
ness dr, and length dx oriented coaxially with the pipe, as shown in Fig.
8–11. The volume element involves only pressure and viscous effects and
thus the pressure and shear forces must balance each other. The pressure
force acting on a submerged plane surface is the product of the pressure at
the centroid of the surface and the surface area. A force balance on the
volume element in the flow direction gives
(8–9)
(2pr dr P)
x
(2pr dr P)
xdx
(2pr dx t)
r
(2pr dx t)
rdr
0
L
h, turbulent
⬇ 10D
L
h, turbulent
1.359DRe
1/4
D
327
CHAPTER 8
u(r)
u
max
x
dx
dr
r
R
P
x
P
x
dx
t
r
t
rdr
FIGURE 8–11
Free-body diagram of a ring-shaped
differential fluid element of radius r,
thickness dr, and length dx oriented
coaxially with a horizontal pipe in
fully developed laminar flow.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 327
which indicates that in fully developed flow in a horizontal pipe, the
viscous and pressure forces balance each other. Dividing by 2pdrdx and
rearranging,
(8–10)
Taking the limit as dr, dx → 0 gives
(8–11)
Substituting t m(du/dr) and taking m constant gives the desired
equation,
(8–12)
The quantity du/dr is negative in pipe flow, and the negative sign is included
to obtain positive values for t. (Or, du/dr du/dy since y R r.) The
left side of Eq. 8–12 is a function of r, and the right side is a function of x.
The equality must hold for any value of r and x, and an equality of the form
f(r) g(x) can be satisfied only if both f (r) and g(x) are equal to the same
constant. Thus we conclude that dP/dx constant. This can be verified by
writing a force balance on a volume element of radius R and thickness dx
(a slice of the pipe), which gives (Fig. 8–12)
(8–13)
Here t
w
is constant since the viscosity and the velocity profile are constants
in the fully developed region. Therefore, dP/dx constant.
Equation 8–12 can be solved by rearranging and integrating it twice to give
(8–14)
The velocity profile u(r) is obtained by applying the boundary conditions
u/r 0 at r 0 (because of symmetry about the centerline) and u 0 at
r R (the no-slip condition at the pipe surface). We get
(8–15)
Therefore, the velocity profile in fully developed laminar flow in a pipe is
parabolic with a maximum at the centerline and minimum (zero) at the pipe
wall. Also, the axial velocity u is positive for any r, and thus the axial pres-
sure gradient dP/dx must be negative (i.e., pressure must decrease in the
flow direction because of viscous effects).
The average velocity is determined from its definition by substituting Eq.
8–15 into Eq. 8–2, and performing the integration. It gives
(8–16)
Combining the last two equations, the velocity profile is rewritten as
(8–17)u(r) 2V
avg
a1
r
2
R
2
b
V
avg
2
R
2
冮
R
0
u(r)r dr
2
R
2
冮
R
0
R
2
4m
a
dP
dx
b a1
r
2
R
2
br dr
R
2
8m
a
dP
dx
b
u(r)
R
2
4m
a
dP
dx
b a1
r
2
R
2
b
u(r)
1
4m
a
dP
dx
b C
1
ln r C
2
dP
dx
2t
w
R
m
r
d
dr
ar
du
dr
b
dP
dx
r
dP
dx
d(rt)
dr
0
r
P
xdx
P
x
dx
(rt)
rdr
(rt)
r
dr
0
328
FLUID MECHANICS
t
w
R
2
P
–
pR
2
(
P
dP) – 2pR dx
t
w
= 0
–=
dP
dx
R
r
x
2
pR dx
t
w
pR
2
(P dP)
p
2
pR
2
P
R
Force balance:
Simplifying:
dx
FIGURE 8–12
Free-body diagram of a fluid disk
element of radius R and length dx in
fully developed laminar flow in a
horizontal pipe.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 328
This is a convenient form for the velocity profile since V
avg
can be deter-
mined easily from the flow rate information.
The maximum velocity occurs at the centerline and is determined from
Eq. 8–17 by substituting r 0,
(8–18)
Therefore, the average velocity in fully developed laminar pipe flow is one-
half of the maximum velocity.
Pressure Drop and Head Loss
A quantity of interest in the analysis of pipe flow is the pressure drop P
since it is directly related to the power requirements of the fan or pump to
maintain flow. We note that dP/dx constant, and integrating from x x
1
where the pressure is P
1
to x x
1
L where the pressure is P
2
gives
(8–19)
Substituting Eq. 8–19 into the V
avg
expression in Eq. 8–16, the pressure
drop can be expressed as
Laminar flow: (8–20)
The symbol is typically used to indicate the difference between the final
and initial values, like y y
2
y
1
. But in fluid flow, P is used to desig-
nate pressure drop, and thus it is P
1
P
2
. A pressure drop due to viscous
effects represents an irreversible pressure loss, and it is called pressure loss
P
L
to emphasize that it is a loss (just like the head loss h
L
, which is pro-
portional to it).
Note from Eq. 8–20 that the pressure drop is proportional to the viscosity
m of the fluid, and P would be zero if there were no friction. Therefore,
the drop of pressure from P
1
to P
2
in this case is due entirely to viscous
effects, and Eq. 8–20 represents the pressure loss P
L
when a fluid of vis-
cosity m flows through a pipe of constant diameter D and length L at aver-
age velocity V
avg
.
In practice, it is found convenient to express the pressure loss for all types
of fully developed internal flows (laminar or turbulent flows, circular or
noncircular pipes, smooth or rough surfaces, horizontal or inclined pipes) as
(Fig. 8–13)
Pressure loss: (8–21)
where rV
2
avg
/2 is the dynamic pressure and f is the Darcy friction factor,
(8–22)
It is also called the Darcy–Weisbach friction factor, named after the
Frenchman Henry Darcy (1803–1858) and the German Julius Weisbach
(1806–1871), the two engineers who provided the greatest contribution in
its development. It should not be confused with the friction coefficient C
f
f
8t
w
rV
2
avg
P
L
f
L
D
rV
2
avg
2
P P
1
P
2
8mLV
avg
R
2
32mLV
avg
D
2
dP
dx
P
2
P
1
L
u
max
2V
avg
329
CHAPTER 8
Pressure loss: ∆P
L
= f
L
V
avg
D 2
21
2g
Head loss: h
L
== f
L
∆P
L
D
rg
D
L
∆P
L
V
avg
r
V
avg
2
2
FIGURE 8–13
The relation for pressure loss (and
head loss) is one of the most general
relations in fluid mechanics, and it is
valid for laminar or turbulent flows,
circular or noncircular pipes, and
pipes with smooth or rough surfaces.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 329
[also called the Fanning friction factor, named after the American engineer
John Fanning (1837–1911)], which is defined as C
f
2t
w
/(rV
2
avg
) f/4.
Setting Eqs. 8–20 and 8–21 equal to each other and solving for f gives the
friction factor for fully developed laminar flow in a circular pipe,
Circular pipe, laminar: (8–23)
This equation shows that in laminar flow, the friction factor is a function of
the Reynolds number only and is independent of the roughness of the pipe
surface.
In the analysis of piping systems, pressure losses are commonly expressed
in terms of the equivalent fluid column height, called the
head loss h
L
. Not-
ing from fluid statics that P rgh and thus a pressure difference of P
corresponds to a fluid height of h P/rg, the pipe head loss is obtained
by dividing P
L
by rg to give
Head loss: (8–24)
The head loss h
L
represents the additional height that the fluid needs to be
raised by a pump in order to overcome the frictional losses in the pipe. The
head loss is caused by viscosity, and it is directly related to the wall shear
stress. Equations 8–21 and 8–24 are valid for both laminar and turbulent
flows in both circular and noncircular pipes, but Eq. 8–23 is valid only for
fully developed laminar flow in circular pipes.
Once the pressure loss (or head loss) is known, the required pumping
power to overcome the pressure loss is determined from
(8–25)
where
V
.
is the volume flow rate and m
.
is the mass flow rate.
The average velocity for laminar flow in a horizontal pipe is, from Eq. 8–20,
Horizontal pipe: (8–26)
Then the volume flow rate for laminar flow through a horizontal pipe of
diameter D and length L becomes
(8–27)
This equation is known as Poiseuille’s law, and this flow is called Hagen–
Poiseuille flow in honor of the works of G. Hagen (1797–1884) and J.
Poiseuille (1799–1869) on the subject. Note from Eq. 8–27 that for a speci-
fied flow rate, the pressure drop and thus the required pumping power is pro-
portional to the length of the pipe and the viscosity of the fluid, but it is
inversely proportional to the fourth power of the radius (or diameter) of the
pipe. Therefore, the pumping power requirement for a piping system can be
reduced by a factor of 16 by doubling the pipe diameter (Fig. 8–14). Of
course the benefits of the reduction in the energy costs must be weighed
against the increased cost of construction due to using a larger-diameter pipe.
The pressure drop P equals the pressure loss P
L
in the case of a hor-
izontal pipe, but this is not the case for inclined pipes or pipes with vari-
able cross-sectional area. This can be demonstrated by writing the energy
V
#
V
avg
A
c
(P
1
P
2
)R
2
8mL
pR
2
(P
1
P
2
)pD
4
128mL
P pD
4
128mL
V
avg
(P
1
P
2
)R
2
8mL
(P
1
P
2
)D
2
32mL
P D
2
32mL
W
#
pump, L
V
#
P
L
V
#
rgh
L
m
#
gh
L
h
L
P
L
rg
f
L
D
V
2
avg
2g
f
64m
rDV
avg
64
Re
330
FLUID MECHANICS
2D
W
pump
= 16 hp
⋅
W
pump
= 1 hp
/4
⋅
D
V
avg
V
avg
FIGURE 8–14
The pumping power requirement for
a laminar flow piping system can be
reduced by a factor of 16 by doubling
the pipe diameter.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 330
equation for steady, incompressible one-dimensional flow in terms of heads
as (see Chap. 5)
(8–28)
where h
pump, u
is the useful pump head delivered to the fluid, h
turbine, e
is the
turbine head extracted from the fluid, h
L
is the irreversible head loss
between sections 1 and 2, V
1
and V
2
are the average velocities at sections
1 and 2, respectively, and a
1
and a
2
are the kinetic energy correction factors
at sections 1 and 2 (it can be shown that a 2 for fully developed laminar
flow and about 1.05 for fully developed turbulent flow). Equation 8–28 can
be rearranged as
(8–29)
Therefore, the pressure drop P P
1
P
2
and pressure loss P
L
rgh
L
for a given flow section are equivalent if (1) the flow section is horizontal
so that there are no hydrostatic or gravity effects (z
1
z
2
), (2) the flow sec-
tion does not involve any work devices such as a pump or a turbine since
they change the fluid pressure (h
pump, u
h
turbine, e
0), (3) the cross-sectional
area of the flow section is constant and thus the average flow velocity is
constant (V
1
V
2
), and (4) the velocity profiles at sections 1 and 2 are the
same shape (a
1
a
2
).
Inclined Pipes
Relations for inclined pipes can be obtained in a similar manner from a force
balance in the direction of flow. The only additional force in this case is the
component of the fluid weight in the flow direction, whose magnitude is
(8–30)
where u is the angle between the horizontal and the flow direction (Fig.
8–15). The force balance in Eq. 8–9 now becomes
(8–31)
which results in the differential equation
(8–32)
Following the same solution procedure, the velocity profile can be shown to be
(8–33)
It can also be shown that the average velocity and the volume flow rate rela-
tions for laminar flow through inclined pipes are, respectively,
(8–34)
which are identical to the corresponding relations for horizontal pipes, except
that P is replaced by P rgL sin u. Therefore, the results already
obtained for horizontal pipes can also be used for inclined pipes provided
that P is replaced by P rgL sin u (Fig. 8–16). Note that u 0 and thus
sin u 0 for uphill flow, and u 0 and thus sin u 0 for downhill flow.
V
avg
(P rgL sin u)D
2
32mL
and
V
#
(P rgL sin u)pD
4
128mL
u(r)
R
2
4m
a
dP
dx
rg sin uba1
r
2
R
2
b
m
r
d
dr
ar
du
dr
b
dP
dx
rg sin u
(2pr dx t)
rdr
rg(2pr dr dx) sin u 0
(2pr dr P)
x
(2pr dr P)
xdx
(2pr dx t)
r
W
x
W sin u rg
V
element
sin u rg(2pr dr dx) sin u
P
1
P
2
r(a
2
V
2
2
a
1
V
2
1
)/2 rg[(z
2
z
1
) h
turbine, e
h
pump, u
h
L
]
P
1
rg
a
1
V
2
1
2g
z
1
h
pump, u
P
2
rg
a
2
V
2
2
2g
z
2
h
turbine, e
h
L
331
CHAPTER 8
u
r
dr
t
r
t
P
x
dx
W sin
W
P
x
x
r
u
u
dx
dr
FIGURE 8–15
Free-body diagram of a ring-shaped
differential fluid element of radius r,
thickness dr, and length dx oriented
coaxially with an inclined pipe in fully
developed laminar flow.
Uphill flow: u > 0 and sin u > 0
Downhill flow: u < 0 and sin u < 0
Horizontal pipe:
V
=
⋅
∆P
p
D
4
128
L
m
⋅
Inclined pipe:
V
=
(∆P –
r
gL sin u)
p
D
4
128
m
L
FIGURE 8–16
The relations developed for fully
developed laminar flow through
horizontal pipes can also be used
for inclined pipes by replacing
P with P rgL sin u.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 331
In inclined pipes, the combined effect of pressure difference and gravity
drives the flow. Gravity helps downhill flow but opposes uphill flow. There-
fore, much greater pressure differences need to be applied to maintain a
specified flow rate in uphill flow although this becomes important only for
liquids, because the density of gases is generally low. In the special case of
no flow (
V
.
0), we have P rgL sin u, which is what we would obtain
from fluid statics (Chap. 3).
Laminar Flow in Noncircular Pipes
The friction factor f relations are given in Table 8–1 for fully developed lam-
inar flow in pipes of various cross sections. The Reynolds number for flow
in these pipes is based on the hydraulic diameter D
h
4A
c
/p, where A
c
is
the cross-sectional area of the pipe and p is its wetted perimeter.
332
FLUID MECHANICS
TABLE 8–1
Friction factor for fully developed laminar flow in pipes of various cross
sections (D
h
4A
c
/p and Re V
avg
D
h
/n)
a/b Friction Factor
Tube Geometry or u° f
Circle — 64.00/Re
Rectangle a/b
1 56.92/Re
2 62.20/Re
3 68.36/Re
4 72.92/Re
6 78.80/Re
8 82.32/Re
96.00/Re
Ellipse a/b
1 64.00/Re
2 67.28/Re
4 72.96/Re
8 76.60/Re
16 78.16/Re
Isosceles triangle u
10° 50.80/Re
30° 52.28/Re
60° 53.32/Re
90° 52.60/Re
120° 50.96/Re
D
b
a
b
a
u
cen72367_ch08.qxd 11/4/04 7:13 PM Page 332
EXAMPLE 8–1 Flow Rates in Horizontal and Inclined Pipes
Oil at 20°C (r 888 kg/m
3
and m 0.800 kg/m · s) is flowing steadily
through a 5-cm-diameter 40-m-long pipe (Fig. 8–17). The pressure at the
pipe inlet and outlet are measured to be 745 and 97 kPa, respectively.
Determine the flow rate of oil through the pipe assuming the pipe is (
a) hor-
izontal, (
b) inclined 15° upward, (c) inclined 15° downward. Also verify that
the flow through the pipe is laminar.
SOLUTION The pressure readings at the inlet and outlet of a pipe are given.
The flow rates are to be determined for three different orientations, and the
flow is to be shown to be laminar.
Assumptions 1 The flow is steady and incompressible. 2 The entrance
effects are negligible, and thus the flow is fully developed. 3 The pipe
involves no components such as bends, valves, and connectors. 4 The piping
section involves no work devices such as a pump or a turbine.
Properties The density and dynamic viscosity of oil are given to be r
888 kg/m
3
and m 0.800 kg/m · s, respectively.
Analysis The pressure drop across the pipe and the pipe cross-sectional
area are
(
a) The flow rate for all three cases can be determined from Eq. 8–34,
where u is the angle the pipe makes with the horizontal. For the horizontal
case, u 0 and thus sin u 0. Therefore,
0.00311 m
3
/s
V
#
horiz
P pD
4
128mL
(648 kPa)p (0.05 m)
4
128(0.800 kg/m s)(40 m)
a
1000 N/m
2
1 kPa
ba
1 kg m/s
2
1 N
b
V
#
(P rgL sin u)pD
4
128mL
A
c
pD
2
/4 p(0.05 m)
2
/4 0.001963 m
2
P P
1
P
2
745 97 648 kPa
333
CHAPTER 8
(b) For uphill flow with an inclination of 15°, we have u 15°, and
(
c) For downhill flow with an inclination of 15°, we have u 15°, and
0.00354 m
3
/s
[648,000 Pa (888 kg/m
3
)(9.81 m/s
2
)(40 m) sin (15)]p(0.05 m)
4
128(0.800 kg/m s)(40 m)
a
1 kg m/s
2
1 Pa m
2
b
V
#
downhill
(P rgL sin u)pD
4
128mL
0.00267 m
3
/s
[648,000 Pa (888 kg/m
3
)(9.81 m/s
2
)(40 m) sin 15°]p(0.05 m)
4
128(0.800 kg/m s)(40 m)
a
1 kg m/s
2
1 Pa m
2
b
V
#
uphill
(P rgL sin u)pD
4
128mL
+15˚
–15˚
Horizontal
FIGURE 8–17
Schematic for Example 8–1.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 333
The flow rate is the highest for the downhill flow case, as expected. The
average fluid velocity and the Reynolds number in this case are
which is much less than 2300. Therefore, the flow is
laminar for all three
cases and the analysis is valid.
Discussion Note that the flow is driven by the combined effect of pressure
difference and gravity. As can be seen from the flow rates we calculated,
gravity opposes uphill flow, but enhances downhill flow. Gravity has no effect
on the flow rate in the horizontal case. Downhill flow can occur even in the
absence of an applied pressure difference. For the case of
P
1
P
2
97 kPa
(i.e., no applied pressure difference), the pressure throughout the entire pipe
would remain constant at 97 Pa, and the fluid would flow through the pipe at
a rate of 0.00043 m
3
/s under the influence of gravity. The flow rate increases
as the tilt angle of the pipe from the horizontal is increased in the negative
direction and would reach its maximum value when the pipe is vertical.
EXAMPLE 8–2 Pressure Drop and Head Loss in a Pipe
Water at 40°F (r 62.42 lbm/ft
3
and m 1.038 10
3
lbm/ft · s) is
flowing through a 0.12-in- ( 0.010 ft) diameter 30-ft-long horizontal pipe
steadily at an average velocity of 3.0 ft/s (Fig. 8–18). Determine (
a) the head
loss, (
b) the pressure drop, and (c) the pumping power requirement to over-
come this pressure drop.
SOLUTION The average flow velocity in a pipe is given. The head loss, the
pressure drop, and the pumping power are to be determined.
Assumptions 1 The flow is steady and incompressible. 2 The entrance
effects are negligible, and thus the flow is fully developed. 3 The pipe
involves no components such as bends, valves, and connectors.
Properties The density and dynamic viscosity of water are given to be r
62.42 lbm/ft
3
and m 1.038 10
3
lbm/ft · s, respectively.
Analysis (a) First we need to determine the flow regime. The Reynolds num-
ber is
which is less than 2300. Therefore, the flow is laminar. Then the friction
factor and the head loss become
(
b) Noting that the pipe is horizontal and its diameter is constant, the pres-
sure drop in the pipe is due entirely to the frictional losses and is equivalent
to the pressure loss,
h
L
f
L
D
V
2
avg
2g
0.0355
30 ft
0.01 ft
(3 ft/s)
2
2(32.2 ft/s
2
)
14.9 ft
f
64
Re
64
1803
0.0355
Re
rV
avg
D
m
(62.42 lbm/ft
3
)(3 ft/s)(0.01 ft)
1.038 10
3
lbm/ft s
1803
Re
rV
avg
D
m
(888 kg/m
3
)(1.80 m/s)(0.05 m)
0.800 kg/m s
100
V
avg
V
#
A
c
0.00354 m
3
/s
0.001963 m
2
1.80 m/s
334
FLUID MECHANICS
3.0 ft/s
30 ft
0.12 in
FIGURE 8–18
Schematic for Example 8–2.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 334
(c) The volume flow rate and the pumping power requirements are
Therefore, power input in the amount of 0.30 W is needed to overcome the
frictional losses in the flow due to viscosity.
Discussion The pressure rise provided by a pump is often listed by a pump
manufacturer in units of head (Chap. 14). Thus, the pump in this flow needs to
provide 14.9 ft of water head in order to overcome the irreversible head loss.
8–5
■
TURBULENT FLOW IN PIPES
Most flows encountered in engineering practice are turbulent, and thus it is
important to understand how turbulence affects wall shear stress. However,
turbulent flow is a complex mechanism dominated by fluctuations, and
despite tremendous amounts of work done in this area by researchers, the
theory of turbulent flow remains largely undeveloped. Therefore, we must
rely on experiments and the empirical or semi-empirical correlations devel-
oped for various situations.
Turbulent flow is characterized by random and rapid fluctuations of
swirling regions of fluid, called eddies, throughout the flow. These fluctua-
tions provide an additional mechanism for momentum and energy transfer.
In laminar flow, fluid particles flow in an orderly manner along pathlines,
and momentum and energy are transferred across streamlines by molecular
diffusion. In turbulent flow, the swirling eddies transport mass, momentum,
and energy to other regions of flow much more rapidly than molecular dif-
fusion, greatly enhancing mass, momentum, and heat transfer. As a result,
turbulent flow is associated with much higher values of friction, heat trans-
fer, and mass transfer coefficients (Fig. 8–19).
Even when the average flow is steady, the eddy motion in turbulent flow
causes significant fluctuations in the values of velocity, temperature, pres-
sure, and even density (in compressible flow). Figure 8–20 shows the varia-
tion of the instantaneous velocity component u with time at a specified loca-
tion, as can be measured with a hot-wire anemometer probe or other
sensitive device. We observe that the instantaneous values of the velocity
fluctuate about an average value, which suggests that the velocity can be
expressed as the sum of an average value u
–
and a fluctuating component u,
(8–35)
This is also the case for other properties such as the velocity component v
in the y-direction, and thus v v
–
v, P P
–
P, and T T
–
T. The
average value of a property at some location is determined by averaging it
over a time interval that is sufficiently large so that the time average levels
off to a constant. Therefore, the time average of fluctuating components is
u u
u
W
#
pump
V
#
P (0.000236 ft
3
/s)(929 lbf/ft
2
) a
1 W
0.737 lbf ft/s
b
0.30 W
V
#
V
avg
A
c
V
avg
(pD
2
/4) (3 ft/s)[p(0.01 ft)
2
/4] 0.000236 ft
3
/s
929 lbf/ft
2
6.45 psi
P P
L
f
L
D
rV
2
avg
2
0.0355
30 ft
0.01 ft
(62.42 lbm/ft
3
)(3 ft/s)
2
2
a
1 lbf
32.2 lbm ft/s
2
b
335
CHAPTER 8
(a) Before
turbulence
22222
55
7
12
7
12
7
12
7
12
7
12
555
(b) After
turbulence
12 2 5 7 5
122
7
2
7
5
12
2
12
7
5
12
572
FIGURE 8–19
The intense mixing in turbulent flow
brings fluid particles at different
momentums into close contact and
thus enhances momentum transfer.
u
u
u
–
Time, t
FIGURE 8–20
Fluctuations of the velocity
component u with time at a specified
location in turbulent flow.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 335
zero, e.g., . The magnitude of u is usually just a few percent of u
–
,but
the high frequencies of eddies (in the order of a thousand per second) makes
them very effective for the transport of momentum, thermal energy, and mass.
In time-averaged stationary turbulent flow, the average values of properties
(indicated by an overbar) are independent of time. The chaotic fluctuations of
fluid particles play a dominant role in pressure drop, and these random
motions must be considered in analyses together with the average velocity.
Perhaps the first thought that comes to mind is to determine the shear
stress in an analogous manner to laminar flow from t m du
–
/dr, where
u
–
(r) is the average velocity profile for turbulent flow. But the experimental
studies show that this is not the case, and the shear stress is much larger due
to the turbulent fluctuations. Therefore, it is convenient to think of the tur-
bulent shear stress as consisting of two parts: the laminar component, which
accounts for the friction between layers in the flow direction (expressed as
t
lam
m du
–
/dr), and the turbulent component, which accounts for the
friction between the fluctuating fluid particles and the fluid body (denoted
as t
turb
and is related to the fluctuation components of velocity). Then the
total shear stress in turbulent flow can be expressed as
(8–36)
The typical average velocity profile and relative magnitudes of laminar and
turbulent components of shear stress for turbulent flow in a pipe are given in
Fig. 8–21. Note that although the velocity profile is approximately parabolic
in laminar flow, it becomes flatter or “fuller” in turbulent flow, with a sharp
drop near the pipe wall. The fullness increases with the Reynolds number,
and the velocity profile becomes more nearly uniform, lending support to the
commonly utilized uniform velocity profile approximation for fully devel-
oped turbulent pipe flow. Keep in mind, however, that the flow speed at the
wall of a stationary pipe is always zero (no-slip condition).
Turbulent Shear Stress
Consider turbulent flow in a horizontal pipe, and the upward eddy motion of
fluid particles in a layer of lower velocity to an adjacent layer of higher
velocity through a differential area dA as a result of the velocity fluctuation
v, as shown in Fig. 8–22. The mass flow rate of the fluid particles rising
through dA is rvdA, and its net effect on the layer above dA is a reduction in
its average flow velocity because of momentum transfer to the fluid particles
with lower average flow velocity. This momentum transfer causes the hori-
zontal velocity of the fluid particles to increase by u, and thus its momen-
tum in the horizontal direction to increase at a rate of (rvdA)u, which must
be equal to the decrease in the momentum of the upper fluid layer. Noting
that force in a given direction is equal to the rate of change of momentum
in that direction, the horizontal force acting on a fluid element above dA
due to the passing of fluid particles through dA is dF (rvdA)(u)
ruvdA. Therefore, the shear force per unit area due to the eddy motion
of fluid particles dF/dA ruv can be viewed as the instantaneous turbu-
lent shear stress. Then the
turbulent shear stress can be expressed as
(8–37)
where is the time average of the product of the fluctuating velocity
components u and v. Note that even though and v
0u 0uv 0
uv
t
turb
ruv
t
total
t
lam
t
turb
u 0
336
FLUID MECHANICS
t
turb
t
lam
u(r)
r
0
r
0
0
t
total
t
FIGURE 8–21
The velocity profile and the variation
of shear stress with radial distance for
turbulent flow in a pipe.
v
r
v
dA
u(y)
u
u
dA
y
FIGURE 8–22
Fluid particle moving upward through
a differential area dA as a result of the
velocity fluctuation v.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 336
(and thus ), and experimental results show that is usually a
negative quantity. Terms such as or are called
Reynolds
stresses
or turbulent stresses.
Many semi-empirical formulations have been developed that model the
Reynolds stress in terms of average velocity gradients in order to provide
mathematical closure to the equations of motion. Such models are called
turbulence models and are discussed in more detail in Chap. 15.
The random eddy motion of groups of particles resembles the random
motion of molecules in a gas—colliding with each other after traveling a
certain distance and exchanging momentum in the process. Therefore,
momentum transport by eddies in turbulent flows is analogous to the molec-
ular momentum diffusion. In many of the simpler turbulence models, turbu-
lent shear stress is expressed in an analogous manner as suggested by the
French mathematician Joseph Boussinesq (1842–1929) in 1877 as
(8–38)
where m
t
is the
eddy viscosity or turbulent viscosity, which accounts for
momentum transport by turbulent eddies. Then the total shear stress can be
expressed conveniently as
(8–39)
where n
t
m
t
/r is the kinematic eddy viscosity or kinematic turbulent
viscosity (also called the eddy diffusivity of momentum). The concept of eddy
viscosity is very appealing, but it is of no practical use unless its value can be
determined. In other words, eddy viscosity must be modeled as a function of
the average flow variables; we call this eddy viscosity closure. For example,
in the early 1900s, the German engineer L. Prandtl introduced the concept of
mixing length l
m
, which is related to the average size of the eddies that are
primarily responsible for mixing, and expressed the turbulent shear stress as
(8–40)
But this concept is also of limited use since l
m
is not a constant for a given
flow (in the vicinity of the wall, for example, l
m
is nearly proportional to the
distance from the wall) and its determination is not easy. Final mathematical
closure is obtained only when l
m
is written as a function of average flow
variables, distance from the wall, etc.
Eddy motion and thus eddy diffusivities are much larger than their molec-
ular counterparts in the core region of a turbulent boundary layer. The eddy
motion loses its intensity close to the wall and diminishes at the wall
because of the no-slip condition (u and v are identically zero at a station-
ary wall). Therefore, the velocity profile is very slowly changing in the core
region of a turbulent boundary layer, but very steep in the thin layer adja-
cent to the wall, resulting in large velocity gradients at the wall surface. So
it is no surprise that the wall shear stress is much larger in turbulent flow
than it is in laminar flow (Fig. 8–23).
Note that molecular diffusivity of momentum n (as well as m) is a fluid
property, and its value is listed in fluid handbooks. Eddy diffusivity n
t
(as
well as m
t
), however, is not a fluid property, and its value depends on flow
t
turb
m
t
u
y
rl
2
m
a
u
y
b
2
t
total
(m m
t
)
u
y
r(n n
t
)
u
y
t
turb
ruv
m
t
u
y
ru
2
ruv
uvu v 0
337
CHAPTER 8
y=0
Turbulent flow
y
u
y
y=0
Laminar flow
y
u
y
ab
ab
FIGURE 8–23
The velocity gradients at the wall, and
thus the wall shear stress, are much
larger for turbulent flow than they are
for laminar flow, even though the
turbulent boundary layer is thicker
than the laminar one for the same
value of free-stream velocity.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 337
conditions. Eddy diffusivity n
t
decreases toward the wall, becoming zero at
the wall. Its value ranges from zero at the wall to several thousand times the
value of the molecular diffusivity in the core region.
Turbulent Velocity Profile
Unlike laminar flow, the expressions for the velocity profile in a turbulent
flow are based on both analysis and measurements, and thus they are
semi-empirical in nature with constants determined from experimental data.
Consider fully developed turbulent flow in a pipe, and let u denote the time-
averaged velocity in the axial direction (and thus drop the overbar from u
–
for simplicity).
Typical velocity profiles for fully developed laminar and turbulent flows
are given in Fig. 8–24. Note that the velocity profile is parabolic in laminar
flow but is much fuller in turbulent flow, with a sharp drop near the pipe
wall. Turbulent flow along a wall can be considered to consist of four
regions, characterized by the distance from the wall. The very thin layer
next to the wall where viscous effects are dominant is the viscous (or lami-
nar or linear or wall) sublayer. The velocity profile in this layer is very
nearly linear, and the flow is streamlined. Next to the viscous sublayer is
the
buffer layer, in which turbulent effects are becoming significant, but the
flow is still dominated by viscous effects. Above the buffer layer is the
overlap (or transition) layer, also called the inertial sublayer, in which the
turbulent effects are much more significant, but still not dominant. Above
that is the outer (or turbulent) layer in the remaining part of the flow in
which turbulent effects dominate over molecular diffusion (viscous) effects.
Flow characteristics are quite different in different regions, and thus it is
difficult to come up with an analytic relation for the velocity profile for the
entire flow as we did for laminar flow. The best approach in the turbulent
case turns out to be to identify the key variables and functional forms using
dimensional analysis, and then to use experimental data to determine the
numerical values of any constants.
The thickness of the viscous sublayer is very small (typically, much less
than 1 percent of the pipe diameter), but this thin layer next to the wall plays
a dominant role on flow characteristics because of the large velocity gradi-
ents it involves. The wall dampens any eddy motion, and thus the flow in this
layer is essentially laminar and the shear stress consists of laminar shear
stress which is proportional to the fluid viscosity. Considering that velocity
changes from zero to nearly the core region value across a layer that is some-
times no thicker than a hair (almost like a step function), we would expect
the velocity profile in this layer to be very nearly linear, and experiments
confirm that. Then the velocity gradient in the viscous sublayer remains
nearly constant at du/dy u/y, and the wall shear stress can be expressed as
(8–41)
where y is the distance from the wall (note that y R r for a circular pipe).
The quantity t
w
/r is frequently encountered in the analysis of turbulent
velocity profiles. The square root of t
w
/r has the dimensions of velocity, and
thus it is convenient to view it as a fictitious velocity called the friction veloc-
ity expressed as . Substituting this into Eq. 8–41, the velocity
profile in the viscous sublayer can be expressed in dimensionless form as
u
*
1t
w
/r
t
w
m
u
y
rn
u
y
or
t
w
r
nu
y
338
FLUID MECHANICS
Laminar flow
u(r)
r
0
Turbulent flow
Turbulent layer
Overlap layer
Buffer layer
Viscous sublaye
r
u(r)
r
0
V
avg
V
avg
FIGURE 8–24
The velocity profile in fully developed
pipe flow is parabolic in laminar flow,
but much fuller in turbulent flow.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 338
Viscous sublayer: (8–42)
This equation is known as the law of the wall, and it is found to satisfacto-
rily correlate with experimental data for smooth surfaces for 0 yu
*
/n 5.
Therefore, the thickness of the viscous sublayer is roughly
Thickness of viscous sublayer: (8–43)
where u
d
is the flow velocity at the edge of the viscous sublayer, which is
closely related to the average velocity in a pipe. Thus we conclude that the
thickness of the viscous sublayer is proportional to the kinematic viscosity
and inversely proportional to the average flow velocity. In other words, the
viscous sublayer is suppressed and it gets thinner as the velocity (and thus
the Reynolds number) increases. Consequently, the velocity profile becomes
nearly flat and thus the velocity distribution becomes more uniform at very
high Reynolds numbers.
The quantity n/u
*
has dimensions of length and is called the viscous
length; it is used to nondimensionalize the distance y from the surface. In
boundary layer analysis, it is convenient to work with nondimensionalized
distance and nondimensionalized velocity defined as
Nondimensionalized variables: (8–44)
Then the law of the wall (Eq. 8–42) becomes simply
Normalized law of the wall: (8–45)
Note that the friction velocity u
*
is used to nondimensionalize both y and u,
and y
resembles the Reynolds number expression.
In the overlap layer, the experimental data for velocity are observed to
line up on a straight line when plotted against the logarithm of distance
from the wall. Dimensional analysis indicates and the experiments confirm
that the velocity in the overlap layer is proportional to the logarithm of dis-
tance, and the velocity profile can be expressed as
The logarithmic law: (8–46)
where k and B are constants whose values are determined experimentally to
be about 0.40 and 5.0, respectively. Equation 8–46 is known as the loga-
rithmic law. Substituting the values of the constants, the velocity profile is
determined to be
Overlap layer: (8–47)
It turns out that the logarithmic law in Eq. 8–47 satisfactorily represents exper-
imental data for the entire flow region except for the regions very close to the
wall and near the pipe center, as shown in Fig. 8–25, and thus it is viewed as a
universal velocity profile for turbulent flow in pipes or over surfaces. Note
from the figure that the logarithmic-law velocity profile is quite accurate for y
30, but neither velocity profile is accurate in the buffer layer, i.e., the region
5 y
30. Also, the viscous sublayer appears much larger in the figure than
it is since we used a logarithmic scale for distance from the wall.
u
u
*
2.5 ln
yu
*
n
5.0 or u
2.5 ln y
5.0
u
u
*
1
k
ln
yu
*
n
B
u
y
y
yu
*
n
and u
u
u
*
y d
sublayer
5n
u
*
25n
u
d
u
u
*
yu
*
n
339
CHAPTER 8
n
Viscous
sublayer
10
0
30
25
20
15
10
5
0
10
1
10
2
y
+
= yu*/
u
+
= u/u*
10
3
10
4
Buffer
layer
Overlap
layer
Turbulent
layer
Eq. 8–47
Eq. 8–42
Experimental data
FIGURE 8–25
Comparison of the law of the wall and
the logarithmic-law velocity profiles
with experimental data for fully
developed turbulent flow in a pipe.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 339
A good approximation for the outer turbulent layer of pipe flow can be
obtained by evaluating the constant B in Eq. 8–46 from the requirement that
maximum velocity in a pipe occurs at the centerline where r 0. Solving
for B from Eq. 8–46 by setting y R r R and u u
max
, and substitut-
ing it back into Eq. 8–46 together with k 0.4 gives
Outer turbulent layer: (8–48)
The deviation of velocity from the centerline value u
max
u is called the
velocity defect, and Eq. 8–48 is called the velocity defect law. This relation
shows that the normalized velocity profile in the core region of turbulent
flow in a pipe depends on the distance from the centerline and is independent
of the viscosity of the fluid. This is not surprising since the eddy motion is
dominant in this region, and the effect of fluid viscosity is negligible.
Numerous other empirical velocity profiles exist for turbulent pipe flow.
Among those, the simplest and the best known is the power-law velocity
profile expressed as
Power-law velocity profile: (8–49)
where the exponent n is a constant whose value depends on the Reynolds
number. The value of n increases with increasing Reynolds number. The
value n 7 generally approximates many flows in practice, giving rise to
the term one-seventh power-law velocity profile.
Various power-law velocity profiles are shown in Fig. 8–26 for n 6, 8,
and 10 together with the velocity profile for fully developed laminar flow
for comparison. Note that the turbulent velocity profile is fuller than the
laminar one, and it becomes more flat as n (and thus the Reynolds number)
increases. Also note that the power-law profile cannot be used to calculate
wall shear stress since it gives a velocity gradient of infinity there, and it
fails to give zero slope at the centerline. But these regions of discrepancy
constitute a small portion of flow, and the power-law profile gives highly
accurate results for turbulent flow through a pipe.
Despite the small thickness of the viscous sublayer (usually much less
than 1 percent of the pipe diameter), the characteristics of the flow in this
layer are very important since they set the stage for flow in the rest of the
pipe. Any irregularity or roughness on the surface disturbs this layer and
affects the flow. Therefore, unlike laminar flow, the friction factor in turbu-
lent flow is a strong function of surface roughness.
It should be kept in mind that roughness is a relative concept, and it has
significance when its height e is comparable to the thickness of the laminar
sublayer (which is a function of the Reynolds number). All materials appear
“rough” under a microscope with sufficient magnification. In fluid mechan-
ics, a surface is characterized as being rough when the hills of roughness
protrude out of the laminar sublayer. A surface is said to be smooth when
the sublayer submerges the roughness elements. Glass and plastic surfaces
are generally considered to be hydrodynamically smooth.
The Moody Chart
The friction factor in fully developed turbulent pipe flow depends on the
Reynolds number and the relative roughness e/D, which is the ratio of the
u
u
max
a
y
R
b
1/n
or
u
u
max
a1
r
R
b
1/n
u
max
u
u
*
2.5 ln
R
R r
340
FLUID MECHANICS
0.20 0.4 0.6 0.8
1
0.8
0.6
0.4
0.2
0
u/u
max
r/R
1
Laminar
n = 6
n = 8
n = 10
FIGURE 8–26
Power-law velocity profiles for
fully developed turbulent flow in
a pipe for different exponents, and
its comparison with the laminar
velocity profile.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 340
mean height of roughness of the pipe to the pipe diameter. The functional
form of this dependence cannot be obtained from a theoretical analysis, and all
available results are obtained from painstaking experiments using artificially
roughened surfaces (usually by gluing sand grains of a known size on the inner
surfaces of the pipes). Most such experiments were conducted by Prandtl’s stu-
dent J. Nikuradse in 1933, followed by the works of others. The friction factor
was calculated from the measurements of the flow rate and the pressure drop.
The experimental results obtained are presented in tabular, graphical, and
functional forms obtained by curve-fitting experimental data. In 1939, Cyril
F. Colebrook (1910–1997) combined the available data for transition and
turbulent flow in smooth as well as rough pipes into the following implicit
relation known as the Colebrook equation:
(8–50)
We note that the logarithm in Eq. 8–50 is a base 10 rather than a natural
logarithm. In 1942, the American engineer Hunter Rouse (1906–1996) veri-
fied Colebrook’s equation and produced a graphical plot of f as a function
of Re and the product . He also presented the laminar flow relation
and a table of commercial pipe roughness. Two years later, Lewis F. Moody
(1880–1953) redrew Rouse’s diagram into the form commonly used today.
The now famous
Moody chart is given in the appendix as Fig. A–12. It
presents the Darcy friction factor for pipe flow as a function of the
Reynolds number and e/D over a wide range. It is probably one of the most
widely accepted and used charts in engineering. Although it is developed for
circular pipes, it can also be used for noncircular pipes by replacing the
diameter by the hydraulic diameter.
Commercially available pipes differ from those used in the experiments in
that the roughness of pipes in the market is not uniform and it is difficult to
give a precise description of it. Equivalent roughness values for some com-
mercial pipes are given in Table 8–2 as well as on the Moody chart. But it
should be kept in mind that these values are for new pipes, and the relative
roughness of pipes may increase with use as a result of corrosion, scale
buildup, and precipitation. As a result, the friction factor may increase by a
factor of 5 to 10. Actual operating conditions must be considered in the
design of piping systems. Also, the Moody chart and its equivalent Cole-
brook equation involve several uncertainties (the roughness size, experimen-
tal error, curve fitting of data, etc.), and thus the results obtained should not
be treated as “exact.” It is usually considered to be accurate to 15 percent
over the entire range in the figure.
The Colebrook equation is implicit in f, and thus the determination of the
friction factor requires some iteration unless an equation solver such as EES
is used. An approximate explicit relation for f was given by S. E. Haaland in
1983 as
(8–51)
The results obtained from this relation are within 2 percent of those
obtained from the Colebrook equation. If more accurate results are desired,
Eq. 8–51 can be used as a good first guess in a Newton iteration when using
a programmable calculator or a spreadsheet to solve for f with Eq. 8–50.
1
2f
⬵ 1.8 logc
6.9
Re
a
e/D
3.7
b
1.11
d
Re1f
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b (turbulent flow)
341
CHAPTER 8
TABLE 8–2
Equivalent roughness values for new
commercial pipes*
Roughness, e
Material ft mm
Glass, plastic 0 (smooth)
Concrete 0.003–0.03 0.9–9
Wood stave 0.0016 0.5
Rubber,
smoothed 0.000033 0.01
Copper or
brass tubing 0.000005 0.0015
Cast iron 0.00085 0.26
Galvanized
iron 0.0005 0.15
Wrought iron 0.00015 0.046
Stainless steel 0.000007 0.002
Commercial
steel 0.00015 0.045
* The uncertainty in these values can be as much
as 60 percent.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 341
We make the following observations from the Moody chart:
• For laminar flow, the friction factor decreases with increasing Reynolds
number, and it is independent of surface roughness.
• The friction factor is a minimum for a smooth pipe (but still not zero
because of the no-slip condition) and increases with roughness (Fig.
8–27). The Colebrook equation in this case (e 0) reduces to the
Prandtl equation expressed as .
• The transition region from the laminar to turbulent regime (2300 Re
4000) is indicated by the shaded area in the Moody chart (Figs. 8–28
and A–12). The flow in this region may be laminar or turbulent,
depending on flow disturbances, or it may alternate between laminar and
turbulent, and thus the friction factor may also alternate between the
values for laminar and turbulent flow. The data in this range are the least
reliable. At small relative roughnesses, the friction factor increases in the
transition region and approaches the value for smooth pipes.
• At very large Reynolds numbers (to the right of the dashed line on the
chart) the friction factor curves corresponding to specified relative
roughness curves are nearly horizontal, and thus the friction factors are
independent of the Reynolds number (Fig. 8–28). The flow in that region
is called fully rough turbulent flow or just fully rough flow because the
thickness of the viscous sublayer decreases with increasing Reynolds
number, and it becomes so thin that it is negligibly small compared to the
surface roughness height. The viscous effects in this case are produced
in the main flow primarily by the protruding roughness elements, and
the contribution of the laminar sublayer is negligible. The Colebrook
equation in the fully rough zone (Re → ) reduces to the von Kármán
equation expressed as which is explicit in
f. Some authors call this zone completely (or fully) turbulent flow, but this
is misleading since the flow to the left of the dashed blue line in Fig. 8–28
is also fully turbulent.
In calculations, we should make sure that we use the actual internal diame-
ter of the pipe, which may be different than the nominal diameter. For
example, the internal diameter of a steel pipe whose nominal diameter is
1 in is 1.049 in (Table 8–3).
1/1f
2.0 log[(e/D)/3.7],
1/1f
2.0 log(Re1f ) 0.8
342
FLUID MECHANICS
Relative Friction
Roughness, Factor,
/Df
0.0* 0.0119
0.00001 0.0119
0.0001 0.0134
0.0005 0.0172
0.001 0.0199
0.005 0.0305
0.01 0.0380
0.05 0.0716
* Smooth surface. All values are for Re 10
6
and are calculated from the Colebrook equation.
FIGURE 8–27
The friction factor is minimum for a
smooth pipe and increases with
roughness.
e
/D = 0.001
0.1
0.01
0.001
10
3
10
4
10
5
10
6
10
7
10
8
Re
ƒ
Transitional
Laminar
Fully rough turbulent flow (ƒ levels off)
e
/D = 0.01
e
/D = 0.0001
e
/D = 0
Smooth turbulent
FIGURE 8–28
At very large Reynolds numbers, the
friction factor curves on the Moody
chart are nearly horizontal, and thus
the friction factors are independent
of the Reynolds number.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 342
Types of Fluid Flow Problems
In the design and analysis of piping systems that involve the use of the
Moody chart (or the Colebrook equation), we usually encounter three types
of problems (the fluid and the roughness of the pipe are assumed to be spec-
ified in all cases) (Fig. 8–29):
1. Determining the pressure drop (or head loss) when the pipe length and
diameter are given for a specified flow rate (or velocity)
2. Determining the flow rate when the pipe length and diameter are given
for a specified pressure drop (or head loss)
3. Determining the pipe diameter when the pipe length and flow rate are
given for a specified pressure drop (or head loss)
Problems of the first type are straightforward and can be solved directly
by using the Moody chart. Problems of the second type and third type are
commonly encountered in engineering design (in the selection of pipe diam-
eter, for example, that minimizes the sum of the construction and pumping
costs), but the use of the Moody chart with such problems requires an itera-
tive approach unless an equation solver is used.
In problems of the second type, the diameter is given but the flow rate is
unknown. A good guess for the friction factor in that case is obtained from
the completely turbulent flow region for the given roughness. This is true
for large Reynolds numbers, which is often the case in practice. Once the
flow rate is obtained, the friction factor can be corrected using the Moody
chart or the Colebrook equation, and the process is repeated until the solu-
tion converges. (Typically only a few iterations are required for convergence
to three or four digits of precision.)
In problems of the third type, the diameter is not known and thus
the Reynolds number and the relative roughness cannot be calculated.
Therefore, we start calculations by assuming a pipe diameter. The pressure
drop calculated for the assumed diameter is then compared to the specified
pressure drop, and calculations are repeated with another pipe diameter in
an iterative fashion until convergence.
To avoid tedious iterations in head loss, flow rate, and diameter calcula-
tions, Swamee and Jain proposed the following explicit relations in 1976
that are accurate to within 2 percent of the Moody chart:
(8–52)
(8–53)
(8–54)
Note that all quantities are dimensional and the units simplify to the
desired unit (for example, to m or ft in the last relation) when consistent
units are used. Noting that the Moody chart is accurate to within 15 percent
of experimental data, we should have no reservation in using these approx-
imate relations in the design of piping systems.
10
6
e/D 10
2
5000 Re 3 10
8
D 0.66ce
1.25
a
L
V
#
2
gh
L
b
4.75
n
V
#
9.4
a
L
gh
L
b
5.2
d
0.04
Re 2000V
#
0.965a
gD
5
h
L
L
b
0.5
lnc
e
3.7D
a
3.17v
2
L
gD
3
h
L
b
0.5
d
10
6
e/D 10
2
3000 Re 3 10
8
h
L
1.07
V
#
2
L
gD
5
elnc
e
3.7D
4.62a
nD
V
#
b
0.9
df
2
343
CHAPTER 8
TABLE 8–3
Standard sizes for Schedule 40
steel pipes
Nominal Actual Inside
Size, in Diameter, in
0.269
0.364
0.493
0.622
0.824
1 1.049
1.610
2 2.067
2.469
3 3.068
5 5.047
10 10.02
2
1
2
1
1
2
3
4
1
2
3
8
1
4
1
8
, ,
Problem
type
, ,
, ,
(or
Given Find
FIGURE 8–29
The three types of problems
encountered in pipe flow.
cen72367_ch08.qxd 11/4/04 7:13 PM Page 343
EXAMPLE 8–3 Determining the Head Loss in a Water Pipe
Water at 60°F (r 62.36 lbm/ft
3
and m 7.536 10
4
lbm/ft · s) is flow-
ing steadily in a 2-in-diameter horizontal pipe made of stainless steel at a rate
of 0.2 ft
3
/s (Fig. 8–30). Determine the pressure drop, the head loss, and the
required pumping power input for flow over a 200-ft-long section of the pipe.
SOLUTION The flow rate through a specified water pipe is given. The pres-
sure drop, the head loss, and the pumping power requirements are to be
determined.
Assumptions 1 The flow is steady and incompressible. 2 The entrance
effects are negligible, and thus the flow is fully developed. 3 The pipe
involves no components such as bends, valves, and connectors. 4 The piping
section involves no work devices such as a pump or a turbine.
Properties The density and dynamic viscosity of water are given to be r
62.36 lbm/ft
3
and m 7.536 10
4
lbm/ft · s, respectively.
Analysis We recognize this as a problem of the first type, since flow rate,
pipe length, and pipe diameter are known. First we calculate the average
velocity and the Reynolds number to determine the flow regime:
which is greater than 4000. Therefore, the flow is turbulent. The relative
roughness of the pipe is calculated using Table 8–2
The friction factor corresponding to this relative roughness and the Reynolds
number can simply be determined from the Moody chart. To avoid any read-
ing error, we determine
f from the Colebrook equation:
Using an equation solver or an iterative scheme, the friction factor is deter-
mined to be
f 0.0174. Then the pressure drop (which is equivalent to
pressure loss in this case), head loss, and the required power input become
Therefore, power input in the amount of 461 W is needed to overcome the
frictional losses in the pipe.
Discussion It is common practice to write our final answers to three signifi-
cant digits, even though we know that the results are accurate to at most two
significant digits because of inherent inaccuracies in the Colebrook equation,
W
#
pump
V
#
P (0.2 ft
3
/s)(1700 lbf/ft
2
)a
1 W
0.737 lbf ft/s
b
461 W
h
L
P
L
rg
f
L
D
V
2
2g
0.0174
200 ft
2/12 ft
(9.17 ft/s)
2
2(32.2 ft/s
2
)
27.3 ft
1700 lbf/ft
2
11.8 psi
P P
L
f
L
D
rV
2
2
0.0174
200 ft
2/12 ft
(62.36 lbm/ft
3
)(9.17 ft/s)
2
2
a
1 lbf
32.2 lbm ft/s
2
b
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b →
1
2f
2.0 log a
0.000042
3.7
2.51
126,4002f
b
e/D
0.000007 ft
2/12 ft
0.000042
Re
rVD
m
(62.36 lbm/ft
3
)(9.17 ft/s)(2/12 ft)
7.536 10
4
lbm/ft s
126,400
V
V
#
A
c
V
#
pD
2
/4
0.2 ft
3
/s
p(2/12 ft)
2
/4
9.17 ft/s
344
FLUID MECHANICS
200 ft
2 in
0.2 ft
3
/s
water
FIGURE 8–30
Schematic for Example 8–3.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 344
as discussed previously. The friction factor could also be determined easily
from the explicit Haaland relation (Eq. 8–51). It would give
f 0.0172,
which is sufficiently close to 0.0174. Also, the friction factor corresponding
to e 0 in this case is 0.0171, which indicates that stainless-steel pipes
can be assumed to be smooth with negligible error.
EXAMPLE 8–4 Determining the Diameter of an Air Duct
Heated air at 1 atm and 35°C is to be transported in a 150-m-long circular
plastic duct at a rate of 0.35 m
3
/s (Fig. 8–31). If the head loss in the pipe
is not to exceed 20 m, determine the minimum diameter of the duct.
SOLUTION The flow rate and the head loss in an air duct are given. The
diameter of the duct is to be determined.
Assumptions 1 The flow is steady and incompressible. 2 The entrance
effects are negligible, and thus the flow is fully developed. 3 The duct
involves no components such as bends, valves, and connectors. 4 Air is an
ideal gas. 5 The duct is smooth since it is made of plastic. 6 The flow is tur-
bulent (to be verified).
Properties The density, dynamic viscosity, and kinematic viscosity of air at
35°C are r 1.145 kg/m
3
, m 1.895 10
5
kg/m · s, and n 1.655
10
5
m
2
/s.
Analysis This is a problem of the third type since it involves the determina-
tion of diameter for specified flow rate and head loss. We can solve this
problem by three different approaches: (1) an iterative approach by assum-
ing a pipe diameter, calculating the head loss, comparing the result to the
specified head loss, and repeating calculations until the calculated head loss
matches the specified value; (2) writing all the relevant equations (leaving
the diameter as an unknown) and solving them simultaneously using an
equation solver; and (3) using the third Swamee–Jain formula. We will
demonstrate the use of the last two approaches.
The average velocity, the Reynolds number, the friction factor, and the
head loss relations can be expressed as (
D is in m, V is in m/s, and Re and f
are dimensionless)
The roughness is approximately zero for a plastic pipe (Table 8–2). There-
fore, this is a set of four equations in four unknowns, and solving them with
an equation solver such as EES gives
Therefore, the diameter of the duct should be more than 26.7 cm if the
head loss is not to exceed 20 m. Note that Re 4000, and thus the turbu-
lent flow assumption is verified.
D 0.267 m, f 0.0180, V 6.24 m/s, and Re 100,800
h
L
f
L
D
V
2
2g
→ 20 f
150 m
D
V
2
2(9.81 m/s
2
)
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b 2.0 loga
2.51
Re2f
b
Re
VD
n
VD
1.655 10
5
m
2
/s
V
V
#
A
c
V
#
pD
2
/4
0.35 m
3
/s
pD
2
/4
345
CHAPTER 8
150 m
D
0.35 m
3
/s
air
FIGURE 8–31
Schematic for Example 8–4
cen72367_ch08.qxd 11/4/04 7:14 PM Page 345
The diameter can also be determined directly from the third Swamee–Jain
formula to be
Discussion Note that the difference between the two results is less than 2
percent. Therefore, the simple Swamee–Jain relation can be used with confi-
dence. Finally, the first (iterative) approach requires an initial guess for
D. If
we use the Swamee–Jain result as our initial guess, the diameter converges
to
D 0.267 m in short order.
EXAMPLE 8–5 Determining the Flow Rate of Air in a Duct
Reconsider Example 8–4. Now the duct length is doubled while its diameter
is maintained constant. If the total head loss is to remain constant, deter-
mine the drop in the flow rate through the duct.
SOLUTION The diameter and the head loss in an air duct are given. The
drop in the flow rate is to be determined.
Analysis This is a problem of the second type since it involves the determi-
nation of the flow rate for a specified pipe diameter and head loss. The solu-
tion involves an iterative approach since the flow rate (and thus the flow
velocity) is not known.
The average velocity, Reynolds number, friction factor, and the head loss
relations can be expressed as (
D is in m, V is in m/s, and Re and f are
dimensionless)
This is a set of four equations in four unknowns and solving them with an
equation solver such as EES gives
Then the drop in the flow rate becomes
Therefore, for a specified head loss (or available head or fan pumping
power), the flow rate drops by about 31 percent from 0.35 to 0.24 m
3
/s
when the duct length doubles.
V
#
drop
V
#
old
V
#
new
0.35 0.24
0.11 m
3
/s (a drop of 31 percent)
V
#
0.24 m
3
/s, f 0.0195, V 4.23 m/s, and Re 68,300
h
L
f
L
D
V
2
2g
→ 20 f
300 m
0.267 m
V
2
2(9.81 m/s
2
)
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b →
1
2f
2.0 loga
2.51
Re2f
b
Re
VD
n
→ Re
V(0.267 m)
1.655 10
5
m
2
/s
V
V
#
A
c
V
#
pD
2
/4
→ V
V
#
p(0.267 m)
2
/4
0.271 m
0.66c0 (1.655 10
5
m
2
/s)(0.35 m
3
/s)
9.4
a
150 m
(9.81 m/s
2
)(20 m)
b
5.2
d
0.04
D 0.66ce
1.25
a
L
V
#
2
gh
L
b
4.75
n
V
#
9.4
a
L
gh
L
b
5.2
d
0.04
346
FLUID MECHANICS
cen72367_ch08.qxd 11/4/04 7:14 PM Page 346
Alternative Solution If a computer is not available (as in an exam situation),
another option is to set up a
manual iteration loop. We have found that the
best convergence is usually realized by first guessing the friction factor
f,
and then solving for the velocity V. The equation for V as a function of
f is
Average velocity through the pipe:
Now that V is calculated, the Reynolds number can be calculated, from
which a
corrected friction factor is obtained from the Moody chart or the
Colebrook equation. We repeat the calculations with the corrected value of
f
until convergence. We guess f 0.04 for illustration:
Iteration f (guess) V, m/s Re Corrected f
1 0.04 2.955 4.724 10
4
0.0212
2 0.0212 4.059 6.489 10
4
0.01973
3 0.01973 4.207 6.727 10
4
0.01957
4 0.01957 4.224 6.754 10
4
0.01956
5 0.01956 4.225 6.756 10
4
0.01956
Notice that the iteration has converged to three digits in only three iterations
and to four digits in only four iterations. The final results are identical to
those obtained with EES, yet do not require a computer.
Discussion The new flow rate can also be determined directly from the sec-
ond Swamee–Jain formula to be
Note that the result from the Swamee–Jain relation is the same (to two sig-
nificant digits) as that obtained with the Colebrook equation using EES or
using our manual iteration technique. Therefore, the simple Swamee–Jain
relation can be used with confidence.
8–6
■
MINOR LOSSES
The fluid in a typical piping system passes through various fittings, valves,
bends, elbows, tees, inlets, exits, enlargements, and contractions in addition
to the pipes. These components interrupt the smooth flow of the fluid and
cause additional losses because of the flow separation and mixing they
induce. In a typical system with long pipes, these losses are minor com-
pared to the total head loss in the pipes (the
major losses) and are called
minor losses. Although this is generally true, in some cases the minor
losses may be greater than the major losses. This is the case, for example, in
systems with several turns and valves in a short distance. The head loss
introduced by a completely open valve, for example, may be negligible. But
a partially closed valve may cause the largest head loss in the system, as
0.24 m
3
/s
lnc0 a
3.17(1.655 10
5
m
2
/s)
2
(300 m)
(9.81 m/s
2
)(0.267 m)
3
(20 m)
b
0.5
d
0.965a
(9.81 m/s
2
)(0.267 m)
5
(20 m)
300 m
b
0.5
V
#
0.965a
gD
5
h
L
L
b
0.5
lnc
e
3.7D
a
3.17v
2
L
gD
3
h
L
b
0.5
d
V
B
2gh
L
f˛ L / D
347
CHAPTER 8
cen72367_ch08.qxd 11/4/04 7:14 PM Page 347
evidenced by the drop in the flow rate. Flow through valves and fittings is
very complex, and a theoretical analysis is generally not plausible. There-
fore, minor losses are determined experimentally, usually by the manufac-
turers of the components.
Minor losses are usually expressed in terms of the loss coefficient K
L
(also called the resistance coefficient), defined as (Fig. 8–32)
Loss coefficient: (8–55)
where h
L
is the additional irreversible head loss in the piping system caused
by insertion of the component, and is defined as h
L
P
L
/rg. For example,
imagine replacing the valve in Fig. 8–32 with a section of constant diameter
pipe from location 1 to location 2. P
L
is defined as the pressure drop from
1 to 2 for the case with the valve, (P
1
P
2
)
valve
, minus the pressure drop
that would occur in the imaginary straight pipe section from 1 to 2 without
the valve, (P
1
P
2
)
pipe
at the same flow rate. While the majority of the
irreversible head loss occurs locally near the valve, some of it occurs down-
stream of the valve due to induced swirling turbulent eddies that are pro-
duced in the valve and continue downstream. These eddies “waste” mechan-
ical energy because they are ultimately dissipated into heat while the flow in
the downstream section of pipe eventually returns to fully developed condi-
tions. When measuring minor losses in some minor loss components, such
as elbows, for example, location 2 must be considerably far downstream
(tens of pipe diameters) in order to fully account for the additional irre-
versible losses due to these decaying eddies.
When the pipe diameter downstream of the component changes, determi-
nation of the minor loss is even more complicated. In all cases, however, it
is based on the additional irreversible loss of mechanical energy that would
otherwise not exist if the minor loss component were not there. For simplic-
ity, you may think of the minor loss as occurring locally across the minor
loss component, but keep in mind that the component influences the flow
for several pipe diameters downstream. By the way, this is the reason why
most flow meter manufacturers recommend installing their flow meter at
least 10 to 20 pipe diameters downstream of any elbows or valves—this
allows the swirling turbulent eddies generated by the elbow or valve to
largely disappear and the velocity profile to become fully developed before
entering the flow meter. (Most flow meters are calibrated with a fully devel-
oped velocity profile at the flow meter inlet, and yield the best accuracy
when such conditions also exist in the actual application.)
When the inlet diameter equals outlet diameter, the loss coefficient of a
component can also be determined by measuring the pressure loss across the
component and dividing it by the dynamic pressure, K
L
P
L
/( rV
2
). When
the loss coefficient for a component is available, the head loss for that com-
ponent is determined from
Minor loss: (8–56)
The loss coefficient, in general, depends on the geometry of the component
and the Reynolds number, just like the friction factor. However, it is usually
assumed to be independent of the Reynolds number. This is a reasonable
approximation since most flows in practice have large Reynolds numbers
h
L
K
L
V
2
2g
1
2
K
L
h
L
V
2
/(2g)
348
FLUID MECHANICS
(P
1
– P
2
)
valve
1 2
∆P
L
= (P
1
– P
2
)
valve
– (P
1
– P
2
)
pipe
V
1 2
V
(P
1
– P
2
)
pipe
Pipe section without valve:
Pipe section with valve:
FIGURE 8–32
For a constant-diameter section of a
pipe with a minor loss component,
the loss coefficient of the component
(such as the gate valve shown) is
determined by measuring the
additional pressure loss it causes
and dividing it by the dynamic
pressure in the pipe.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 348
and the loss coefficients (including the friction factor) tend to be indepen-
dent of the Reynolds number at large Reynolds numbers.
Minor losses are also expressed in terms of the equivalent length L
equiv
,
defined as (Fig. 8–33)
Equivalent length: (8–57)
where f is the friction factor and D is the diameter of the pipe that contains
the component. The head loss caused by the component is equivalent to the
head loss caused by a section of the pipe whose length is L
equiv
. Therefore,
the contribution of a component to the head loss can be accounted for by
simply adding L
equiv
to the total pipe length.
Both approaches are used in practice, but the use of loss coefficients is
more common. Therefore, we will also use that approach in this book. Once
all the loss coefficients are available, the total head loss in a piping system
is determined from
Total head loss (general):
(8–58)
where i represents each pipe section with constant diameter and j represents
each component that causes a minor loss. If the entire piping system being
analyzed has a constant diameter, Eq. 8–58 reduces to
Total head loss (D constant): (8–59)
where V is the average flow velocity through the entire system (note that
V constant since D constant).
Representative loss coefficients K
L
are given in Table 8–4 for inlets, exits,
bends, sudden and gradual area changes, and valves. There is considerable
uncertainty in these values since the loss coefficients, in general, vary with
the pipe diameter, the surface roughness, the Reynolds number, and the
details of the design. The loss coefficients of two seemingly identical valves
by two different manufacturers, for example, can differ by a factor of 2 or
more. Therefore, the particular manufacturer’s data should be consulted in
the final design of piping systems rather than relying on the representative
values in handbooks.
The head loss at the inlet of a pipe is a strong function of geometry. It is
almost negligible for well-rounded inlets (K
L
0.03 for r/D 0.2), but
increases to about 0.50 for sharp-edged inlets (Fig. 8–34). That is, a sharp-
edged inlet causes half of the velocity head to be lost as the fluid enters the
pipe. This is because the fluid cannot make sharp 90° turns easily, espe-
cially at high velocities. As a result, the flow separates at the corners, and
the flow is constricted into the vena contracta region formed in the midsec-
tion of the pipe (Fig. 8–35). Therefore, a sharp-edged inlet acts like a flow
constriction. The velocity increases in the vena contracta region (and the
pressure decreases) because of the reduced effective flow area and then
decreases as the flow fills the entire cross section of the pipe. There would
be negligible loss if the pressure were increased in accordance with
Bernoulli’s equation (the velocity head would simply be converted into
pressure head). However, this deceleration process is far from ideal and the
h
L, total
af
L
D
a
K
L
b
V
2
2g
a
i
f
i
L
i
D
i
V
2
i
2g
a
j
K
L, j
V
2
j
2g
h
L, total
h
L, major
h
L, minor
h
L
K
L
V
2
2g
f
L
equiv
D
V
2
2g
→ L
equiv
D
f
K
L
349
CHAPTER 8
D
∆P = P
1
– P
2
= P
3
– P
4
D
3 4
1
2
FIGURE 8–33
The head loss caused by a component
(such as the angle valve shown) is
equivalent to the head loss caused by a
section of the pipe whose length is the
equivalent length.
Well-rounded inlet
K
L
= 0.03
Sharp-edged inlet
K
L
= 0.50
Recirculating flow
DD
r
FIGURE 8–34
The head loss at the inlet of a pipe is
almost negligible for well-rounded
inlets (K
L
0.03 for r/D 0.2)
but increases to about 0.50 for
sharp-edged inlets.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 349
TABLE 8–4
Loss coefficients K
L
of various pipe components for turbulent flow (for use in the relation h
L
K
L
V
2
/(2g), where V is the
average velocity in the pipe that contains the component)*
Pipe Inlet
Reentrant: K
L
0.80 Sharp-edged: K
L
0.50 Well-rounded (r/D 0.2): K
L
0.03
(
t D and I 艐 0.1D) Slightly rounded (r/D 0.1): K
L
0.12
(see Fig. 8–36)
Pipe Exit
Reentrant: K
L
a Sharp-edged: K
L
a Rounded: K
L
a
Note: The kinetic energy correction factor is a 2 for fully developed laminar flow, and a ⬇ 1 for fully developed turbulent flow.
Sudden Expansion and Contraction (based on the velocity in the smaller-diameter pipe)
Sudden expansion:
Sudden contraction:
See chart.
Gradual Expansion and Contraction (based on the velocity in the smaller-diameter pipe)
Expansion: Contraction (for u 20°):
K
L
0.02 for u 20° K
L
0.30 for d/D 0.2
K
L
0.04 for u 45° K
L
0.25 for d/D 0.4
K
L
0.07 for u 60° K
L
0.15 for d/D 0.6
K
L
0.10 for d/D 0.8
K
L
a1
d
2
D
2
b
2
D
V
l
t
D
V
D
V
r
V V
V
V
dD
V
dD
0.6
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1.0
K
L
d
2
/D
2
K
L
for sudden
contraction
V
d D
u
V
D d
u
cen72367_ch08.qxd 11/4/04 7:14 PM Page 350
TABLE 8–4 (CONCLUDED)
Bends and Branches
90° smooth bend: 90° miter bend 90° miter bend 45° threaded elbow:
Flanged: K
L
0.3 (without vanes): K
L
1.1 (with vanes): K
L
0.2 K
L
0.4
Threaded:
K
L
0.9
180° return bend: Tee (branch flow): Tee (line flow): Threaded union:
Flanged: K
L
0.2 Flanged: K
L
1.0 Flanged: K
L
0.2 K
L
0.08
Threaded:
K
L
1.5 Threaded: K
L
2.0 Threaded: K
L
0.9
Valves
Globe valve, fully open: K
L
10 Gate valve, fully open: K
L
0.2
Angle valve, fully open: K
L
5 K
L
0.3
Ball valve, fully open: K
L
0.05 K
L
2.1
Swing check valve: K
L
2 K
L
17
* These are representative values for loss coefficients. Actual values strongly depend on the design and manufacture of the components and may differ from the
given values considerably (especially for valves). Actual manufacturer’s data should be used in the final design.
3
4
closed:
1
2
closed:
1
4
closed:
V V V
V
45°
V
V
V
V
21
Head
Pressure head
converted to
velocity head
Remaining
pressure head
Remaining
velocity head
Lost velocity head
Total
head
Pressure
head
P
0
rg
P
1
rg
P
2
rg
V
1
2
2g
V
2
2
/2g
K
L
V
2
/2g
0
Vena contracta
Separated
flow
1
2
1
2
FIGURE 8–35
Graphical representation of flow
contraction and the associated head
loss at a sharp-edged pipe inlet.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 351
viscous dissipation caused by intense mixing and the turbulent eddies con-
vert part of the kinetic energy into frictional heating, as evidenced by a
slight rise in fluid temperature. The end result is a drop in velocity without
much pressure recovery, and the inlet loss is a measure of this irreversible
pressure drop.
Even slight rounding of the edges can result in significant reduction of K
L
,
as shown in Fig. 8–36. The loss coefficient rises sharply (to about K
L
0.8)
when the pipe protrudes into the reservoir since some fluid near the edge
in this case is forced to make a 180° turn.
The loss coefficient for a submerged pipe exit is often listed in hand-
books as K
L
1. More precisely, however, K
L
is equal to the kinetic
energy correction factor a at the exit of the pipe. Although a is indeed
close to 1 for fully developed turbulent pipe flow, it is equal to 2 for fully
developed laminar pipe flow. To avoid possible errors when analyzing
laminar pipe flow, then, it is best to always set K
L
a at a submerged
pipe exit. At any such exit, whether laminar or turbulent, the fluid leaving
the pipe loses all of its kinetic energy as it mixes with the reservoir fluid
and eventually comes to rest through the irreversible action of viscosity.
This is true, regardless of the shape of the exit (Table 8–4 and Fig. 8–37).
Therefore, there is no need to round the pipe exits.
Piping systems often involve sudden or gradual expansion or contraction
sections to accommodate changes in flow rates or properties such as density
and velocity. The losses are usually much greater in the case of sudden expan-
sion and contraction (or wide-angle expansion) because of flow separation.
By combining the conservation of mass, momentum, and energy equations,
the loss coefficient for the case of sudden expansion is approximated as
(8–60)
where A
small
and A
large
are the cross-sectional areas of the small and large
pipes, respectively. Note that K
L
0 when there is no area change (A
small
A
large
) and K
L
1 when a pipe discharges into a reservoir (A
large
A
small
). No such relation exists for a sudden contraction, and the K
L
val-
ues in that case can be read from the chart in Table 8–4. The losses due to
expansion and contraction can be reduced significantly by installing conical
gradual area changers (nozzles and diffusers) between the small and large
K
L
a1
A
small
A
large
b
2
(sudden expansion)
352
FLUID MECHANICS
0.050 0.10 0.15 0.20
0.5
0.4
0.3
0.2
0.1
0
r/D
K
L
0.25
D
r
FIGURE 8–36
The effect of rounding of a pipe inlet
on the loss coefficient.
From ASHRAE Handbook of Fundamentals.
Mixing
Entrained
ambient fluid
Submerged
outlet
FIGURE 8–37
All the kinetic energy of the flow is
“lost” (turned into thermal energy)
through friction as the jet decelerates
and mixes with ambient fluid
downstream of a submerged outlet.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 352
pipes. The K
L
values for representative cases of gradual expansion and con-
traction are given in Table 8–4. Note that in head loss calculations, the
velocity in the small pipe is to be used as the reference velocity in Eq. 8–56.
Losses during expansion are usually much higher than the losses during
contraction because of flow separation.
Piping systems also involve changes in direction without a change in
diameter, and such flow sections are called bends or elbows. The losses in
these devices are due to flow separation (just like a car being thrown off the
road when it enters a turn too fast) on the inner side and the swirling
secondary flows caused by different path lengths. The losses during changes
of direction can be minimized by making the turn “easy” on the fluid by
using circular arcs (like the 90° elbow) instead of sharp turns (like miter
bends) (Fig. 8–38). But the use of sharp turns (and thus suffering a penalty
in loss coefficient) may be necessary when the turning space is limited. In
such cases, the losses can be minimized by utilizing properly placed guide
vanes to help the flow turn in an orderly manner without being thrown off
the course. The loss coefficients for some elbows and miter bends as well as
tees are given in Table 8–4. These coefficients do not include the frictional
losses along the pipe bend. Such losses should be calculated as in straight
pipes (using the length of the centerline as the pipe length) and added to
other losses.
Valves are commonly used in piping systems to control the flow rates by
simply altering the head loss until the desired flow rate is achieved. For
valves it is desirable to have a very low loss coefficient when they are fully
open so that they cause minimal head loss during full-load operation. Sev-
eral different valve designs, each with its own advantages and disadvan-
tages, are in common use today. The gate valve slides up and down like a
gate, the globe valve closes a hole placed in the valve, the angle valve is a
globe valve with a 90° turn, and the check valve allows the fluid to flow
only in one direction like a diode in an electric circuit. Table 8–4 lists the
representative loss coefficients of the popular designs. Note that the loss
coefficient increases drastically as a valve is closed (Fig. 8–39). Also, the
deviation in the loss coefficients for different manufacturers is greatest for
valves because of their complex geometries.
EXAMPLE 8–6 Head Loss and Pressure Rise
during Gradual Expansion
A 6-cm-diameter horizontal water pipe expands gradually to a 9-cm-diameter
pipe (Fig. 8–40). The walls of the expansion section are angled 30° from the
horizontal. The average velocity and pressure of water before the expansion
section are 7 m/s and 150 kPa, respectively. Determine the head loss in the
expansion section and the pressure in the larger-diameter pipe.
SOLUTION A horizontal water pipe expands gradually into a larger-diameter
pipe. The head loss and pressure after the expansion are to be determined.
Assumptions 1 The flow is steady and incompressible. 2 The flow at sec-
tions 1 and 2 is fully developed and turbulent with a
1
a
2
艑 1.06.
Properties We take the density of water to be r 1000 kg/m
3
. The loss coef-
ficient for gradual expansion of u 60° total included angle is
K
L
0.07.
353
CHAPTER 8
Flanged elbow
K
L
= 0.3
Sharp turn
K
L
= 1.1
FIGURE 8–38
The losses during changes of direction
can be minimized by making the turn
“easy” on the fluid by using circular
arcs instead of sharp turns.
V
2
=
V
1
V
constriction
>
V
1
V
1
V
2
Constriction
A globe
valve
FIGURE 8–39
The large head loss in a partially
closed valve is due to irreversible
deceleration, flow separation, and
mixing of high-velocity fluid coming
from the narrow valve passage.
9 cm6 cm
Water
7 m/s
150 kPa
1 2
FIGURE 8–40
Schematic for Example 8–6.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 353
Analysis Noting that the density of water remains constant, the downstream
velocity of water is determined from conservation of mass to be
Then the irreversible head loss in the expansion section becomes
Noting that
z
1
z
2
and there are no pumps or turbines involved, the energy
equation for the expansion section can be expressed in terms of heads as
Solving for
P
2
and substituting,
Therefore, despite the head (and pressure) loss, the pressure
increases from
150 to 169 kPa after the expansion. This is due to the conversion of
dynamic pressure to static pressure when the average flow velocity is
decreased in the larger pipe.
Discussion It is common knowledge that higher pressure upstream is neces-
sary to cause flow, and it may come as a surprise to you that the downstream
pressure has
increased after the expansion, despite the loss. This is because
the flow is driven by the sum of the three heads that comprise the total head
(namely, the pressure head, velocity head, and elevation head). During flow
expansion, the higher velocity head upstream is converted to pressure head
downstream, and this increase outweighs the nonrecoverable head loss. Also,
you may be tempted to solve this problem using the Bernoulli equation.
Such a solution would ignore the head (and the associated pressure) loss
and result in an incorrect higher pressure for the fluid downstream.
8–7
■
PIPING NETWORKS AND PUMP SELECTION
Most piping systems encountered in practice such as the water distribution
systems in cities or commercial or residential establishments involve numer-
ous parallel and series connections as well as several sources (supply of
fluid into the system) and loads (discharges of fluid from the system) (Fig.
8–41). A piping project may involve the design of a new system or the
169 kPa
a
1 kN
1000 kg m/s
ba
1 kPa
1 kN/m
2
b
e
1.06(7 m/s)
2
1.06(3.11 m/s)
2
2
(9.81 m/s
2
)(0.175 m)f
P
2
P
1
re
a
1
V
2
1
a
2
V
2
2
2
gh
L
f (150 kPa) (1000 kg/m
3
)
→
P
1
rg
a
1
V
2
1
2g
P
2
rg
a
2
V
2
2
2g
h
L
P
1
rg
a
1
V
2
1
2g
z
1
h
pump, u
P
2
rg
a
2
V
2
2
2g
z
2
h
turbine, e
h
L
h
L
K
L
V
2
1
2g
(0.07)
(7 m/s)
2
2(9.81 m/s
2
)
0.175 m
V
2
(0.06 m)
2
(0.09 m)
2
(7 m/s) 3.11 m/s
m
#
1
m
#
2
→ rV
1
A
1
rV
2
A
2
→ V
2
A
1
A
2
V
1
D
2
1
D
2
2
V
1
354
FLUID MECHANICS
FIGURE 8–41
A piping network in an industrial
facility.
Courtesy UMDE Engineering, Contracting,
and Trading. Used by permission.
S
00
¡
cen72367_ch08.qxd 11/4/04 7:14 PM Page 354
expansion of an existing system. The engineering objective in such projects
is to design a piping system that will deliver the specified flow rates at spec-
ified pressures reliably at minimum total (initial plus operating and mainte-
nance) cost. Once the layout of the system is prepared, the determination of
the pipe diameters and the pressures throughout the system, while remain-
ing within the budget constraints, typically requires solving the system
repeatedly until the optimal solution is reached. Computer modeling and
analysis of such systems make this tedious task a simple chore.
Piping systems typically involve several pipes connected to each other in
series and/or in parallel, as shown in Figs. 8–42 and 8–43. When the pipes
are connected in series, the flow rate through the entire system remains
constant regardless of the diameters of the individual pipes in the system.
This is a natural consequence of the conservation of mass principle for
steady incompressible flow. The total head loss in this case is equal to the
sum of the head losses in individual pipes in the system, including the
minor losses. The expansion or contraction losses at connections are consid-
ered to belong to the smaller-diameter pipe since the expansion and contrac-
tion loss coefficients are defined on the basis of the average velocity in the
smaller-diameter pipe.
For a pipe that branches out into two (or more) parallel pipes and then
rejoins at a junction downstream, the total flow rate is the sum of the flow
rates in the individual pipes. The pressure drop (or head loss) in each indi-
vidual pipe connected in parallel must be the same since P P
A
P
B
and
the junction pressures P
A
and P
B
are the same for all the individual pipes.
For a system of two parallel pipes 1 and 2 between junctions A and B with
negligible minor losses, this can be expressed as
Then the ratio of the average velocities and the flow rates in the two parallel
pipes become
Therefore, the relative flow rates in parallel pipes are established from the
requirement that the head loss in each pipe be the same. This result can be
extended to any number of pipes connected in parallel. The result is also
valid for pipes for which the minor losses are significant if the equivalent
lengths for components that contribute to minor losses are added to the pipe
V
1
V
2
a
f
2
f
1
L
2
L
1
D
1
D
2
b
1/2
and
V
#
1
V
#
2
A
c, 1
V
1
A
c, 2
V
2
D
2
1
D
2
2
a
f
2
f
1
L
2
L
1
D
1
D
2
b
1/2
h
L, 1
h
L, 2
→ f
1
L
1
D
1
V
2
1
2g
f
2
L
2
D
2
V
2
2
2g
355
CHAPTER 8
A
f
A
, L
A
, D
A
V
A
=
V
B
h
L, 1-2
= h
L, A
+ h
L, B
⋅⋅
f
B
, L
B
, D
B
B
12
FIGURE 8–42
For pipes in series, the flow rate is the
same in each pipe, and the total head
loss is the sum of the head losses in
individual pipes.
P
A
A
B
P
B
< P
A
A B
h
L, 1
= h
L, 2
V
A
=
V
1
+
V
2
=
V
B
⋅⋅⋅⋅
f
1
, L
1
, D
1
f
2
, L
2
, D
2
FIGURE 8–43
For pipes in parallel, the head loss is
the same in each pipe, and the total
flow rate is the sum of the flow rates
in individual pipes.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 355
length. Note that the flow rate in one of the parallel branches is proportional
to its diameter to the power 5/2 and is inversely proportional to the square
root of its length and friction factor.
The analysis of piping networks, no matter how complex they are, is
based on two simple principles:
1. Conservation of mass throughout the system must be satisfied. This is
done by requiring the total flow into a junction to be equal to the total
flow out of the junction for all junctions in the system. Also, the flow
rate must remain constant in pipes connected in series regardless of the
changes in diameters.
2. Pressure drop (and thus head loss) between two junctions must be the
same for all paths between the two junctions. This is because pressure is
a point function and it cannot have two values at a specified point. In
practice this rule is used by requiring that the algebraic sum of head
losses in a loop (for all loops) be equal to zero. (A head loss is taken to
be positive for flow in the clockwise direction and negative for flow in
the counterclockwise direction.)
Therefore, the analysis of piping networks is very similar to the analysis of
electric circuits, with flow rate corresponding to electric current and pressure
corresponding to electric potential. However, the situation is much more
complex here since, unlike the electrical resistance, the “flow resistance” is a
highly nonlinear function. Therefore, the analysis of piping networks
requires the simultaneous solution of a system of nonlinear equations. The
analysis of such systems is beyond the scope of this introductory text.
Piping Systems with Pumps and Turbines
When a piping system involves a pump and/or turbine, the steady-flow
energy equation on a unit-mass basis can be expressed as (see Section 5–7)
(8–61)
It can also be expressed in terms of heads as
(8–62)
where h
pump, u
w
pump, u
/g is the useful pump head delivered to the fluid,
h
turbine, e
w
turbine, e
/g is the turbine head extracted from the fluid, a is the
kinetic energy correction factor whose value is nearly 1 for most (turbulent)
flows encountered in practice, and h
L
is the total head loss in piping (includ-
ing the minor losses if they are significant) between points 1 and 2. The
pump head is zero if the piping system does not involve a pump or a fan,
the turbine head is zero if the system does not involve a turbine, and both
are zero if the system does not involve any mechanical work-producing or
work-consuming devices.
Many practical piping systems involve a pump to move a fluid from one
reservoir to another. Taking points 1 and 2 to be at the free surfaces of the
reservoirs, the energy equation in this case reduces for the useful pump head
required to (Fig. 8–44)
(8–63)h
pump, u
(z
2
z
1
) h
L
P
1
rg
a
1
V
2
1
2g
z
1
h
pump, u
P
2
rg
a
2
V
2
2
2g
z
2
h
turbine, e
h
L
P
1
r
a
1
V
2
1
2
gz
1
w
pump, u
P
2
r
a
2
V
2
2
2
gz
2
w
turbine, e
gh
L
356
FLUID MECHANICS
cen72367_ch08.qxd 11/4/04 7:14 PM Page 356
since the velocities at free surfaces are negligible and the pressures are at
atmospheric pressure. Therefore, the useful pump head is equal to the eleva-
tion difference between the two reservoirs plus the head loss. If the head
loss is negligible compared to z
2
z
1
, the useful pump head is simply equal
to the elevation difference between the two reservoirs. In the case of z
1
z
2
(the first reservoir being at a higher elevation than the second one) with no
pump, the flow is driven by gravity at a flow rate that causes a head loss
equal to the elevation difference. A similar argument can be given for the
turbine head for a hydroelectric power plant by replacing h
pump, u
in Eq. 8–63
by h
turbine, e
.
Once the useful pump head is known, the mechanical power that needs to
be delivered by the pump to the fluid and the electric power consumed by
the motor of the pump for a specified flow rate are determined from
(8–64)
where h
pump–motor
is the efficiency of the pump–motor combination, which is
the product of the pump and the motor efficiencies (Fig. 8–45). The
pump–motor efficiency is defined as the ratio of the net mechanical energy
delivered to the fluid by the pump to the electric energy consumed by the
motor of the pump, and it usually ranges between 50 and 85 percent.
The head loss of a piping system increases (usually quadratically) with the
flow rate. A plot of required useful pump head h
pump, u
as a function of flow
rate is called the system (or demand) curve. The head produced by a pump
is not a constant either. Both the pump head and the pump efficiency vary
with the flow rate, and pump manufacturers supply this variation in tabular
or graphical form, as shown in Fig. 8–46. These experimentally determined
h
pump, u
and h
pump, u
versus
V
.
curves are called characteristic (or supply or
performance) curves. Note that the flow rate of a pump increases as the
required head decreases. The intersection point of the pump head curve with
the vertical axis typically represents the maximum head the pump can pro-
vide, while the intersection point with the horizontal axis indicates the maxi-
mum flow rate (called the free delivery) that the pump can supply.
The efficiency of a pump is sufficiently high for a certain range of head
and flow rate combination. Therefore, a pump that can supply the required
head and flow rate is not necessarily a good choice for a piping system
unless the efficiency of the pump at those conditions is sufficiently high.
The pump installed in a piping system will operate at the point where the
system curve and the characteristic curve intersect. This point of intersec-
tion is called the operating point, as shown in Fig. 8–46. The useful head
W
#
pump, shaft
r
V
#
gh
pump, u
h
pump
and W
#
elect
r
V
#
gh
pump, u
h
pump–motor
357
CHAPTER 8
z
1
z
2
Pump
Control volume
boundary
h
pump, u
= (z
2
– z
1
) + h
L
W
pump, u
= r
V
gh
pump, u
⋅⋅
1
2
FIGURE 8–44
When a pump moves a fluid from one
reservoir to another, the useful pump
head requirement is equal to the
elevation difference between the two
reservoirs plus the head loss.
Liquid in
Liquid out
Motor
h
motor
= 0.90
h
pump
= 0.70
h
pump–motor
= h
pump
h
motor
= 0.70 0.90 = 0.63
Pump
FIGURE 8–45
The efficiency of the pump–motor
combination is the product of the
pump and the motor efficiencies.
Courtesy Yunus Çengel
cen72367_ch08.qxd 11/4/04 7:14 PM Page 357
produced by the pump at this point matches the head requirements of the
system at that flow rate. Also, the efficiency of the pump during operation is
the value corresponding to that flow rate.
EXAMPLE 8–7 Pumping Water through Two Parallel Pipes
Water at 20°C is to be pumped from a reservoir (
z
A
5 m) to another reser-
voir at a higher elevation (
z
B
13 m) through two 36-m-long pipes con-
nected in parallel, as shown in Fig. 8–47. The pipes are made of commercial
steel, and the diameters of the two pipes are 4 and 8 cm. Water is to be
pumped by a 70 percent efficient motor–pump combination that draws 8 kW
of electric power during operation. The minor losses and the head loss in
pipes that connect the parallel pipes to the two reservoirs are considered to
be negligible. Determine the total flow rate between the reservoirs and the
flow rate through each of the parallel pipes.
SOLUTION The pumping power input to a piping system with two parallel
pipes is given. The flow rates are to be determined.
358
FLUID MECHANICS
1
2
z
A
= 5 m
L
1
= 36 m
D
1
= 4 cm
Control
volume
boundary
A
Pump
z
B
= 13 m
B
D
2
= 8 cm
L =36m
FIGURE 8–47
The piping system discussed in
Example 8–7.
40
30
20
10
0
80
60
40
20
0
100
123
Flow rate, m
3
/s
4560
Head, m
Pump efficiency, % h
pump
Operating
point
No pipe is attached
to the pump (no load
to maximize flow rate)
System curve
Pump exit is closed to produce maximum head
h
pump, u
h
pump
Supply
curve
FIGURE 8–46
Characteristic pump curves for
centrifugal pumps, the system curve
for a piping system, and the operating
point.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 358
Assumptions 1 The flow is steady and incompressible. 2 The entrance
effects are negligible, and thus the flow is fully developed. 3 The elevations
of the reservoirs remain constant. 4 The minor losses and the head loss in
pipes other than the parallel pipes are said to be negligible. 5 Flows through
both pipes are turbulent (to be verified).
Properties The density and dynamic viscosity of water at 20°C are r
998 kg/m
3
and m 1.002 10
3
kg/m · s. The roughness of commer-
cial steel pipe is e 0.000045 m.
Analysis This problem cannot be solved directly since the velocities (or flow
rates) in the pipes are not known. Therefore, we would normally use a trial-
and-error approach here. However, nowadays equation solvers such as EES
are widely available, and thus we will simply set up the equations to be
solved by an equation solver. The useful head supplied by the pump to the
fluid is determined from
(1)
We choose points
A and B at the free surfaces of the two reservoirs. Noting
that the fluid at both points is open to the atmosphere (and thus
P
A
P
B
P
atm
) and that the fluid velocities at both points are zero (V
A
V
B
0), the
energy equation for a control volume between these two points simplifies to
or
(2)
where
(3)(4)
We designate the 4-cm-diameter pipe by 1 and the 8-cm-diameter pipe by
2. The average velocity, the Reynolds number, the friction factor, and the
head loss in each pipe are expressed as
(5)
(6)
(7)
(8)
(9)
→
1
2f
1
2.0 loga
0.000045
3.7 0.04
2.51
Re
1
2f
1
b
1
2f
1
2.0 loga
e/D
1
3.7
2.51
Re
1
2f
1
b
Re
2
rV
2
D
2
m
→ Re
2
(998 kg/m
3
)V
2
(0.08 m)
1.002 10
3
kg/m s
Re
1
rV
1
D
1
m
→ Re
1
(998 kg/m
3
)V
1
(0.04 m)
1.002 10
3
kg/m s
V
2
V
#
2
A
c, 2
V
#
2
pD
2
2
/4
→ V
2
V
#
2
p(0.08 m)
2
/4
V
1
V
#
1
A
c, 1
V
#
1
pD
2
1
/4
→ V
1
V
#
1
p(0.04 m)
2
/4
h
L
h
L, 1
h
L, 2
h
pump, u
(13 5) h
L
(z
B
z
A
) h
L
P
A
rg
a
A
V
2
A
2g
z
A
h
pump, u
P
B
rg
a
B
V
2
B
2g
z
B
h
L
→ h
pump, u
W
#
elect
r
V
#
gh
pump, u
h
pumpmotor
→ 8000 W
(998 kg/m
3
)
V
#
(9.81 m/s
2
)h
pump, u
0.70
359
CHAPTER 8
00
QQ
cen72367_ch08.qxd 11/4/04 7:14 PM Page 359
(10)
(11)
(12)
(13)
This is a system of 13 equations in 13 unknowns, and their simultaneous
solution by an equation solver gives
Note that Re 4000 for both pipes, and thus the assumption of turbulent
flow is verified.
Discussion The two parallel pipes are identical, except the diameter of the
first pipe is half the diameter of the second one. But only 14 percent of the
water flows through the first pipe. This shows the strong dependence of
the flow rate (and the head loss) on diameter. Also, it can be shown that if
the free surfaces of the two reservoirs were at the same elevation (and thus
z
A
z
B
), the flow rate would increase by 20 percent from 0.0300 to
0.0361 m
3
/s. Alternately, if the reservoirs were as given but the irreversible
head losses were negligible, the flow rate would become 0.0715 m
3
/s (an
increase of 138 percent).
EXAMPLE 8–8 Gravity-Driven Water Flow in a Pipe
Water at 10°C flows from a large reservoir to a smaller one through a 5-cm-
diameter cast iron piping system, as shown in Fig. 8–48. Determine the ele-
vation
z
1
for a flow rate of 6 L/s.
Re
1
131,600, Re
2
410,000, f
1
0.0221, f
2
0.0182
V
1
3.30 m/s, V
2
5.15 m/s, h
L
h
L, 1
h
L, 2
11.1 m, h
pump
19.1 m
V
#
0.0300 m
3
/s,
V
#
1
0.00415 m
3
/s, V
#
2
0.0259 m
3
/s
V
#
V
#
1
V
#
2
h
L, 2
f
2
L
2
D
2
V
2
2
2g
→ h
L, 2
f
2
36 m
0.08 m
V
2
2
2(9.81 m/s
2
)
h
L, 1
f
1
L
1
D
1
V
2
1
2g
→ h
L, 1
f
1
36 m
0.04 m
V
2
1
2(9.81 m/s
2
)
→
1
2f
2
2.0 loga
0.000045
3.7 0.08
2.51
Re
2
2f
2
b
1
2f
2
2.0 loga
e/D
2
3.7
2.51
Re
2
2f
2
b
360
FLUID MECHANICS
1
z
1
= ?
2
z
2
= 4 m
D = 5 cm
9 m
80 m
Standard elbow,
flanged, K
L
= 0.3
Gate valve,
fully open
K
L
= 0.2
Sharp-edged
entrance, K
L
= 0.5
Control
volume
boundary
Exit, K
L
= 1.06
FIGURE 8–48
The piping system discussed in
Example 8–8.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 360
SOLUTION The flow rate through a piping system connecting two reservoirs
is given. The elevation of the source is to be determined.
Assumptions 1 The flow is steady and incompressible. 2 The elevations of
the reservoirs remain constant. 3 There are no pumps or turbines in the line.
Properties The density and dynamic viscosity of water at 10°C are r
999.7 kg/m
3
and m 1.307 10
3
kg/m · s. The roughness of cast iron
pipe is e 0.00026 m.
Analysis The piping system involves 89 m of piping, a sharp-edged
entrance (
K
L
0.5), two standard flanged elbows (K
L
0.3 each), a fully
open gate valve (
K
L
0.2), and a submerged exit (K
L
1.06). We choose
points 1 and 2 at the free surfaces of the two reservoirs. Noting that the
fluid at both points is open to the atmosphere (and thus
P
1
P
2
P
atm
)
and that the fluid velocities at both points are zero (
V
1
V
2
0), the
energy equation for a control volume between these two points simplifies to
where
since the diameter of the piping system is constant. The average velocity in
the pipe and the Reynolds number are
The flow is turbulent since Re 4000. Noting that e/
D 0.00026/0.05
0.0052, the friction factor can be determined from the Colebrook equa-
tion (or the Moody chart),
It gives
f 0.0315. The sum of the loss coefficients is
Then the total head loss and the elevation of the source become
Therefore, the free surface of the first reservoir must be 31.9 m above the
ground level to ensure water flow between the two reservoirs at the specified
rate.
Discussion Note that fL/D 56.1 in this case, which is about 24 times the
total minor loss coefficient. Therefore, ignoring the sources of minor losses
in this case would result in about 4 percent error.
It can be shown that the total head loss would be 35.9 m (instead of
27.9 m) if the valve were three-fourths closed, and it would drop to 24.8 m
if the pipe between the two reservoirs were straight at the ground level (thus
z
1
z
2
h
L
4 27.9 31.9 m
h
L
af
L
D
a
K
L
b
V
2
2g
a0.0315
89 m
0.05 m
2.36b
(3.06 m/s)
2
2(9.81 m/s
2
)
27.9 m
0.5 2 0.3 0.2 1.06 2.36
a
K
L
K
L, entrance
2K
L, elbow
K
L, valve
K
L, exit
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b →
1
2f
2.0 loga
0.0052
3.7
2.51
117,0002f
b
Re
rVD
m
(999.7 kg/m
3
)(3.06 m/s)(0.05 m)
1.307 10
3
kg/m s
117,000
V
V
#
A
c
V
#
pD
2
/4
0.006 m
3
/s
p(0.05 m)
2
/4
3.06 m/s
h
L
h
L, total
h
L, major
h
L, minor
af
L
D
a
K
L
b
V
2
2g
P
1
rg
a
1
V
2
1
2g
z
1
P
2
rg
a
2
V
2
2
2g
z
2
h
L
→ z
1
z
2
h
L
361
CHAPTER 8
00
QQ
cen72367_ch08.qxd 11/4/04 7:14 PM Page 361
eliminating the elbows and the vertical section of the pipe). The head loss
could be reduced further (from 24.8 to 24.6 m) by rounding the entrance.
The head loss can be reduced significantly (from 27.9 to 16.0 m) by replac-
ing the cast iron pipes by smooth pipes such as those made of plastic.
EXAMPLE 8–9 Effect of Flushing on Flow Rate from a Shower
The bathroom plumbing of a building consists of 1.5-cm-diameter copper
pipes with threaded connectors, as shown in Fig. 8–49. (
a) If the gage pres-
sure at the inlet of the system is 200 kPa during a shower and the toilet
reservoir is full (no flow in that branch), determine the flow rate of water
through the shower head. (
b) Determine the effect of flushing of the toilet on
the flow rate through the shower head. Take the loss coefficients of the
shower head and the reservoir to be 12 and 14, respectively.
SOLUTION The cold-water plumbing system of a bathroom is given. The
flow rate through the shower and the effect of flushing the toilet on the flow
rate are to be determined.
Assumptions 1 The flow is steady and incompressible. 2 The flow is turbu-
lent and fully developed. 3 The reservoir is open to the atmosphere. 4 The
velocity heads are negligible.
Properties The properties of water at 20°C are r 998 kg/m
3
, m 1.002
10
3
kg/m · s, and n m/r 1.004 10
6
m
2
/s. The roughness of
copper pipes is e 1.5 10
6
m.
Analysis This is a problem of the second type since it involves the determi-
nation of the flow rate for a specified pipe diameter and pressure drop. The
solution involves an iterative approach since the flow rate (and thus the flow
velocity) is not known.
(
a) The piping system of the shower alone involves 11 m of piping, a tee
with line flow (
K
L
0.9), two standard elbows (K
L
0.9 each), a fully open
globe valve (
K
L
10), and a shower head (K
L
12). Therefore, 兺K
L
0.9
2 0.9 10 12 24.7. Noting that the shower head is open to the
atmosphere, and the velocity heads are negligible, the energy equation for a
control volume between points 1 and 2 simplifies to
→
P
1, gage
rg
(z
2
z
1
) h
L
P
1
rg
a
1
V
2
1
2g
z
1
h
pump, u
P
2
rg
a
2
V
2
2
2g
z
2
h
turbine, e
h
L
362
FLUID MECHANICS
5 m 4 m
Toilet reservoir
with float
K
L
= 14
K
L
= 0.9
K
L
= 2
K
L
= 10
K
L
= 12
Shower head
Globe valve,
fully open
K
L
= 10
Cold
water
1 m
2 m
3
1
2
FIGURE 8–49
Schematic for Example 8–9.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 362
Therefore, the head loss is
Also,
since the diameter of the piping system is constant. The average velocity in
the pipe, the Reynolds number, and the friction factor are
This is a set of four equations with four unknowns, and solving them with an
equation solver such as EES gives
Therefore, the flow rate of water through the shower head is
0.53 L/s.
(b) When the toilet is flushed, the float moves and opens the valve. The dis-
charged water starts to refill the reservoir, resulting in parallel flow after the
tee connection. The head loss and minor loss coefficients for the shower
branch were determined in (
a) to be h
L, 2
18.4 m and
K
L, 2
24.7,
respectively. The corresponding quantities for the reservoir branch can be
determined similarly to be
The relevant equations in this case are
Re
1
V
1
(0.015 m)
1.004 10
6
m
2
/s
, Re
2
V
2
(0.015 m)
1.004 10
6
m
2
/s
, Re
3
V
3
(0.015 m)
1.004 10
6
m
2
/s
V
1
V
#
1
p(0.015 m)
2
/4
, V
2
V
#
2
p(0.015 m)
2
/4
, V
3
V
#
3
p(0.015 m)
2
/4
h
L, 3
f
1
5 m
0.015 m
V
2
1
2(9.81 m/s
2
)
af
3
1 m
0.015 m
26.9b
V
2
3
2(9.81 m/s
2
)
19.4
h
L, 2
f
1
5 m
0.015 m
V
2
1
2(9.81 m/s
2
)
af
2
6 m
0.015 m
24.7b
V
2
2
2(9.81 m/s
2
)
18.4
V
#
1
V
#
2
V
#
3
K
L, 3
2 10 0.9 14 26.9
h
L, 3
200,000 N/m
2
(998 kg/m
3
)(9.81 m/s
2
)
1 m 19.4 m
V
#
0.00053 m
3
/s, f 0.0218, V 2.98 m/s, and Re 44,550
→
1
2f
2.0 loga
1.5 10
6
m
3.7(0.015 m)
2.51
Re2f
b
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b
Re
VD
n
→ Re
V(0.015 m)
1.004 10
6
m
2
/s
V
V
#
A
c
V
#
pD
2
/4
→ V
V
#
p(0.015 m)
2
/4
h
L
af
L
D
a
K
L
b
V
2
2g
→ 18.4 af
11 m
0.015 m
24.7b
V
2
2(9.81 m/s
2
)
h
L
200,000 N/m
2
(998 kg/m
3
)(9.81 m/s
2
)
2 m 18.4 m
363
CHAPTER 8
cen72367_ch08.qxd 11/4/04 7:14 PM Page 363
Solving these 12 equations in 12 unknowns simultaneously using an equa-
tion solver, the flow rates are determined to be
Therefore, the flushing of the toilet
reduces the flow rate of cold water through
the shower by 21 percent
from 0.53 to 0.42 L/s, causing the shower water to
suddenly get very hot (Fig. 8–50).
Discussion If the velocity heads were considered, the flow rate through the
shower would be 0.43 instead of 0.42 L/s. Therefore, the assumption of
negligible velocity heads is reasonable in this case
Note that a leak in a piping system will cause the same effect, and thus
an unexplained drop in flow rate at an end point may signal a leak in the
system.
8–8
■
FLOW RATE AND VELOCITY MEASUREMENT
A major application area of fluid mechanics is the determination of the flow
rate of fluids, and numerous devices have been developed over the years for
the purpose of flow metering. Flowmeters range widely in their level of
sophistication, size, cost, accuracy, versatility, capacity, pressure drop, and
the operating principle. We give an overview of the meters commonly used
to measure the flow rate of liquids and gases flowing through pipes or
ducts. We limit our consideration to incompressible flow.
Some flowmeters measure the flow rate directly by discharging and
recharging a measuring chamber of known volume continuously and keep-
ing track of the number of discharges per unit time. But most flowmeters
measure the flow rate indirectly—they measure the average velocity V or a
quantity that is related to average velocity such as pressure and drag, and
determine the volume flow rate
V
.
from
(8–65)
where A
c
is the cross-sectional area of flow. Therefore, measuring the flow
rate is usually done by measuring flow velocity, and most flowmeters are
simply velocimeters used for the purpose of metering flow.
The velocity in a pipe varies from zero at the wall to a maximum at the
center, and it is important to keep this in mind when taking velocity mea-
surements. For laminar flow, for example, the average velocity is half the
centerline velocity. But this is not the case in turbulent flow, and it may be
necessary to take the weighted average of several local velocity measure-
ments to determine the average velocity.
V
#
VA
c
V
#
1
0.00090 m
3
/s,
V
#
2
0.00042 m
3
/s, and
V
#
3
0.00048 m
3
/s
1
2f
3
2.0 loga
1.5 10
6
m
3.7(0.015 m)
2.51
Re
3
2f
3
b
1
2f
2
2.0 loga
1.5 10
6
m
3.7(0.015 m)
2.51
Re
2
2f
2
b
1
2f
1
2.0 loga
1.5 10
6
m
3.7(0.015 m)
2.51
Re
1
2f
1
b
364
FLUID MECHANICS
FIGURE 8–50
Flow rate of cold water through a
shower may be affected significantly
by the flushing of a nearby toilet.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 364
The flow rate measurement techniques range from very crude to very ele-
gant. The flow rate of water through a garden hose, for example, can be
measured simply by collecting the water in a bucket of known volume and
dividing the amount collected by the collection time (Fig. 8–51). A crude
way of estimating the flow velocity of a river is to drop a float on the river
and measure the drift time between two specified locations. At the other
extreme, some flowmeters use the propagation of sound in flowing fluids
while others use the electromotive force generated when a fluid passes
through a magnetic field. In this section we discuss devices that are com-
monly used to measure velocity and flow rate, starting with the Pitot-static
probe introduced in Chap. 5.
Pitot and Pitot-Static Probes
Pitot probes (also called Pitot tubes) and Pitot-static probes, named after
the French engineer Henri de Pitot (1695–1771), are widely used for flow
rate measurement. A Pitot probe is just a tube with a pressure tap at the
stagnation point that measures stagnation pressure, while a Pitot-static probe
has both a stagnation pressure tap and several circumferential static pressure
taps and it measures both stagnation and static pressures (Figs. 8–52 and
8–53). Pitot was the first person to measure velocity with the upstream
pointed tube, while French engineer Henry Darcy (1803–1858) developed
most of the features of the instruments we use today, including the use of
small openings and the placement of the static tube on the same assembly.
Therefore, it is more appropriate to call the Pitot-static probes Pitot–Darcy
probes.
The Pitot-static probe measures local velocity by measuring the pressure
difference in conjunction with the Bernoulli equation. It consists of a slen-
der double-tube aligned with the flow and connected to a differential pres-
sure meter. The inner tube is fully open to flow at the nose, and thus it mea-
sures the stagnation pressure at that location (point 1). The outer tube is
sealed at the nose, but it has holes on the side of the outer wall (point 2) and
thus it measures the static pressure. For incompressible flow with suffi-
ciently high velocities (so that the frictional effects between points 1 and 2
are negligible), the Bernoulli equation is applicable and can be expressed as
(8–66)
P
1
rg
V
2
1
2g
z
1
P
2
rg
V
2
2
2g
z
2
365
CHAPTER 8
Nozzle
Bucket
Garden
hose
Stopwatch
FIGURE 8–51
A primitive (but fairly accurate) way
of measuring the flow rate of water
through a garden hose involves
collecting water in a bucket and
recording the collection time.
Stagnation
pressure
To stagnation pressure meter To stagnation pressure meter
To static pressure meter
Pitot-static probe
Pitot probe
(a)(b)
Stagnation
pressure
Static
pressure
VV
FIGURE 8–52
(a) A Pitot probe measures stagnation
pressure at the nose of the probe,
while (b) a Pitot-static probe measures
both stagnation pressure and static
pressure, from which the flow
speed can be calculated.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 365
Noting that z
1
≅ z
2
since the static pressure holes of the Pitot-static probe are
arranged circumferentially around the tube and V
1
0 because of the stag-
nation conditions, the flow velocity V V
2
becomes
Pitot formula: (8–67)
which is known as the Pitot formula. If the velocity is measured at a loca-
tion where the local velocity is equal to the average flow velocity, the vol-
ume flow rate can be determined from
V
.
VA
c
.
The Pitot-static probe is a simple, inexpensive, and highly reliable device
since it has no moving parts (Fig. 8–54). It also causes very small pressure
drop and usually does not disturb the flow appreciably. However, it is
important that it be properly aligned with the flow to avoid significant errors
that may be caused by misalignment. Also, the difference between the static
and stagnation pressures (which is the dynamic pressure) is proportional to
the density of the fluid and the square of the flow velocity. It can be used to
measure velocity in both liquids and gases. Noting that gases have low den-
sities, the flow velocity should be sufficiently high when the Pitot-static probe
is used for gas flow such that a measurable dynamic pressure develops.
Obstruction Flowmeters: Orifice, Venturi,
and Nozzle Meters
Consider incompressible steady flow of a fluid in a horizontal pipe of diam-
eter D that is constricted to a flow area of diameter d, as shown in Fig. 8–55.
The mass balance and the Bernoulli equations between a location before the
constriction (point 1) and the location where constriction occurs (point 2)
can be written as
Mass balance: (8–68)
Bernoulli equation (z
1
z
2
): (8–69)
Combining Eqs. 8–68 and 8–69 and solving for velocity V
2
gives
Obstruction (with no loss): (8–70)
where b d/D is the diameter ratio. Once V
2
is known, the flow rate can be
determined from
V
.
A
2
V
2
(pd
2
/4)V
2
.
This simple analysis shows that the flow rate through a pipe can be deter-
mined by constricting the flow and measuring the decrease in pressure due
to the increase in velocity at the constriction site. Noting that the pressure
drop between two points along the flow can be measured easily by a differ-
ential pressure transducer or manometer, it appears that a simple flow rate
measurement device can be built by obstructing the flow. Flowmeters based
on this principle are called obstruction flowmeters and are widely used to
measure flow rates of gases and liquids.
The velocity in Eq. 8–70 is obtained by assuming no loss, and thus it is
the maximum velocity that can occur at the constriction site. In reality,
some pressure losses due to frictional effects are inevitable, and thus the
velocity will be less. Also, the fluid stream will continue to contract past the
V
2
B
2(P
1
P
2
)
r(1 b
4
)
P
1
rg
V
2
1
2g
P
2
rg
V
2
2
2g
V
#
A
1
V
1
A
2
V
2
→ V
1
(A
2
/A
1
)V
2
(d/D)
2
V
2
V
B
2(P
1
P
2
)
r
366
FLUID MECHANICS
Wind tunnel wall
Flexible
tubing
Differential pressure transducer
or manometer to measure P
1
– P
2
P
1
– P
2
Flow
Pitot-static probe
Stagnation
pressure, P
1
Static
pressure, P
2
FIGURE 8–53
Measuring flow velocity with a Pitot-
static probe. (A manometer may also
be used in place of the differential
pressure transducer.)
FIGURE 8–54
Close-up of a Pitot-static probe,
showing the stagnation pressure hole
and two of the five static
circumferential pressure holes.
Photo by Po-Ya Abel Chuang.
1 2
Dd
Obstruction
FIGURE 8–55
Flow through a constriction in a pipe.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 366
obstruction, and the vena contracta area is less than the flow area of the
obstruction. Both losses can be accounted for by incorporating a correction
factor called the discharge coefficient C
d
whose value (which is less than
1) is determined experimentally. Then the flow rate for obstruction flowme-
ters can be expressed as
Obstruction flowmeters: (8–71)
where A
0
A
2
pd
2
/4 is the cross-sectional area of the hole and b d/D
is the ratio of hole diameter to pipe diameter. The value of C
d
depends on
both b and the Reynolds number Re V
1
D/n, and charts and curve-fit cor-
relations for C
d
are available for various types of obstruction meters.
Of the numerous types of obstruction meters available, those most widely
used are orifice meters, flow nozzles, and Venturi meters (Fig. 8–56). The
experimentally determined data for discharge coefficients are expressed as
(Miller, 1997)
Orifice meters: (8–72)
Nozzle meters: (8–73)
These relations are valid for 0.25 b 0.75 and 10
4
Re 10
7
. Precise
values of C
d
depend on the particular design of the obstruction, and thus the
manufacturer’s data should be consulted when available. For flows with
high Reynolds numbers (Re 30,000), the value of C
d
can be taken to be
0.96 for flow nozzles and 0.61 for orifices.
Owing to its streamlined design, the discharge coefficients of Venturi
meters are very high, ranging between 0.95 and 0.99 (the higher values are
for the higher Reynolds numbers) for most flows. In the absence of specific
data, we can take C
d
0.98 for Venturi meters. Also, the Reynolds number
depends on the flow velocity, which is not known a priori. Therefore, the
solution is iterative in nature when curve-fit correlations are used for C
d
.
The orifice meter has the simplest design and it occupies minimal space
as it consists of a plate with a hole in the middle, but there are considerable
variations in design (Fig. 8–57). Some orifice meters are sharp-edged, while
C
d
0.9975
6.53b
0.5
Re
0.5
C
d
0.5959 0.0312b
2.1
0.184b
8
91.71b
2.5
Re
0.75
V
#
A
0
C
d
B
2(P
1
P
2
)
r(1 b
4
)
367
CHAPTER 8
D
(c) Venturi meter
D
d
d
(b) Flow nozzle
21°
15°
(a) Orifice meter
Dd
FIGURE 8–56
Common types of obstruction meters.
Flow
Housing
Magnet
Bellows
Orifice
P
1
V
1
V
2
> V
1
P
1
> P
2
P
2
V
2
Force
FIGURE 8–57
An orifice meter and schematic
showing its built-in pressure
transducer and digital readout.
Courtesy KOBOLD Instruments, Pittsburgh, PA.
www.koboldusa.com. Used by permission.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 367
others are beveled or rounded. The sudden change in the flow area in orifice
meters causes considerable swirl and thus significant head loss or perma-
nent pressure loss, as shown in Fig. 8–58. In nozzle meters, the plate is
replaced by a nozzle, and thus the flow in the nozzle is streamlined. As a
result, the vena contracta is practically eliminated and the head loss is small.
However, flow nozzle meters are more expensive than orifice meters.
The Venturi meter, invented by the American engineer Clemans Herschel
(1842–1930) and named by him after the Italian Giovanni Venturi (1746–
1822) for his pioneering work on conical flow sections, is the most accurate
flowmeter in this group, but it is also the most expensive. Its gradual con-
traction and expansion prevent flow separation and swirling, and it suffers
only frictional losses on the inner wall surfaces. Venturi meters cause very
low head losses, as shown in Fig. 8–59, and thus they should be preferred
for applications that cannot allow large pressure drops. The irreversible head
loss for Venturi meters due to friction is only about 10 percent.
EXAMPLE 8–10 Measuring Flow Rate with an Orifice Meter
The flow rate of methanol at 20°C (r 788.4 kg/m
3
and m 5.857
10
4
kg/m · s) through a 4-cm-diameter pipe is to be measured with a
3-cm-diameter orifice meter equipped with a mercury manometer across the
orifice place, as shown in Fig. 8–60. If the differential height of the
manometer is read to be 11 cm, determine the flow rate of methanol
through the pipe and the average flow velocity.
SOLUTION The flow rate of methanol is to be measured with an orifice
meter. For a given pressure drop across the orifice plate, the flow rate and
the average flow velocity are to be determined.
Assumptions 1 The flow is steady and incompressible. 2 The discharge coef-
ficient of the orifice meter is
C
d
0.61.
368
FLUID MECHANICS
Pressure drop
across orifice
HGL
Lost pressure
Orifice
meter
P
1
P
2
P
3
Recovered
pressure
FIGURE 8–58
The variation of pressure along a flow
section with an orifice meter as
measured with piezometer tubes; the
lost pressure and the pressure recovery
are shown.
Orifice with
flange taps
Short cone Venturi
Long cone Venturi
Lo-loss tube
Flow
nozzle
Fraction of pressure loss, %
d/D ratio, b
0
10
20
30
40
50
60
70
80
90
1.00
0 0.10 0.20 0.30 0.40 0.50 0.700.60 0.80 0.90 1.00
FIGURE 8–59
The fraction of pressure (or head) loss
for various obstruction meters.
From ASME Fluid Meters. Used by permission of
ASME International.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 368
Properties The density and dynamic viscosity of methanol are given to be
r 788.4 kg/m
3
and m 5.857 10
4
kg/m · s, respectively. We take the
density of mercury to be 13,600 kg/m
3
.
Analysis The diameter ratio and the throat area of the orifice are
The pressure drop across the orifice plate can be expressed as
Then the flow rate relation for obstruction meters becomes
Substituting, the flow rate is determined to be
which is equivalent to 3.09 L/s. The average flow velocity in the pipe is
determined by dividing the flow rate by the cross-sectional area of the pipe,
Discussion The Reynolds number of flow through the pipe is
Substituting b 0.75 and Re 1.32 10
5
into the orifice discharge coef-
ficient relation
gives
C
d
0.601, which is very close to the assumed value of 0.61. Using
this refined value of
C
d
, the flow rate becomes 3.04 L/s, which differs from
our original result by only 1.6 percent. Therefore, it is convenient to analyze
orifice meters using the recommended value of
C
d
0.61 for the discharge
coefficient, and then to verify the assumed value. If the problem is solved
using an equation solver such as EES, then the problem can be formulated
using the curve-fit formula for
C
d
(which depends on the Reynolds number),
and all equations can be solved simultaneously by letting the equation solver
perform the iterations as necessary.
Positive Displacement Flowmeters
When we buy gasoline for the car, we are interested in the total amount of
gasoline that flows through the nozzle during the period we fill the tank
rather than the flow rate of gasoline. Likewise, we care about the total
C
d
0.5959 0.0312b
2.1
0.184b
8
91.71b
2.5
Re
0.75
Re
rVD
m
(788.4 kg/m
3
)(2.46 m/s)(0.04 m)
5.857 10
4
kg/m s
1.32 10
5
V
V
#
A
c
V
#
pD
2
/4
3.09 10
3
m
3
/s
p(0.04 m)
2
/4
2.46 m/s
3.09 10
3
m
3
/s
V
#
(7.069 10
4
m
2
)(0.61)
B
2(13,600/788.4 1)(9.81 m/s
2
)(0.11 m)
1 0.75
4
V
#
A
0
C
d
B
2(P
1
P
2
)
r(1 b
4
)
A
0
C
d
B
2(r
Hg
r
met
)gh
r
met
(1 b
4
)
A
0
C
d
B
2(r
Hg
/r
met
1)gh
1 b
4
P P
1
P
2
(r
Hg
r
met
)gh
A
0
pd
2
4
p(0.03 m)
2
4
7.069 10
4
m
2
b
d
D
3
4
0.75
369
CHAPTER 8
1 2
11 cm
Mercury
manometer
FIGURE 8–60
Schematic for the orifice meter
considered in Example 8–10.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 369
amount of water or natural gas we use in our homes during a billing period.
In these and many other applications, the quantity of interest is the total
amount of mass or volume of a fluid that passes through a cross section of a
pipe over a certain period of time rather than the instantaneous value of
flow rate, and positive displacement flowmeters are well suited for such
applications. There are numerous types of displacement meters, and they are
based on continuous filling and discharging of the measuring chamber. They
operate by trapping a certain amount of incoming fluid, displacing it to the
discharge side of the meter, and counting the number of such discharge–
recharge cycles to determine the total amount of fluid displaced. The clear-
ance between the impeller and its casing must be controlled carefully to pre-
vent leakage and thus to avoid error.
Figure 8–61 shows a positive displacement flowmeter with two rotating
impellers driven by the flowing liquid. Each impeller has three gear lobes,
and a pulsed output signal is generated each time a lobe passes by a nonin-
trusive sensor. Each pulse represents a known volume of liquid that is cap-
tured in between the lobes of the impellers, and an electronic controller con-
verts the pulses to volume units. This particular meter has a quoted accuracy
of 0.1 percent, has a low pressure drop, and can be used with high- or low-
viscosity liquids at temperatures up to 230°C and pressures up to 7 MPa for
flow rates of up to 700 gal/min (or 50 L/s).
The most widely used flowmeters to measure liquid volumes are nutating
disk flowmeters, shown in Fig. 8–62. They are commonly used as water
and gasoline meters. The liquid enters the nutating disk meter through the
chamber (A). This causes the disk (B) to nutate or wobble and results in the
rotation of a spindle (C) and the excitation of a magnet (D). This signal is
transmitted through the casing of the meter to a second magnet (E). The
total volume is obtained by counting the number of these signals during a
discharge process.
Quantities of gas flows, such as the amount of natural gas used in build-
ings, are commonly metered by using bellows flowmeters that displace a
certain amount of gas volume during each revolution.
Turbine Flowmeters
We all know from experience that a propeller held against the wind rotates,
and the rate of rotation increases as the wind velocity increases. You may
370
FLUID MECHANICS
FIGURE 8–61
A positive displacement flowmeter
with double helical three-lobe
impeller design.
Courtesy Flow Technology, Inc.
Source: www.ftimeters.com.
A
B
D
C
E
FIGURE 8–62
A nutating disk flowmeter.
(a) Courtesy Badger Meter, Inc.
Source: www.badgermeter.com.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 370
also have seen that the turbine blades of wind turbines rotate rather slowly
at low winds, but quite fast at high winds. These observations suggest that
the flow velocity in a pipe can be measured by placing a freely rotating
propeller inside a pipe section and doing the necessary calibration. Flow
measurement devices that work on this principle are called turbine flowme-
ters or sometimes propeller flowmeters, although the latter is a misnomer
since, by definition, propellers add energy to a fluid, while turbines extract
energy from a fluid.
A turbine flowmeter consists of a cylindrical flow section that houses a
turbine (a vaned rotor) that is free to rotate, additional stationary vanes at
the inlet to straighten the flow, and a sensor that generates a pulse each time
a marked point on the turbine passes by to determine the rate of rotation.
The rotational speed of the turbine is nearly proportional to the flow rate of
the fluid. Turbine flowmeters give highly accurate results (as accurate as
0.25 percent) over a wide range of flow rates when calibrated properly for
the anticipated flow conditions. Turbine flowmeters have very few blades
(sometimes just two blades) when used to measure liquid flow, but several
blades when used to measure gas flow to ensure adequate torque generation.
The head loss caused by the turbine is very small.
Turbine flowmeters have been used extensively for flow measurement
since the 1940s because of their simplicity, low cost, and accuracy over a
wide range of flow conditions. They are made commercially available for
both liquids and gases and for pipes of practically all sizes. Turbine flowme-
ters are also commonly used to measure flow velocities in unconfined flows
such as winds, rivers, and ocean currents. The handheld device shown in
Fig. 8–63b is used to measure wind velocity.
Paddlewheel Flowmeters
Paddlewheel flowmeters are low-cost alternatives to turbine flowmeters for
flows where very high accuracy is not required. In paddlewheel flowmeters,
the paddlewheel (the rotor and the blades) is perpendicular to the flow, as
shown in Fig. 8–64, rather than parallel as was the case with turbine
371
CHAPTER 8
FIGURE 8–63
(a) An in-line turbine flowmeter
to measure liquid flow, with flow from
left to right, (b) a close-up view of the
turbine blades inside the flowmeter,
looking down the axis with flow into
the page, and (c) a handheld turbine
flowmeter to measure wind speed,
measuring no flow so that the turbine
blades are visible. the flowmeter in (c)
also measures the air termperature for
convenience.
Photos by John M. Cimbala.
(a)(b)
(C)
cen72367_ch08.qxd 11/4/04 7:34 PM Page 371
flowmeters. The paddles cover only a portion of the flow cross section (typ-
ically, less than half), and thus the head loss is much smaller compared to
that of turbine flowmeters, but the depth of insertion of the paddlewheel
into the flow is of critical importance for accuracy. Also, no strainers are
required since the paddlewheels are not susceptible to fouling. A sensor
detects the passage of each of the paddlewheel blades and transmits a sig-
nal. A microprocessor then converts this rotational speed information to
flow rate or integrated flow quantity.
Variable-Area Flowmeters (Rotameters)
A simple, reliable, inexpensive, and easy-to-install flowmeter with low pres-
sure drop and no electrical connections that gives a direct reading of flow
rate for a wide range of liquids and gases is the variable-area flowmeter,
also called a rotameter or floatmeter. A variable-area flowmeter consists of
a vertical tapered conical transparent tube made of glass or plastic with a
float inside that is free to move, as shown in Fig. 8–65. As fluid flows
through the tapered tube, the float rises within the tube to a location where
the float weight, drag force, and buoyancy force balance each other and the
net force acting on the float is zero. The flow rate is determined by simply
matching the position of the float against the graduated flow scale outside
the tapered transparent tube.
We know from experience that high winds knock down trees, break power
lines, and blow away hats or umbrellas. This is because the drag force
increases with flow velocity. The weight and the buoyancy force acting on
the float are constant, but the drag force changes with flow velocity. Also,
the velocity along the tapered tube decreases in the flow direction because
of the increase in the cross-sectional area. There is a certain velocity that
generates enough drag to balance the float weight and the buoyancy force,
and the location at which this velocity occurs around the float is the location
where the float settles. The degree of tapering of the tube can be made such
that the vertical rise changes linearly with flow rate, and thus the tube can
be calibrated linearly for flow rates. The transparent tube also allows the
fluid to be seen during flow.
There are numerous kinds of variable-area flowmeters. The gravity-based
flowmeter discussed previously must be positioned vertically, with fluid
entering from the bottom and leaving from the top. In spring-opposed
372
FLUID MECHANICS
Paddlewheel
sensor
Sensor
housing
Flow
Truseal
locknut
Retainer cap
FIGURE 8–64
Paddlewheel flowmeter to measure
liquid flow, with flow from left to
right, and a schematic diagram of
its operation.
Photo by John M. Cimbala.
FIGURE 8–65
Two types of variable-area
flowmeters: (a) an ordinary
gravity-based meter and
(b) a spring-opposed meter.
(a) Photo by John M. Cimbala and (b) Courtesy
Insite, Universal Flow Monitors, Inc. Used by
permission. Sources: www.omega.com and
www.flowmeters.com.
(a)
(b)
cen72367_ch08.qxd 11/4/04 7:14 PM Page 372
flowmeters, the drag force is balanced by the spring force, and such
flowmeters can be installed horizontally. Another type of flowmeter uses a
loose-fitting piston instead of a float.
The accuracy of variable-area flowmeters is typically 5 percent. There-
fore, these flowmeters are not appropriate for applications that require preci-
sion measurements. However, some manufacturers quote accuracies of the
order of 1 percent. Also, these meters depend on visual checking of the loca-
tion of the float, and thus they cannot be used to measure the flow rate of
fluids that are opaque or dirty, or fluids that coat the float since such fluids
block visual access. Finally, glass tubes are prone to breakage and thus they
pose a safety hazard if toxic fluids are handled. In such applications, vari-
able-area flowmeters should be installed at locations with minimum traffic.
Ultrasonic Flowmeters
It is a common observation that when a stone is dropped into calm water,
the waves that are generated spread out as concentric circles uniformly in all
directions. But when a stone is thrown into flowing water such as a river,
the waves move much faster in the flow direction (the wave and flow veloc-
ities are added since they are in the same direction) compared to the waves
moving in the upstream direction (the wave and flow velocities are sub-
tracted since they are in opposite directions). As a result, the waves appear
spread out downstream while they appear tightly packed upstream. The dif-
ference between the number of waves in the upstream and downstream parts
of the flow per unit length is proportional to the flow velocity, and this sug-
gests that flow velocity can be measured by comparing the propagation of
waves in the forward and backward directions to flow. Ultrasonic flowme-
ters operate on this principle, using sound waves in the ultrasonic range
(typically at a frequency of 1 MHz).
Ultrasonic (or acoustic) flowmeters operate by generating sound waves
with a transducer and measuring the propagation of those waves through a
flowing fluid. There are two basic kinds of ultrasonic flowmeters: transit
time and Doppler-effect (or frequency shift) flowmeters. The transit time
flowmeter transmits sound waves in the upstream and downstream direc-
tions and measures the difference in travel time. A typical transit time ultra-
sonic meter is shown schematically in Fig. 8–66. It involves two transducers
that alternately transmit and receive ultrasonic waves, one in the direction of
flow and the other in the opposite direction. The travel time for each direc-
tion can be measured accurately, and the difference in the travel time can be
calculated. The average flow velocity V in the pipe is proportional to this
travel time difference t, and can be determined from
(8–74)
where L is the distance between the transducers and K is a constant.
Doppler-Effect Ultrasonic Flowmeters
You have probably noticed that when a fast-moving car approaches with its
horn blowing, the tone of the high-pitched sound of the horn drops to a
lower pitch as the car passes by. This is due to the sonic waves being com-
pressed in front of the car and being spread out behind it. This shift in fre-
quency is called the Doppler effect, and it forms the basis for the operation
of most ultrasonic flowmeters.
V KL t
373
CHAPTER 8
Flow
A Reflect-mode
configuration
B
Top view
FIGURE 8–66
The operation of a transit time
ultrasonic flowmeter equipped with
two transducers, www.flocat.com.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 373
Doppler-effect ultrasonic flowmeters measure the average flow velocity
along the sonic path. This is done by clamping a piezoelectric transducer on
the outside surface of a pipe (or pressing the transducer against the pipe for
handheld units). The transducer transmits a sound wave at a fixed frequency
through the pipe wall and into the flowing liquid. The waves reflected by
impurities, such as suspended solid particles or entrained gas bubbles, are
relayed to a receiving transducer. The change in the frequency of the reflected
waves is proportional to the flow velocity, and a microprocessor determines
the flow velocity by comparing the frequency shift between the transmitted
and reflected signals (Figs. 8–67 and 8–68). The flow rate and the total
amount of flow can also be determined using the measured velocity by prop-
erly configuring the flowmeter for the given pipe and flow conditions.
The operation of ultrasonic flowmeters depends on the ultrasound waves
being reflected off discontinuities in density. Ordinary ultrasonic flowmeters
require the liquid to contain impurities in concentrations greater than 25
parts per million (ppm) in sizes greater than at least 30 m. But advanced
ultrasonic units can also measure the velocity of clean liquids by sensing the
waves reflected off turbulent swirls and eddies in the flow stream, provided
that they are installed at locations where such disturbances are nonsymmet-
rical and at a high level, such as a flow section just downstream of a 90°
elbow.
Ultrasonic flowmeters have the following advantages:
• They are easy and quick to install by clamping them on the outside of
pipes of 0.6 cm to over 3 m in diameter, and even on open channels.
• They are nonintrusive. Since the meters clamp on, there is no need to stop
operation and drill holes into piping, and no production downtime.
• There is no pressure drop since the meters do not interfere with the flow.
374
FLUID MECHANICS
Transmitting
element
Receiving
element
Flow
direction
Reflectors
FIGURE 8–67
The operation of a Doppler-effect
ultrasonic flowmeter equipped
with a transducer pressed
on the outer surface of a pipe.
FIGURE 8–68
Ultrasonic flowmeters enable one
to measure flow velocity without
even contacting the fluid by simply
pressing a transducer on the outer
surface of the pipe.
Photo by John M. Cimbala.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 374
• Since there is no direct contact with the fluid, there is no danger of
corrosion or clogging.
• They are suitable for a wide range of fluids from toxic chemicals to
slurries to clean liquids, for permanent or temporary flow measurement.
• There are no moving parts, and thus the meters provide reliable and
maintenance-free operation.
• They can also measure flow quantities in reverse flow.
• The quoted accuracies are 1 to 2 percent.
Ultrasonic flowmeters are noninvasive devices, and the ultrasonic transduc-
ers can effectively transmit signals through polyvinyl chloride (PVC), steel,
iron, and glass pipe walls. However, coated pipes and concrete pipes are not
suitable for this measurement technique since they absorb ultrasonic waves.
Electromagnetic Flowmeters
It has been known since Faraday’s experiments in the 1830s that when a
conductor is moved in a magnetic field, an electromotive force develops
across that conductor as a result of magnetic induction. Faraday’s law states
that the voltage induced across any conductor as it moves at right angles
through a magnetic field is proportional to the velocity of that conductor.
This suggests that we may be able to determine flow velocity by replacing
the solid conductor by a conducting fluid, and electromagnetic flowmeters
do just that. Electromagnetic flowmeters have been in use since the mid-
1950s, and they come in various designs such as full-flow and insertion
types.
A full-flow electromagnetic flowmeter is a nonintrusive device that con-
sists of a magnetic coil that encircles the pipe, and two electrodes drilled
into the pipe along a diameter flush with the inner surface of the pipe so
that the electrodes are in contact with the fluid but do not interfere with the
flow and thus do not cause any head loss (Fig. 8–69a). The electrodes are
connected to a voltmeter. The coils generate a magnetic field when sub-
jected to electric current, and the voltmeter measures the electric potential
375
CHAPTER 8
(a) Full-flow electromagnetic flowmeter (b) Insertion electromagnetic flowmeter
Electrodes
Flow
Flow
E
E
Flow
FIGURE 8–69
(a) Full-flow and (b) insertion
electromagnetic flowmeters,
www.flocat.com.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 375
difference between the electrodes. This potential difference is proportional
to the flow velocity of the conducting fluid, and thus the flow velocity can
be calculated by relating it to the voltage generated.
Insertion electromagnetic flowmeters operate similarly, but the magnetic
field is confined within a flow channel at the tip of a rod inserted into the
flow, as shown in Fig. 8–69b.
Electromagnetic flowmeters are well-suited for measuring flow velocities
of liquid metals such as mercury, sodium, and potassium that are used in
some nuclear reactors. They can also be used for liquids that are poor con-
ductors, such as water, provided that they contain an adequate amount of
charged particles. Blood and seawater, for example, contain sufficient
amounts of ions, and thus electromagnetic flowmeters can be used to mea-
sure their flow rates. Electromagnetic flowmeters can also be used to mea-
sure the flow rates of chemicals, pharmaceuticals, cosmetics, corrosive liq-
uids, beverages, fertilizers, and numerous slurries and sludges, provided that
the substances have high enough electrical conductivities. Electromagnetic
flowmeters are not suitable for use with distilled or deionized water.
Electromagnetic flowmeters measure flow velocity indirectly, and thus
careful calibration is important during installation. Their use is limited by
their relatively high cost, power consumption, and the restrictions on the
types of suitable fluids with which they can be used.
Vortex Flowmeters
You have probably noticed that when a flow stream such as a river encoun-
ters an obstruction such as a rock, the fluid separates and moves around the
rock. But the presence of the rock is felt for some distance downstream via
the swirls generated by it.
Most flows encountered in practice are turbulent, and a disk or a short
cylinder placed in the flow coaxially sheds vortices (see also Chap. 4). It is
observed that these vortices are shed periodically, and the shedding fre-
quency is proportional to the average flow velocity. This suggests that the
flow rate can be determined by generating vortices in the flow by placing an
obstruction along the flow and measuring the shedding frequency. The flow
measurement devices that work on this principle are called vortex flowme-
ters. The Strouhal number, defined as St fd/V, where f is the vortex shed-
ding frequency, d is the characteristic diameter or width of the obstruction,
and V is the velocity of the flow impinging on the obstruction, also remains
constant in this case, provided that the flow velocity is high enough.
A vortex flowmeter consists of a sharp-edged bluff body (strut) placed in
the flow that serves as the vortex generator, and a detector (such as a pres-
sure transducer that records the oscillation in pressure) placed a short dis-
tance downstream on the inner surface of the casing to measure the shed-
ding frequency. The detector can be an ultrasonic, electronic, or fiber-optic
sensor that monitors the changes in the vortex pattern and transmits a pul-
sating output signal (Fig. 8–70). A microprocessor then uses the frequency
information to calculate and display the flow velocity or flow rate. The fre-
quency of vortex shedding is proportional to the average velocity over a
wide range of Reynolds numbers, and vortex flowmeters operate reliably
and accurately at Reynolds numbers from 10
4
to 10
7
.
376
FLUID MECHANICS
Bluff
body (strut)
Vortex
swirl
Transmitting
transducer
Receiving
transducer
Flow
FIGURE 8–70
The operation of a vortex flowmeter,
www.flocat.com.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 376
The vortex flowmeter has the advantage that it has no moving parts and
thus is inherently reliable, versatile, and very accurate (usually 1 percent
over a wide range of flow rates), but it obstructs flow and thus causes con-
siderable head loss.
Thermal (Hot-Wire and Hot-Film) Anemometers
Thermal anemometers were introduced in the late 1950s and have been in
common use since then in fluid research facilities and labs. As the name
implies, thermal anemometers involve an electrically heated sensor, as
shown in Fig. 8–71, and utilize a thermal effect to measure flow velocity.
Thermal anemometers have extremely small sensors, and thus they can be
used to measure the instantaneous velocity at any point in the flow without
appreciably disturbing the flow. They can take thousands of velocity mea-
surements per second with excellent spatial and temporal resolution, and
thus they can be used to study the details of fluctuations in turbulent flow.
They can measure velocities in liquids and gases accurately over a wide
range—from a few centimeters to over a hundred meters per second.
A thermal anemometer is called a
hot-wire anemometer if the sensing
element is a wire, and a
hot-film anemometer if the sensor is a thin metal-
lic film (less than 0.1 m thick) mounted usually on a relatively thick
ceramic support having a diameter of about 50 m. The hot-wire anemome-
ter is characterized by its very small sensor wire—usually a few microns in
diameter and a couple of millimeters in length. The sensor is usually made
of platinum, tungsten, or platinum–iridium alloys, and it is attached to the
probe through holders. The fine wire sensor of a hot-wire anemometer is
very fragile because of its small size and can easily break if the liquid or
gas contains excessive amounts of contaminants or particulate matter. This
is especially of consequence at high velocities. In such cases, the more
rugged hot-film probes should be used. But the sensor of the hot-film probe
is larger, has significantly lower frequency response, and interferes more
with the flow; thus it is not always suitable for studying the fine details of
turbulent flow.
The operating principle of a constant-temperature anemometer (CTA),
which is the most common type and is shown schematically in Fig. 8–72, is
as follows: the sensor is electrically heated to a specified temperature (typi-
cally about 200°C). The sensor tends to cool as it loses heat to the surround-
ing flowing fluid, but electronic controls maintain the sensor at a constant
377
CHAPTER 8
Electric
current I
Sensor (a thin wire
approximately1 mm long with
a diameter of 5 mm)
Wire
support
Flow
velocity V
FIGURE 8–71
The electrically heated sensor and its
support of a hot-wire probe.
Bridge
Connector box
and computer
CTA
Signal
conditioner
Sensor
Flow
Filter Gain
Servo
loop
Probe
FIGURE 8–72
Schematic of a thermal
anemometer system.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 377
temperature by varying the electric current (which is done by varying the
voltage) as needed. The higher the flow velocity, the higher the rate of heat
transfer from the sensor, and thus the larger the voltage that needs to be
applied across the sensor to maintain it at constant temperature. There is a
close correlation between the flow velocity and voltage, and the flow veloc-
ity can be determined by measuring the voltage applied by an amplifier or
the electric current passing through the sensor.
The sensor is maintained at a constant temperature during operation, and
thus its thermal energy content remains constant. The conservation of
energy principle requires that the electrical Joule heating W
.
elect
I
2
R
w
E
2
/R
w
of the sensor must be equal to the total rate of heat loss from the
sensor Q
.
total
, which consists of convection heat transfer since conduction to
the wire supports and radiation to the surrounding surfaces are small and
can be disregarded. Using proper relations for forced convection, the energy
balance can be expressed by King’s law as
(8–75)
where E is the voltage, and the values of the constants a, b,and n are cali-
brated for a given probe. Once the voltage is measured, this relation gives
the flow velocity V directly.
Most hot-wire sensors have a diameter of 5 mm and a length of approxi-
mately 1 mm and are made of tungsten. The wire is spot-welded to needle-
shaped prongs embedded in a probe body, which is connected to the
anemometer electronics. Thermal anemometers can be used to measure two-
or three-dimensional velocity components simultaneously by using probes
with two or three sensors, respectively (Fig. 8–73). When selecting probes,
consideration should be given to the type and the contamination level of the
fluid, the number of velocity components to be measured, the required spa-
tial and temporal resolution, and the location of measurement.
Laser Doppler Velocimetry
Laser Doppler velocimetry (LDV),
also called laser velocimetry (LV) or
laser Doppler anemometry (LDA), is an optical technique to measure flow
velocity at any desired point without disturbing the flow. Unlike thermal
anemometry, LDV involves no probes or wires inserted into the flow, and
thus it is a nonintrusive method. Like thermal anemometry, it can accurately
measure velocity at a very small volume, and thus it can also be used to
study the details of flow at a locality, including turbulent fluctuations, and it
can be traversed through the entire flow field without intrusion.
The LDV technique was developed in the mid-1960s and has found wide-
spread acceptance because of the high accuracy it provides for both gas and
E
2
a bV
n
378
FLUID MECHANICS
(a)(c)(b)
FIGURE 8–73
Thermal anemometer probes
with single, double, and triple
sensors to measure (a) one-, (b) two-,
and (c) three-dimensional velocity
components simultaneously.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 378
liquid flows; the high spatial resolution it offers; and, in recent years, its
ability to measure all three velocity components. Its drawbacks are the rela-
tively high cost; the requirement for sufficient transparency between the
laser source, the target location in the flow, and the photodetector; and the
requirement for careful alignment of emitted and reflected beams for accu-
racy. The latter drawback is eliminated for the case of a fiber-optic LDV
system, since it is aligned at the factory.
The operating principle of LDV is based on sending a highly coherent
monochromatic (all waves are in phase and at the same wavelength) light
beam toward the target, collecting the light reflected by small particles in
the target area, determining the change in frequency of the reflected radia-
tion due to the Doppler effect, and relating this frequency shift to the flow
velocity of the fluid at the target area.
LDV systems are available in many different configurations. A basic dual-
beam LDV system to measure a single velocity component is shown in Fig.
8–74. In the heart of all LDV systems is a laser power source, which is usu-
ally a helium–neon or argon-ion laser with a power output of 10 mW to 20
W. Lasers are preferred over other light sources since laser beams are highly
coherent and highly focused. The helium–neon laser, for example, emits
radiation at a wavelength of 0.6328 m, which is in the reddish-orange
color range. The laser beam is first split into two parallel beams of equal
intensity by a half-silvered mirror called a beam splitter. Both beams then
pass through a converging lens that focuses the beams at a point in the flow
(the target). The small fluid volume where the two beams intersect is the
region where the velocity is measured and is called the measurement volume
or the focal volume. The measurement volume resembles an ellipsoid, typi-
cally of 0.1 mm diameter and 0.5 mm in length. The laser light is scattered
by particles passing through this measurement volume, and the light scat-
tered in a certain direction is collected by a receiving lens and is passed
through a photodetector that converts the fluctuations in light intensity into
fluctuations in a voltage signal. Finally, a signal processor determines the
frequency of the voltage signal and thus the velocity of the flow.
The waves of the two laser beams that cross in the measurement volume
are shown schematically in Fig. 8–75. The waves of the two beams interfere
in the measurement volume, creating a bright fringe where they are in phase
and thus support each other, and creating a dark fringe where they are out of
phase and thus cancel each other. The bright and dark fringes form lines
379
CHAPTER 8
Beam splitter
Mirror
Laser
Bragg cell
Measurement
volume
Receiving
lens
Photodetector
Sending lens
V
a
FIGURE 8–74
A dual-beam LDV system in forward
scatter mode.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 379
parallel to the midplane between the two incident laser beams. Using
trigonometry, the spacing s between the fringe lines, which can be viewed
as the wavelength of fringes, can be shown to be s l/[2 sin(a/2)], where
l is the wavelength of the laser beam and a is the angle between the two
laser beams. When a particle traverses these fringe lines at velocity V, the
frequency of the scattered fringe lines is
(8–76)
This fundamental relation shows the flow velocity to be proportional to the
frequency and is known as the LDV equation. As a particle passes through
the measurement volume, the reflected light is bright, then dark, then bright,
etc., because of the fringe pattern, and the flow velocity is determined by
measuring the frequency of the reflected light. The velocity profile at a
cross section of a pipe can be obtained by mapping the flow across the pipe
(Fig. 8–76).
The LDV method obviously depends on the presence of scattered fringe
lines, and thus the flow must contain a sufficient amount of small particles
called seeds or seeding particles. These particles must be small enough to
follow the flow closely so that the particle velocity is equal to the flow
velocity, but large enough (relative to the wavelength of the laser light) to
scatter an adequate amount of light. Particles with a diameter of 1 m usu-
ally serve the purpose well. Some fluids such as tap water naturally contain
an adequate amount of such particles, and no seeding is necessary. Gases
such as air are commonly seeded with smoke or with particles made of
latex, oil, or other materials. By using three laser beam pairs at different
wavelengths, the LDV system is also used to obtain all three velocity com-
ponents at any point in the flow.
Particle Image Velocimetry
Particle image velocimetry (PIV)
is a double-pulsed laser technique used
to measure the instantaneous velocity distribution in a plane of flow by pho-
tographically determining the displacement of particles in the plane during a
very short time interval. Unlike methods like hot-wire anemometry and
LDV that measure velocity at a point, PIV provides velocity values simulta-
neously throughout an entire cross section, and thus it is a whole-field tech-
nique. PIV combines the accuracy of LDV with the capability of flow visu-
alization and provides instantaneous flow field mapping. The entire
instantaneous velocity profile at a cross section of pipe, for example, can be
obtained with a single PIV measurement. A PIV system can be viewed as a
camera that can take a snapshot of velocity distribution at any desired plane
in a flow. Ordinary flow visualization gives a qualitative picture of the
details of flow. PIV also provides an accurate quantitative description of
various flow quantities such as the velocity field, and thus the capability to
analyze the flow numerically using the velocity data provided. Because of
its whole-field capability, PIV is also used to validate computational fluid
dynamics (CFD) codes (Chap. 15).
The PIV technique has been used since the mid-1980s, and its use and
capabilities have grown in recent years with improvements in frame grabber
f
V
s
2V sin(a/2)
l
380
FLUID MECHANICS
Fringe
lines
V
Laser
beams
Measurement
volume
Fringe
lines
s
a
l
FIGURE 8–75
Fringes that form as a result of the
interference at the intersection of two
laser beams of an LDV system (lines
represent peaks of waves). The top
diagram is a close-up view of two
fringes.
5
4
3
2
1
–80
(m/s)
–60 –40
x (mm)
–20 0
FIGURE 8–76
A time-averaged velocity profile in
turbulent pipe flow obtained by an
LDV system.
Courtesy Dantec Dynamics, Inc.
www.dantecmt.com. Used by permission.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 380
and charge-coupled device (CCD) camera technologies. The accuracy, flexi-
bility, and versatility of PIV systems with their ability to capture whole-field
images with submicrosecond exposure time have made them extremely
valuable tools in the study of supersonic flows, explosions, flame propaga-
tion, bubble growth and collapse, turbulence, and unsteady flow.
The PIV technique for velocity measurement consists of two main steps:
visualization and image processing. The first step is to seed the flow with
suitable particles in order to trace the fluid motion. Then a pulse of laser
light sheet illuminates a thin slice of the flow field at the desired plane, and
the positions of particles in that plane are determined by detecting the light
scattered by particles on a digital video or photographic camera positioned
at right angles to the light sheet (Fig. 8–77). After a very short time period
t (typically in s), the particles are illuminated again by a second pulse of
laser light sheet, and their new positions are recorded. Using the informa-
tion on these two superimposed camera images, the particle displacements
s are determined for all particles, and the magnitude of velocity of the par-
ticles in the plane of the laser light sheet is determined from s/t. The
direction of motion of the particles is also determined from the two posi-
tions, so that two components of velocity in the plane are calculated. The
built-in algorithms of PIV systems determine the velocities at thousands of
area elements called interrogation regions throughout the entire plane
and display the velocity field on the computer monitor in any desired form
(Fig. 8–78).
The PIV technique relies on the laser light scattered by particles, and thus
the flow must be seeded if necessary with particles, also called markers, in
order to obtain an adequate reflected signal. Seed particles must be able to
follow the pathlines in the flow for their motion to be representative of the
381
CHAPTER 8
Computer
Synchronizer
Pulser
Pulsed Nd:YAG
laser
Sheet-forming optics
Seeded flow
Beam dump
Video camera
FIGURE 8–77
A PIV system to study flame
stabilization.
Courtesy of TSI Incorporated (www.tsi.com).
Used by permission.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 381
flow, and this requires the particle density to be equal to the fluid density
(so that they are neutrally buoyant) or the particles to be so small (typically
m-sized) that their movement relative to the fluid is insignificant. A vari-
ety of such particles is available to seed gas or liquid flow. Very small parti-
cles must be used in high-speed flows. Silicon carbide particles (mean
diameter of 1.5 m) are suitable for both liquid and gas flow, titanium diox-
ide particles (mean diameter of 0.2 m) are usually used for gas flow and
are suitable for high-temperature applications, and polystyrene latex parti-
cles (nominal diameter of 1.0 m) are suitable for low-temperature applica-
tions. Metallic-coated particles (mean diameter of 9.0 m) are also used to
seed water flows for LDV measurements because of their high reflectivity.
Gas bubbles as well as droplets of some liquids such as olive oil or silicon
oil are also used as seeding particles after they are atomized to m-sized
spheres.
A variety of laser light sources such as argon, copper vapor, and Nd:YAG
can be used with PIV systems, depending on the requirements for pulse
duration, power, and time between pulses. Nd:YAG lasers are commonly
used in PIV systems over a wide range of applications. A beam delivery
system such as a light arm or a fiber-optic system is used to generate and
deliver a high-energy pulsed laser sheet at a specified thickness.
With PIV, other flow properties such as vorticity and strain rates can also
be obtained, and the details of turbulence can be studied. Recent advances
in PIV technology have made it possible to obtain three-dimensional veloc-
ity profiles at a cross section of a flow using two cameras (Fig. 8–79). This
is done by recording the images of the target plane simultaneously by both
cameras at different angles, processing the information to produce two sepa-
rate two-dimensional velocity maps, and combining these two maps to gen-
erate the instantaneous three-dimensional velocity field.
382
FLUID MECHANICS
FIGURE 8–78
Instantaneous velocity field in the
wake region of a car as measured by
a PIV system in a wind tunnel. The
velocity vectors are superimposed
on a contour plot of pressure. The
interface between two adjacent gray-
scale levels is an isobar.
Courtesy Dantec Dynamics, Inc.
www.dantecmt.com.
Jet flow
Main
flow
Jet trajectory
Stereoscopic
camera setup
Light-guide
delivery of
laser sheet
Field of view
x
y
FIGURE 8–79
A three-dimensional PIV system set
up to study the mixing of an air jet
with cross duct flow.
Courtesy TSI Incorporated (www.tsi.com).
Used by permission.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 382
383
CHAPTER 8
Guest Author: Lorenz Sigurdson, Vortex Fluid
Dynamics Lab, University of Alberta
The Bernoulli equation is the most beloved of all fluid mechanical equations
because it is a scalar equation and has a vast range of applications. One very
valuable use is in the development of Bernoulli obstruction theory. This the-
ory allows an estimate of the flow velocity from the measured pressure drop
between locations upstream and downstream of an obstruction in a pipe flow.
The volume flow rate can be calculated by using the Bernoulli equation, con-
servation of mass, and the obstruction geometry. The cheapest obstruction to
produce is a plate with a circular orifice in it. There are hundreds of thou-
sands of orifice plate flowmeters in use in North America. It is the accepted
international standard of measurement of volume flow rates. The accuracy
can become very important in industries such as natural gas pipelining where
the commodity is bought and sold based on measurements from these
meters. Some pipes carry more than a million dollars per hour of natural gas.
For practical purposes, meter calibration is required because, although the
pipe and orifice diameter may be known, the flow separates from the lip of
the orifice and creates a flow tube narrower than the orifice diameter. The
flow is accelerated through this vena contracta. Figure 8–80 shows the flow
downstream of the orifice visualized by using a smoke-wire to introduce
streaklines in a transparent flowmeter. The calibration assumes that there is no
pulsation in the pipe flow. However, this is not the case in practice if there is a
reciprocating compressor in the pipeline, or a loose flapping valve. Figure
8–81 shows what can happen to the vena contracta in this circumstance, if the
frequency of the pulsation is near a resonance frequency of the turbulent flow
structures. The vena contracta diameter is reduced. Stop reading and ask
yourself, “Will this cause a flow rate underprediction or overprediction?”
Conservation of mass and the narrower vena contracta mean a higher aver-
age velocity there than without pulsation. The Bernoulli equation says that
the pressure will be lower there as a result, meaning a larger pressure drop
and an overprediction. Errors as high as 40 percent have been found at high
pulsation levels. For the natural gas pipeline mentioned, that could mean
paying (or earning) $400,000 too much per hour! Characteristic instabilities
that have previously been found in shear flows, jet flows, and reattaching
flows (Sigurdson, 1995; Sigurdson and Chapple, 1997) also exist down-
stream of the orifice plate. Thankfully, meter installation designers can now
avoid the dangerously resonant pulsation frequencies associated with these
instabilities, thereby minimizing flowmeter error.
References
Sigurdson, L. W., “The Structure and Control of a Turbulent Reattaching Flow,”
J. Fluid Mechanics, 298, pp. 139–165, 1995.
Sigurdson, L. W., and Chapple, D., “Visualization of Acoustically Pulsated Flow
through an Orifice Plate Flow Meter,” Proc. 1st Pacific Image Processing and
Flow Visualization Conf., Honolulu, HI, February 23–26, 1997.
Sigurdson, L. W., and Chapple, D., “A Turbulent Mechanism for Pulsation—
Induced Orifice Plate Flow Meter Error,” Proc. 13th Australasian Fluid
Mechanics Conf., December 13–18, 1998, Monash U., Melbourne, Australia,
Thompson, M.C., and Hourigan, K., eds., 1, pp. 67–70, 1998.
APPLICATION SPOTLIGHT ■ How Orifice Plate Flowmeters Work, or Do Not Work
D
v
59mm
Re9000
No Pulsation
FIGURE 8–80
Smoke-wire streakline photograph of
orifice plate flowmeter with no
pulsation present, Reynolds number
9000. D
v
indicates the estimated
vena contracta diameter. A hot-wire
probe can be seen along the pipe
centerline.
From Sigurdson and Chapple (1998).
D
v
57mm
Re9000
Stj0.42
u
7
/U
v
13%
CD13%
68mm
FIGURE 8–81
Smoke-wire streakline photograph of
orifice plate flowmeter with pulsation
present, showing a large effect. The
meter is in error by 13%. Reynolds
number 9000. The vena contracta
diameter D
v
is reduced from the no
pulsation case of Fig. 8–80.
From Sigurdson and Chapple (1998).
cen72367_ch08.qxd 11/4/04 7:14 PM Page 383
384
FLUID MECHANICS
SUMMARY
In internal flow, a pipe is completely filled with a fluid. Lam-
inar flow is characterized by smooth streamlines and highly
ordered motion, and turbulent flow is characterized by veloc-
ity fluctuations and highly disordered motion. The Reynolds
number is defined as
Under most practical conditions, the flow in a pipe is laminar
at Re 2300, turbulent at Re 4000, and transitional in
between.
The region of the flow in which the effects of the viscous
shearing forces are felt is called the velocity boundary layer.
The region from the pipe inlet to the point at which the
boundary layer merges at the centerline is called the hydrody-
namic entrance region, and the length of this region is called
the hydrodynamic entry length L
h
. It is given by
The friction coefficient in the fully developed flow region
remains constant. The maximum and average velocities in
fully developed laminar flow in a circular pipe are
The volume flow rate and the pressure drop for laminar flow
in a horizontal pipe are
These results for horizontal pipes can also be used for inclined
pipes provided that P is replaced by P rgL sin u,
The pressure loss and head loss for all types of internal flows
(laminar or turbulent, in circular or noncircular pipes, smooth
or rough surfaces) are expressed as
where rV
2
/2 is the dynamic pressure and the dimensionless
quantity f is the friction factor. For fully developed laminar
flow in a circular pipe, the friction factor is f 64/Re.
For noncircular pipes, the diameter in the previous rela-
tions is replaced by the hydraulic diameter defined as D
h
4A
c
/p, where A
c
is the cross-sectional area of the pipe and
p is its wetted perimeter.
P
L
f
L
D
rV
2
2
and h
L
P
L
rg
f
L
D
V
2
2g
V
#
(P rgL sin u)pD
4
128mL
V
avg
(P rgL sin u)D
2
32mL
and
V
#
V
avg
A
c
PpD
4
128mL
and P
32mLV
avg
D
2
u
max
2V
avg
and V
avg
PD
2
32mL
L
h, laminar
⬵ 0.05 Re D and L
h, turbulent
⬵ 10D
Re
Inertial forces
Viscous forces
V
avg
D
n
rV
avg
D
m
In fully developed turbulent flow, the friction factor
depends on the Reynolds number and the relative roughness
e/D. The friction factor in turbulent flow is given by the
Colebrook equation, expressed as
The plot of this formula is known as the Moody chart. The
design and analysis of piping systems involve the determina-
tion of the head loss, flow rate, or the pipe diameter. Tedious
iterations in these calculations can be avoided by the approx-
imate Swamee–Jain formulas expressed as
The losses that occur in piping components such as fittings,
valves, bends, elbows, tees, inlets, exits, enlargements, and
contractions are called minor losses. The minor losses are
usually expressed in terms of the loss coefficient K
L
. The
head loss for a component is determined from
When all the loss coefficients are available, the total head
loss in a piping system is determined from
If the entire piping system has a constant diameter, the total
head loss reduces to
The analysis of a piping system is based on two simple prin-
ciples: (1) The conservation of mass throughout the system
must be satisfied and (2) the pressure drop between two
points must be the same for all paths between the two points.
h
L, total
af
L
D
a
K
L
b
V
2
2g
h
L, total
h
L, major
h
L, minor
a
i
f
i
L
i
D
i
V
2
i
2g
a
j
K
L, j
V
2
j
2g
h
L
K
L
V
2
2g
10
6
e/D 10
2
5000 Re 3 10
8
D 0.66ce
1.25
a
L
V
#
2
gh
L
b
4.75
n
V
#
9.4
a
L
gh
L
b
5.2
d
0.04
Re 2000
V
#
0.965a
gD
5
h
L
L
b
0.5
lnc
e
3.7D
a
3.17n
2
L
gD
3
h
L
b
0.5
d
10
6
e/D 10
2
3000 Re 3 10
8
h
L
1.07
V
#
2
L
gD
5
elnc
e
3.7D
4.62a
nD
V
#
b
0.9
df
2
1
2f
2.0 loga
e/D
3.7
2.51
Re2f
b
cen72367_ch08.qxd 11/4/04 7:14 PM Page 384
CHAPTER 8
385
When the pipes are connected in series, the flow rate through
the entire system remains constant regardless of the diameters
of the individual pipes. For a pipe that branches out into two
(or more) parallel pipes and then rejoins at a junction down-
stream, the total flow rate is the sum of the flow rates in the
individual pipes but the head loss in each branch is the same.
When a piping system involves a pump and/or turbine, the
steady-flow energy equation is expressed as
When the useful pump head h
pump, u
is known, the mechanical
power that needs to be supplied by the pump to the fluid and
the electric power consumed by the motor of the pump for a
specified flow rate are determined from
W
#
pump, shaft
r
V
#
gh
pump, u
h
pump
and W
#
elect
r
V
#
gh
pump, u
h
pump–motor
P
2
rg
a
2
V
2
2
2g
z
2
h
turbine, e
h
L
P
1
rg
a
1
V
2
1
2g
z
1
h
pump, u
where h
pump–motor
is the efficiency of the pump–motor combi-
nation, which is the product of the pump and the motor effi-
ciencies.
The plot of the head loss versus the flow rate
V
.
is called
the system curve. The head produced by a pump is not a con-
stant, and the curves of h
pump, u
and h
pump
versus
V
.
are called
the characteristic curves. A pump installed in a piping sys-
tem operates at the operating point, which is the point of
intersection of the system curve and the characteristic curve.
Flow measurement techniques and devices can be consid-
ered in three major categories: (1) volume (or mass) flow rate
measurement techniques and devices such as obstruction
flowmeters, turbine meters, positive displacement flowme-
ters, rotameters, and ultrasonic meters; (2) point velocity
measurement techniques such as the Pitot-static probes, hot-
wires, and LDV; and (3) whole-field velocity measurement
techniques such as PIV.
The emphasis in this chapter has been on flow through
pipes. A detailed treatment of numerous types of pumps and
turbines, including their operation principles and performance
parameters, is given in Chap. 14.
REFERENCES AND SUGGESTED READING
1. H. S. Bean (ed.). Fluid Meters: Their Theory and
Applications, 6th ed. New York: American Society of
Mechanical Engineers, 1971.
2. M. S. Bhatti and R. K. Shah. “Turbulent and Transition
Flow Convective Heat Transfer in Ducts.” In Handbook of
Single-Phase Convective Heat Transfer, ed. S. Kakaç, R.
K. Shah, and W. Aung. New York: Wiley Interscience,
1987.
3. C. F. Colebrook. “Turbulent Flow in Pipes, with Particular
Reference to the Transition between the Smooth and
Rough Pipe Laws,” Journal of the Institute of Civil
Engineers London. 11 (1939), pp. 133–156.
4. C. T. Crowe, J. A. Roberson, and D. F. Elger. Engineering
Fluid Mechanics, 7th ed. New York: Wiley, 2001.
5. F. Durst, A. Melling, and J. H. Whitelaw. Principles and
Practice of Laser-Doppler Anemometry, 2nd ed. New
York: Academic, 1981.
6. R. W. Fox and A. T. McDonald. Introduction to Fluid
Mechanics, 5th ed. New York: Wiley, 1999.
7. Fundamentals of Orifice Meter Measurement. Houston,
TX: Daniel Measurement and Control, 1997.
8. S. E. Haaland. “Simple and Explicit Formulas for the
Friction Factor in Turbulent Pipe Flow,” Journal of Fluids
Engineering, March 1983, pp. 89–90.
9. I. E. Idelchik. Handbook of Hydraulic Resistance, 3rd ed.
Boca Raton, FL: CRC Press, 1993.
10. W. M. Kays and M. E. Crawford. Convective Heat and
Mass Transfer, 3rd ed. New York: McGraw-Hill, 1993.
11. R. W. Miller. Flow Measurement Engineering Handbook,
3rd ed. New York: McGraw-Hill, 1997.
12. L. F. Moody. “Friction Factors for Pipe Flows,”
Transactions of the ASME 66 (1944), pp. 671–684.
13. B. R. Munson, D. F. Young, and T. Okiishi. Fundamentals
of Fluid Mechanics, 4th ed. New York: Wiley, 2002.
14. O. Reynolds. “On the Experimental Investigation of the
Circumstances Which Determine Whether the Motion of
Water Shall Be Direct or Sinuous, and the Law of
Resistance in Parallel Channels.” Philosophical
Transactions of the Royal Society of London, 174 (1883),
pp. 935–982.
15. H. Schlichting. Boundary Layer Theory, 7th ed. New
York: McGraw-Hill, 1979.
16. R. K. Shah and M. S. Bhatti. “Laminar Convective Heat
Transfer in Ducts.” In Handbook of Single-Phase
Convective Heat Transfer, ed. S. Kakaç, R. K. Shah, and
W. Aung. New York: Wiley Interscience, 1987.
17. P. L. Skousen. Valve Handbook. New York: McGraw-Hill,
1998.
18. P. K. Swamee and A. K. Jain. “Explicit Equations for
Pipe-Flow Problems,” Journal of the Hydraulics Division.
ASCE 102, no. HY5 (May 1976), pp. 657–664.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 385
19. G. Vass. “Ultrasonic Flowmeter Basics,” Sensors, 14, no.
10 (1997).
20. A. J. Wheeler and A. R. Ganji. Introduction to
Engineering Experimentation. Englewood Cliffs, NJ:
Prentice-Hall, 1996.
386
FLUID MECHANICS
21. F. M. White. Fluid Mechanics, 5th ed. New York:
McGraw-Hill, 2003.
22. W. Zhi-qing. “Study on Correction Coefficients of
Laminar and Turbulent Entrance Region Effects in Round
Pipes,” Applied Mathematical Mechanics, 3 (1982), p. 433.
* Problems designated by a “C” are concept questions, and
students are encouraged to answer them all. Problems designated
by an “E” are in English units, and the SI users can ignore them.
Problems with the icon are solved using EES, and complete
solutions together with parametric studies are included on the
enclosed DVD. Problems with the icon are comprehensive in
nature and are intended to be solved with a computer, preferably
using the EES software that accompanies this text.
PROBLEMS*
Laminar and Turbulent Flow
8–1C Why are liquids usually transported in circular pipes?
8–2C What is the physical significance of the Reynolds
number? How is it defined for (a) flow in a circular pipe of
inner diameter D and (b) flow in a rectangular duct of cross
section a b?
8–10C Consider laminar flow in a circular pipe. Will the
wall shear stress t
w
be higher near the inlet of the pipe or
near the exit? Why? What would your response be if the flow
were turbulent?
8–11C How does surface roughness affect the pressure
drop in a pipe if the flow is turbulent? What would your
response be if the flow were laminar?
Fully Developed Flow in Pipes
8–12C How does the wall shear stress t
w
vary along the
flow direction in the fully developed region in (a) laminar
flow and (b) turbulent flow?
8–13C What fluid property is responsible for the develop-
ment of the velocity boundary layer? For what kinds of fluids
will there be no velocity boundary layer in a pipe?
8–14C In the fully developed region of flow in a circular
pipe, will the velocity profile change in the flow direction?
8–15C How is the friction factor for flow in a pipe related
to the pressure loss? How is the pressure loss related to the
pumping power requirement for a given mass flow rate?
8–16C Someone claims that the shear stress at the center of
a circular pipe during fully developed laminar flow is zero.
Do you agree with this claim? Explain.
8–17C Someone claims that in fully developed turbulent
flow in a pipe, the shear stress is a maximum at the pipe sur-
face. Do you agree with this claim? Explain.
8–18C Consider fully developed flow in a circular pipe
with negligible entrance effects. If the length of the pipe is
doubled, the head loss will (a) double, (b) more than double,
(c) less than double, (d) reduce by half, or (e) remain con-
stant.
8–19C Someone claims that the volume flow rate in a cir-
cular pipe with laminar flow can be determined by measuring
the velocity at the centerline in the fully developed region,
multiplying it by the cross-sectional area, and dividing the
result by 2. Do you agree? Explain.
8–20C Someone claims that the average velocity in a circu-
lar pipe in fully developed laminar flow can be determined
by simply measuring the velocity at R/2 (midway between
the wall surface and the centerline). Do you agree? Explain.
a
b
D
FIGURE P8–2C
8–3C Consider a person walking first in air and then in
water at the same speed. For which motion will the Reynolds
number be higher?
8–4C Show that the Reynolds number for flow in a circular
pipe of diameter D can be expressed as Re 4m
.
/(pDm).
8–5C Which fluid at room temperature requires a larger
pump to flow at a specified velocity in a given pipe: water or
engine oil? Why?
8–6C What is the generally accepted value of the Reynolds
number above which the flow in smooth pipes is turbulent?
8–7C Consider the flow of air and water in pipes of the
same diameter, at the same temperature, and at the same
mean velocity. Which flow is more likely to be turbulent?
Why?
8–8C What is hydraulic diameter? How is it defined? What
is it equal to for a circular pipe of diameter D?
8–9C How is the hydrodynamic entry length defined for
flow in a pipe? Is the entry length longer in laminar or turbu-
lent flow?
cen72367_ch08.qxd 11/4/04 7:14 PM Page 386
CHAPTER 8
387
8–21C Consider fully developed laminar flow in a circular
pipe. If the diameter of the pipe is reduced by half while the
flow rate and the pipe length are held constant, the head loss
will (a) double, (b) triple, (c) quadruple, (d) increase by a
factor of 8, or (e) increase by a factor of 16.
8–22C What is the physical mechanism that causes the fric-
tion factor to be higher in turbulent flow?
8–23C What is turbulent viscosity? What is it caused by?
8–24C The head loss for a certain circular pipe is given by
h
L
0.0826fL(
V
.
2
/D
5
), where f is the friction factor (dimen-
sionless), L is the pipe length,
V
.
is the volumetric flow rate,
and D is the pipe diameter. Determine if the 0.0826 is a dimen-
sional or dimensionless constant. Is this equation dimension-
ally homogeneous as it stands?
8–25C Consider fully developed laminar flow in a circular
pipe. If the viscosity of the fluid is reduced by half by heat-
ing while the flow rate is held constant, how will the head
loss change?
8–26C How is head loss related to pressure loss? For a
given fluid, explain how you would convert head loss to pres-
sure loss.
8–27C Consider laminar flow of air in a circular pipe with
perfectly smooth surfaces. Do you think the friction factor for
this flow will be zero? Explain.
8–28C Explain why the friction factor is independent of the
Reynolds number at very large Reynolds numbers.
8–29E Oil at 80°F (r 56.8 lbm/ft
3
and m 0.0278 lbm/ft
· s) is flowing steadily in a 0.5-in-diameter, 120-ft-long pipe.
During the flow, the pressure at the pipe inlet and exit is
measured to be 120 psi and 14 psi, respectively. Determine
the flow rate of oil through the pipe assuming the pipe is
(a) horizontal, (b) inclined 20° upward, and (c) inclined 20°
downward.
8–30 Oil with a density of 850 kg/m
3
and kinematic viscos-
ity of 0.00062 m
2
/s is being discharged by a 5-mm-diameter,
40-m-long horizontal pipe from a storage tank open to the
atmosphere. The height of the liquid level above the center of
the pipe is 3 m. Disregarding the minor losses, determine the
flow rate of oil through the pipe.
8–31 Water at 10°C (r 999.7 kg/m
3
and m 1.307
10
3
kg/m · s) is flowing steadily in a 0.20-cm-diameter,
15-m-long pipe at an average velocity of 1.2 m/s. Determine
(a) the pressure drop, (b) the head loss, and (c) the pumping
power requirement to overcome this pressure drop. Answers:
(a) 188 kPa, (b) 19.2 m, (c) 0.71 W
8–32 Water at 15°C (r 999.1 kg/m
3
and m 1.138
10
3
kg/m · s) is flowing steadily in a 30-m-long and
4-cm-diameter horizontal pipe made of stainless steel at a
rate of 8 L/s. Determine (a) the pressure drop, (b) the head
loss, and (c) the pumping power requirement to overcome
this pressure drop.
Oil
tank
3 m
5 mm
FIGURE P8–30
30 m
4 cm
8 L/s
FIGURE P8–32
8–33E Heated air at 1 atm and 100°F is to be transported in
a 400-ft-long circular plastic duct at a rate of 12 ft
3
/s. If the
head loss in the pipe is not to exceed 50 ft, determine the
minimum diameter of the duct.
8–34 In fully developed laminar flow in a circular pipe, the
velocity at R/2 (midway between the wall surface and the
centerline) is measured to be 6 m/s. Determine the velocity at
the center of the pipe.
Answer: 8 m/s
8–35 The velocity profile in fully developed laminar flow
in a circular pipe of inner radius R 2 cm, in m/s, is given
by u(r) 4(1 r
2
/R
2
). Determine the average and maximum
velocities in the pipe and the volume flow rate.
R = 2 cm
u(r) = 4 1 –
r
2
––
R
2
a
b
FIGURE P8–35
8–36 Repeat Prob. 8–35 for a pipe of inner radius 7 cm.
8–37 Consider an air solar collector that is 1 m wide and 5
m long and has a constant spacing of 3 cm between the glass
cover and the collector plate. Air flows at an average temper-
ature of 45°C at a rate of 0.15 m
3
/s through the 1-m-wide
edge of the collector along the 5-m-long passageway. Disre-
garding the entrance and roughness effects, determine the
pressure drop in the collector.
Answer: 29 Pa
cen72367_ch08.qxd 11/4/04 7:14 PM Page 387
8–38 Consider the flow of oil with r 894 kg/m
3
and m
2.33 kg/m · s in a 40-cm-diameter pipeline at an average
velocity of 0.5 m/s. A 300-m-long section of the pipeline
passes through the icy waters of a lake. Disregarding the
entrance effects, determine the pumping power required to
overcome the pressure losses and to maintain the flow of oil
in the pipe.
8–39 Consider laminar flow of a fluid through a square
channel with smooth surfaces. Now the average velocity of
the fluid is doubled. Determine the change in the head loss of
the fluid. Assume the flow regime remains unchanged.
8–40 Repeat Prob. 8–39 for turbulent flow in smooth pipes
for which the friction factor is given as f 0.184Re
0.2
.
What would your answer be for fully turbulent flow in a
rough pipe?
8–41 Air enters a 7-m-long section of a rectangular duct of
cross section 15 cm 20 cm made of commercial steel at 1
atm and 35°C at an average velocity of 7 m/s. Disregarding
the entrance effects, determine the fan power needed to over-
come the pressure losses in this section of the duct.
Answer:
4.9 W
388
FLUID MECHANICS
atmosphere at 88 kPa. The absolute pressure 15 m before the
exit is measured to be 135 kPa. Determine the flow rate of oil
through the pipe if the pipe is (a) horizontal, (b) inclined 8°
upward from the horizontal, and (c) inclined 8° downward
from the horizontal.
Collector plate
Insulation
Glass cover
5 m
Air
0.15 m
3
/s
FIGURE P8–37
7 m
15 cm
20 cm
Air
7 m/s
FIGURE P8–41
8–42E Water at 60°F passes through 0.75-in-internal-
diameter copper tubes at a rate of 1.2 lbm/s. Determine the
pumping power per ft of pipe length required to maintain this
flow at the specified rate.
8–43 Oil with r 876 kg/m
3
and m 0.24 kg/m · s is flow-
ing through a 1.5-cm-diameter pipe that discharges into the
1.5 cm
15 m
135 kPa
Oil
FIGURE P8–43
8–44 Glycerin at 40°C with r 1252 kg/m
3
and m
0.27 kg/m · s is flowing through a 2-cm-diameter, 25-m-
long pipe that discharges into the atmosphere at 100 kPa. The
flow rate through the pipe is 0.035 L/s. (a) Determine the
absolute pressure 25 m before the pipe exit. (b) At what angle
u must the pipe be inclined downward from the horizontal for
the pressure in the entire pipe to be atmospheric pressure and
the flow rate to be maintained the same?
8–45 In an air heating system, heated air at 40°C and
105 kPa absolute is distributed through a 0.2 m 0.3 m rec-
tangular duct made of commercial steel at a rate of 0.5 m
3
/s.
Determine the pressure drop and head loss through a 40-m-
long section of the duct. Answers: 128 Pa, 93.8 m
8–46 Glycerin at 40°C with r 1252 kg/m
3
and m
0.27 kg/m · s is flowing through a 5-cm-diameter horizontal
smooth pipe with an average velocity of 3.5 m/s. Determine
the pressure drop per 10 m of the pipe.
8–47 Reconsider Prob. 8–46. Using EES (or other)
software, investigate the effect of the pipe diam-
eter on the pressure drop for the same constant flow rate. Let
the pipe diameter vary from 1 to 10 cm in increments of 1
cm. Tabulate and plot the results, and draw conclusions.
8–48E Air at 1 atm and 60°F is flowing through a 1 ft 1ft
square duct made of commercial steel at a rate of 1200 cfm.
Determine the pressure drop and head loss per ft of the duct.
1 ft
1 ft
Air
1200 ft
3
/min
FIGURE P8–48E
8–49 Liquid ammonia at 20°C is flowing through a 30-
m-long section of a 5-mm-diameter copper tube at a rate of
cen72367_ch08.qxd 11/4/04 7:14 PM Page 388
CHAPTER 8
389
0.15 kg/s. Determine the pressure drop, the head loss, and the
pumping power required to overcome the frictional losses in
the tube.
Answers: 4792 kPa, 743 m, 1.08 kW
8–50 Shell-and-tube heat exchangers with hundreds
of tubes housed in a shell are commonly used in
practice for heat transfer between two fluids. Such a heat
exchanger used in an active solar hot-water system transfers
heat from a water-antifreeze solution flowing through the
shell and the solar collector to fresh water flowing through
the tubes at an average temperature of 60°C at a rate of
15 L/s. The heat exchanger contains 80 brass tubes 1 cm in
inner diameter and 1.5 m in length. Disregarding inlet, exit,
and header losses, determine the pressure drop across a single
tube and the pumping power required by the tube-side fluid
of the heat exchanger.
After operating a long time, 1-mm-thick scale builds up on
the inner surfaces with an equivalent roughness of 0.4 mm.
For the same pumping power input, determine the percent
reduction in the flow rate of water through the tubes.
into the miter elbows or to replace the sharp turns in 90°
miter elbows by smooth curved bends. Which approach will
result in a greater reduction in pumping power requirements?
8–58 Water is to be withdrawn from a 3-m-high water
reservoir by drilling a 1.5-cm-diameter hole at the bottom
surface. Disregarding the effect of the kinetic energy correc-
tion factor, determine the flow rate of water through the hole
if (a) the entrance of the hole is well-rounded and (b) the
entrance is sharp-edged.
8–59 Consider flow from a water reservoir through a circu-
lar hole of diameter D at the side wall at a vertical distance H
from the free surface. The flow rate through an actual hole
with a sharp-edged entrance (K
L
0.5) will be considerably
less than the flow rate calculated assuming “frictionless” flow
and thus zero loss for the hole. Disregarding the effect of the
kinetic energy correction factor, obtain a relation for the
“equivalent diameter” of the sharp-edged hole for use in fric-
tionless flow relations.
1.5 m
80 tubes
1 cm
Water
FIGURE P8–50
Minor Losses
8–51C
What is minor loss in pipe flow? How is the minor
loss coefficient K
L
defined?
8–52C Define equivalent length for minor loss in pipe flow.
How is it related to the minor loss coefficient?
8–53C The effect of rounding of a pipe inlet on the loss
coefficient is (a) negligible, (b) somewhat significant, or
(c) very significant.
8–54C The effect of rounding of a pipe exit on the loss
coefficient is (a) negligible, (b) somewhat significant, or
(c) very significant.
8–55C Which has a greater minor loss coefficient during
pipe flow: gradual expansion or gradual contraction? Why?
8–56C A piping system involves sharp turns, and thus large
minor head losses. One way of reducing the head loss is to
replace the sharp turns by circular elbows. What is another
way?
8–57C During a retrofitting project of a fluid flow system
to reduce the pumping power, it is proposed to install vanes
D
Frictionless flow Actual flow
D
equiv.
FIGURE P8–59
8–60
Repeat Prob. 8–59 for a slightly rounded entrance
(K
L
0.12).
8–61 A horizontal pipe has an abrupt expansion from D
1
8 cm to D
2
16 cm. The water velocity in the smaller
section is 10 m/s and the flow is turbulent. The pressure in
the smaller section is P
1
300 kPa. Taking the kinetic
energy correction factor to be 1.06 at both the inlet and the
outlet, determine the downstream pressure P
2
, and estimate
the error that would have occurred if Bernoulli’s equation had
been used.
Answers: 321 kPa, 28 kPa
D
2
= 16 cm
D
1
= 8 cm
10 m/s
300 kPa
Water
FIGURE P8–61
Piping Systems and Pump Selection
8–62C
A piping system involves two pipes of different
diameters (but of identical length, material, and roughness)
connected in series. How would you compare the (a) flow
rates and (b) pressure drops in these two pipes?
cen72367_ch08.qxd 11/4/04 7:14 PM Page 389
8–63C A piping system involves two pipes of different
diameters (but of identical length, material, and roughness)
connected in parallel. How would you compare the (a) flow
rates and (b) pressure drops in these two pipes?
8–64C A piping system involves two pipes of identical
diameters but of different lengths connected in parallel. How
would you compare the pressure drops in these two pipes?
8–65C Water is pumped from a large lower reservoir to a
higher reservoir. Someone claims that if the head loss is
negligible, the required pump head is equal to the elevation
difference between the free surfaces of the two reservoirs.
Do you agree?
8–66C A piping system equipped with a pump is operating
steadily. Explain how the operating point (the flow rate and
the head loss) is established.
8–67C For a piping system, define the system curve, the
characteristic curve, and the operating point on a head versus
flow rate chart.
8–68 Water at 20°C is to be pumped from a reservoir
(z
A
2 m) to another reservoir at a higher ele-
vation (z
B
9 m) through two 25-m-long plastic pipes con-
nected in parallel. The diameters of the two pipes are 3 cm
and 5 cm. Water is to be pumped by a 68 percent efficient
motor–pump unit that draws 7 kW of electric power during
operation. The minor losses and the head loss in the pipes
that connect the parallel pipes to the two reservoirs are con-
sidered to be negligible. Determine the total flow rate
between the reservoirs and the flow rates through each of the
parallel pipes.
390
FLUID MECHANICS
8–70 A 3-m-diameter tank is initially filled with water 2 m
above the center of a sharp-edged 10-cm-diameter orifice.
The tank water surface is open to the atmosphere, and the
orifice drains to the atmosphere. Neglecting the effect of the
kinetic energy correction factor, calculate (a) the initial
velocity from the tank and (b) the time required to empty the
tank. Does the loss coefficient of the orifice cause a signifi-
cant increase in the draining time of the tank?
Pump
Reservoir A
z
A
= 2 m
25 m
3 cm
5 cm
Reservoir B
z
B
= 9 m
FIGURE P8–68
8–69E Water at 70°F flows by gravity from a large reservoir
at a high elevation to a smaller one through a 120-ft-long, 2-
in-diameter cast iron piping system that includes four stan-
dard flanged elbows, a well-rounded entrance, a sharp-edged
exit, and a fully open gate valve. Taking the free surface of
the lower reservoir as the reference level, determine the ele-
vation z
1
of the higher reservoir for a flow rate of 10 ft
3
/min.
Answer: 23.1 ft
Water
tank
2 m
3 m
Sharp-edged
orifice
FIGURE P8–70
8–71 A 3-m-diameter tank is initially filled with water 2 m
above the center of a sharp-edged 10-cm-diameter orifice.
The tank water surface is open to the atmosphere, and the
orifice drains to the atmosphere through a 100-m-long pipe.
The friction coefficient of the pipe can be taken to be 0.015
and the effect of the kinetic energy correction factor can be
neglected. Determine (a) the initial velocity from the tank
and (b) the time required to empty the tank.
8–72 Reconsider Prob. 8–71. In order to drain the tank faster,
a pump is installed near the tank exit. Determine how much
pump power input is necessary to establish an average water
velocity of 4 m/s when the tank is full at z 2 m. Also,
assuming the discharge velocity to remain constant, estimate
the time required to drain the tank.
Someone suggests that it makes no difference whether the
pump is located at the beginning or at the end of the pipe,
and that the performance will be the same in either case, but
Water
tank
Pump
4 m/s
2 m
3 m
FIGURE P8–72
cen72367_ch08.qxd 11/4/04 7:14 PM Page 390
CHAPTER 8
391
another person argues that placing the pump near the end
of the pipe may cause cavitation. The water temperature
is 30°C, so the water vapor pressure is P
v
4.246 kPa
0.43 m-H
2
O, and the system is located at sea level. Investi-
gate if there is the possibility of cavitation and if we should
be concerned about the location of the pump.
8–73 Oil at 20°C is flowing through a vertical glass funnel
that consists of a 15-cm-high cylindrical reservoir and a 1-
cm-diameter, 25-cm-high pipe. The funnel is always main-
tained full by the addition of oil from a tank. Assuming the
entrance effects to be negligible, determine the flow rate of
oil through the funnel and calculate the “funnel effective-
ness,” which can be defined as the ratio of the actual flow
rate through the funnel to the maximum flow rate for the
“frictionless” case. Answers: 4.09 10
6
m
3
/s, 1.86 percent
8–76E A farmer is to pump water at 70°F from a river to a
water storage tank nearby using a 125-ft-long, 5-in-diameter
plastic pipe with three flanged 90° smooth bends. The water
velocity near the river surface is 6 ft/s, and the pipe inlet is
placed in the river normal to the flow direction of water to
take advantage of the dynamic pressure. The elevation differ-
ence between the river and the free surface of the tank is
12 ft. For a flow rate of 1.5 ft
3
/s and an overall pump effi-
ciency of 70 percent, determine the required electric power
input to the pump.
8–77E Reconsider Prob. 8–76E. Using EES (or other)
software, investigate the effect of the pipe
diameter on the required electric power input to the pump.
Let the pipe diameter vary from 1 to 10 in, in increments of
1 in. Tabulate and plot the results, and draw conclusions.
8–78 A water tank filled with solar-heated water at 40°C is
to be used for showers in a field using gravity-driven flow.
The system includes 20 m of 1.5-cm-diameter galvanized
iron piping with four miter bends (90°) without vanes and a
wide-open globe valve. If water is to flow at a rate of 0.7 L/s
through the shower head, determine how high the water level
in the tank must be from the exit level of the shower. Disre-
gard the losses at the entrance and at the shower head, and
neglect the effect of the kinetic energy correction factor.
8–79 Two water reservoirs A and B are connected to each
other through a 40-m-long, 2-cm-diameter cast iron pipe with
a sharp-edged entrance. The pipe also involves a swing check
valve and a fully open gate valve. The water level in both
reservoirs is the same, but reservoir A is pressurized by com-
pressed air while reservoir B is open to the atmosphere at
88 kPa. If the initial flow rate through the pipe is 1.2 L/s,
determine the absolute air pressure on top of reservoir A.
Take the water temperature to be 10°C. Answer: 733 kPa
15 cm
25 cm
1 cm
Oil
Oil
FIGURE P8–73
8–74 Repeat Prob. 8–73 assuming (a) the diameter of the
pipe is doubled and (b) the length of the pipe is doubled.
8–75 Water at 15°C is drained from a large reservoir using
two horizontal plastic pipes connected in series. The first pipe
is 20 m long and has a 10-cm diameter, while the second pipe
is 35 m long and has a 4-cm diameter. The water level in the
reservoir is 18 m above the centerline of the pipe. The pipe
entrance is sharp-edged, and the contraction between the two
pipes is sudden. Neglecting the effect of the kinetic energy
correction factor, determine the discharge rate of water from
the reservoir.
Water
tank
18 m
20 m
35 m
FIGURE P8–75
Air
2 cm
40 m
FIGURE P8–79
8–80
A vented tanker is to be filled with fuel oil with r
920 kg/m
3
and m 0.045 kg/m · s from an underground
reservoir using a 20-m-long, 5-cm-diameter plastic hose with
a slightly rounded entrance and two 90° smooth bends. The
elevation difference between the oil level in the reservoir and
the top of the tanker where the hose is discharged is 5 m. The
capacity of the tanker is 18 m
3
and the filling time is 30 min.
Taking the kinetic energy correction factor at hose discharge
cen72367_ch08.qxd 11/4/04 7:14 PM Page 391
to be 1.05 and assuming an overall pump efficiency of
82 percent, determine the required power input to the pump.
392
FLUID MECHANICS
ligible because of the large length-to-diameter ratio and the
relatively small number of components that cause minor
losses. (a) Assuming the pump–motor efficiency to be 74 per-
cent, determine the electric power consumption of the system
for pumping. Would you recommend the use of a single large
pump or several smaller pumps of the same total pumping
power scattered along the pipeline? Explain. (b) Determine
the daily cost of power consumption of the system if the unit
cost of electricity is $0.06/kWh. (c) The temperature of geo-
thermal water is estimated to drop 0.5°C during this long
flow. Determine if the frictional heating during flow can
make up for this drop in temperature.
8–85 Repeat Prob. 8–84 for cast iron pipes of the same
diameter.
8–86E A clothes dryer discharges air at 1 atm and 120°F at
a rate of 1.2 ft
3
/s when its 5-in-diameter, well-rounded vent
with negligible loss is not connected to any duct. Determine
the flow rate when the vent is connected to a 15-ft-long, 5-in-
diameter duct made of galvanized iron, with three 90°
flanged smooth bends. Take the friction factor of the duct to
be 0.019, and assume the fan power input to remain constant.
3000 m
30 cm
0.4 m
3
/s
30 cm
1000 m
A
B
FIGURE P8–82
8–81 Two pipes of identical length and material are con-
nected in parallel. The diameter of pipe A is twice the diame-
ter of pipe B. Assuming the friction factor to be the same in
both cases and disregarding minor losses, determine the ratio
of the flow rates in the two pipes.
8–82 A certain part of cast iron piping of a water distribu-
tion system involves a parallel section. Both parallel pipes
have a diameter of 30 cm, and the flow is fully turbulent. One
of the branches (pipe A) is 1000 m long while the other
branch (pipe B) is 3000 m long. If the flow rate through pipe
A is 0.4 m
3
/s, determine the flow rate through pipe B. Disre-
gard minor losses and assume the water temperature to be
15°C. Show that the flow is fully turbulent, and thus the fric-
tion factor is independent of Reynolds number.
Answer:
0.231 m
3
/s
Hot ai
r
Clothes drier
15 ft
5 in
FIGURE P8–86E
8–87 In large buildings, hot water in a water tank is circu-
lated through a loop so that the user doesn’t have to wait for
all the water in long piping to drain before hot water starts
coming out. A certain recirculating loop involves 40-m-long,
1.2-cm-diameter cast iron pipes with six 90° threaded smooth
bends and two fully open gate valves. If the average flow
velocity through the loop is 2.5 m/s, determine the required
power input for the recirculating pump. Take the average
water temperature to be 60°C and the efficiency of the pump
to be 70 percent.
Answer: 0.217 kW
8–88 Reconsider Prob. 8–87. Using EES (or other)
software, investigate the effect of the average
flow velocity on the power input to the recirculating pump.
5 cm
Pump
5 m
20 m
Tanker
18 m
3
FIGURE P8–80
8–83 Repeat Prob. 8–82 assuming pipe A has a halfway-
closed gate valve (K
L
2.1) while pipe B has a fully open
globe valve (K
L
10), and the other minor losses are negli-
gible. Assume the flow to be fully turbulent.
8–84 A geothermal district heating system involves the
transport of geothermal water at 110°C from a geothermal
well to a city at about the same elevation for a distance of
12 km at a rate of 1.5 m
3
/s in 60-cm-diameter stainless-steel
pipes. The fluid pressures at the wellhead and the arrival
point in the city are to be the same. The minor losses are neg-
cen72367_ch08.qxd 11/4/04 7:14 PM Page 392
CHAPTER 8
393
Let the velocity vary from 0 to 3 m/s in increments of
0.3 m/s. Tabulate and plot the results.
8–89 Repeat Prob. 8–87 for plastic pipes.
Flow Rate and Velocity Measurements
8–90C What are the primary considerations when selecting
a flowmeter to measure the flow rate of a fluid?
8–91C Explain how flow rate is measured with a Pitot-static
tube, and discuss its advantages and disadvantages with respect
to cost, pressure drop, reliability, and accuracy.
8–92C Explain how flow rate is measured with obstruction-
type flowmeters. Compare orifice meters, flow nozzles, and
Venturi meters with respect to cost, size, head loss, and
accuracy.
8–93C How do positive displacement flowmeters operate?
Why are they commonly used to meter gasoline, water, and
natural gas?
8–94C Explain how flow rate is measured with a turbine
flowmeter, and discuss how they compare to other types of
flowmeters with respect to cost, head loss, and accuracy.
8–95C What is the operating principle of variable-area
flowmeters (rotameters)? How do they compare to other
types of flowmeters with respect to cost, head loss, and relia-
bility?
8–96C What is the difference between the operating princi-
ples of thermal and laser Doppler anemometers?
8–97C What is the difference between laser Doppler velo-
cimetry (LDV) and particle image velocimetry (PIV)?
8–98 The flow rate of ammonia at 10°C (r 624.6 kg/m
3
and m 1.697 10
4
kg/m · s) through a 3-cm-diameter
pipe is to be measured with a 1.5-cm-diameter flow nozzle
equipped with a differential pressure gage. If the gage reads a
pressure differential of 4 kPa, determine the flow rate of
ammonia through the pipe, and the average flow velocity.
8–99 The flow rate of water through a 10-cm-diameter pipe
is to be determined by measuring the water velocity at sev-
eral locations along a cross section. For the set of measure-
ments given in the table, determine the flow rate.
r, cm V, m/s
0 6.4
1 6.1
2 5.2
3 4.4
4 2.0
5 0.0
8–100E An orifice with a 2-in-diameter opening is used to
measure the mass flow rate of water at 60°F (r 62.36 lbm/ft
3
and m 7.536 10
4
lbm/ft · s) through a horizontal
4-in-diameter pipe. A mercury manometer is used to measure
the pressure difference across the orifice. If the differential
height of the manometer is read to be 6 in, determine the vol-
ume flow rate of water through the pipe, the average velocity,
and the head loss caused by the orifice meter.
2 in4 in
6 in
FIGURE P8–100E
5 cm 3 cm
P
Differential
pressure
g
a
g
e
FIGURE P8–103
8–101E Repeat Prob. 8–100E for a differential height of
9 in.
8–102 The flow rate of water at 20°C (r 998 kg/m
3
and
m 1.002 10
3
kg/m · s) through a 50-cm-diameter pipe
is measured with an orifice meter with a 30-cm-diameter
opening to be 250 L/s. Determine the pressure difference
indicated by the orifice meter and the head loss.
8–103 A Venturi meter equipped with a differential pres-
sure gage is used to measure the flow rate of water at 15°C
(r 999.1 kg/m
3
) through a 5-cm-diameter horizontal pipe.
The diameter of the Venturi neck is 3 cm, and the measured
pressure drop is 5 kPa. Taking the discharge coefficient to be
0.98, determine the volume flow rate of water and the aver-
age velocity through the pipe.
Answers: 2.35 L/s and 1.20 m/s
cen72367_ch08.qxd 11/4/04 7:14 PM Page 393
8–104 Reconsider Prob. 8–103. Letting the pressure
drop vary from 1 kPa to 10 kPa, evaluate the
flow rate at intervals of 1 kPa, and plot it against the pressure
drop.
8–105 The mass flow rate of air at 20°C (r 1.204 kg/m
3
)
through a 15-cm-diameter duct is measured with a Venturi
meter equipped with a water manometer. The Venturi neck
has a diameter of 6 cm, and the manometer has a maximum
differential height of 40 cm. Taking the discharge coefficient
to be 0.98, determine the maximum mass flow rate of air this
Venturi meter can measure.
Answer: 0.273 kg/s
394
FLUID MECHANICS
flow rate of liquid propane at 10°C (r 514.7 kg/m
3
)
through an 8-cm-diameter vertical pipe. For a discharge coef-
ficient of 0.98, determine the volume flow rate of propane
through the pipe.
8–108 A flow nozzle equipped with a differential pressure
gage is used to measure the flow rate of water at 10°C (r
999.7 kg/m
3
and m 1.307 10
3
kg/m · s) through a 3-
cm-diameter horizontal pipe. The nozzle exit diameter is 1.5
cm, and the measured pressure drop is 3 kPa. Determine the
volume flow rate of water, the average velocity through the
pipe, and the head loss.
15 cm 6 cm
Water
manometer
h
FIGURE P8–105
8 cm
5 cm
30 cm
P = 7 kPa
FIGURE P8–107
8–106 Repeat Prob. 8–105 for a Venturi neck diameter of
7.5 cm.
8–107 A vertical Venturi meter equipped with a differential
pressure gage shown in Fig. P8–107 is used to measure the
1.5 cm3 cm
P = 3 kPa
Differential
pressure gage?
FIGURE P8–108
2 cm4 cm
32 cm
Water
FIGURE P8–110
8–109 A 16-L kerosene tank (r 820 kg/m
3
) is filled with
a 2-cm-diameter hose equipped with a 1.5-cm-diameter noz-
zle meter. If it takes 20 s to fill the tank, determine the pres-
sure difference indicated by the nozzle meter.
8–110 The flow rate of water at 20°C (r 998 kg/m
3
and
m 1.002 10
3
kg/m · s) through a 4-cm-diameter pipe is
measured with a 2-cm-diameter nozzle meter equipped with
an inverted air–water manometer. If the manometer indicates
cen72367_ch08.qxd 11/4/04 7:14 PM Page 394
CHAPTER 8
395
a differential water height of 32 cm, determine the volume
flow rate of water and the head loss caused by the nozzle
meter.
8–111E The volume flow rate of liquid refrigerant-134a at
10°F (r 83.31 lbm/ft
3
) is to be measured with a horizontal
Venturi meter with a diameter of 5 in at the inlet and 2 in at
the throat. If a differential pressure meter indicates a pressure
drop of 7.4 psi, determine the flow rate of the refrigerant.
Take the discharge coefficient of the Venturi meter to be 0.98.
Review Problems
8–112
The compressed air requirements of a manufacturing
facility are met by a 150-hp compressor that draws in air
from the outside through an 8-m-long, 20-cm-diameter duct
made of thin galvanized iron sheets. The compressor takes in
air at a rate of 0.27 m
3
/s at the outdoor conditions of 15°C
and 95 kPa. Disregarding any minor losses, determine the
useful power used by the compressor to overcome the fric-
tional losses in this duct.
Answer: 9.66 W
tion of a circular stainless-steel duct of 20-cm diameter
passes through the water. Air flows through the underwater
section of the duct at 3 m/s at an average temperature of
15°C. For an overall fan efficiency of 62 percent, determine
the fan power needed to overcome the flow resistance in this
section of the duct.
8–114 The velocity profile in fully developed laminar flow
in a circular pipe, in m/s, is given by u(r) 6(1 100r
2
),
where r is the radial distance from the centerline of the pipe
in m. Determine (a) the radius of the pipe, (b) the average
velocity through the pipe, and (c) the maximum velocity in
the pipe.
8–115E The velocity profile in a fully developed laminar
flow of water at 40°F in a 80-ft-long horizontal circular pipe,
in ft/s, is given by u(r) 0.8(1 625r
2
), where r is the radial
distance from the centerline of the pipe in ft. Determine
(a) the volume flow rate of water through the pipe, (b) the
pressure drop across the pipe, and (c) the useful pumping
power required to overcome this pressure drop.
8–116E Repeat Prob. 8–115E assuming the pipe is inclined
12° from the horizontal and the flow is uphill.
8–117 Consider flow from a reservoir through a horizontal
pipe of length L and diameter D that penetrates into the side
wall at a vertical distance H from the free surface. The flow
rate through an actual pipe with a reentrant section (K
L
0.8) will be considerably less than the flow rate through
the hole calculated assuming “frictionless” flow and thus zero
loss. Obtain a relation for the “equivalent diameter” of the
reentrant pipe for use in relations for frictionless flow
through a hole and determine its value for a pipe friction fac-
tor, length, and diameter of 0.018, 10 m, and 0.04 m, respec-
tively. Assume the friction factor of the pipe to remain con-
stant and the effect of the kinetic energy correction factor to
be negligible.
8–118 Water is to be withdrawn from a 5-m-high water
reservoir by drilling a well-rounded 3-cm-diameter hole with
negligible loss at the bottom surface and attaching a horizon-
tal 90° bend of negligible length. Taking the kinetic energy
correction factor to be 1.05, determine the flow rate of water
through the bend if (a) the bend is a flanged smooth bend
8 m
20
cm
Air, 0.27 m
3
/s
15°C, 95 kPa
Air
compressor
150 hp
FIGURE P8–112
8–113 A house built on a riverside is to be cooled in sum-
mer by utilizing the cool water of the river. A 15-m-long sec-
River
Air
Air, 3 m/s
FIGURE P8–113
5 m
FIGURE P8–118
cen72367_ch08.qxd 11/4/04 7:14 PM Page 395
and (b) the bend is a miter bend without vanes. Answers:
(a) 0.00603 m
3
/s, (b) 0.00478 m
3
/s
8–119 In a geothermal district heating system, 10,000
kg/s of hot water must be delivered a distance
of 10 km in a horizontal pipe. The minor losses are negligi-
ble, and the only significant energy loss will arise from pipe
friction. The friction factor can be taken to be 0.015. Specify-
ing a larger-diameter pipe would reduce water velocity,
velocity head, pipe friction, and thus power consumption. But
a larger pipe would also cost more money initially to pur-
chase and install. Otherwise stated, there is an optimum pipe
diameter that will minimize the sum of pipe cost and future
electric power cost.
Assume the system will run 24 h/day, every day, for 30
years. During this time the cost of electricity will remain con-
stant at $0.06/kWh. Assume system performance stays con-
stant over the decades (this may not be true, especially if
highly mineralized water is passed through the pipeline—
scale may form). The pump has an overall efficiency of 80
percent. The cost to purchase, install, and insulate a 10-km
pipe depends on the diameter D and is given by Cost
$10
6
D
2
, where D is in m. Assuming zero inflation and
interest rate for simplicity and zero salvage value and zero
maintenance cost, determine the optimum pipe diameter.
8–120 Water at 15°C is to be discharged from a reservoir at
a rate of 18 L/s using two horizontal cast iron pipes con-
nected in series and a pump between them. The first pipe is
20 m long and has a 6-cm diameter, while the second pipe is
35 m long and has a 4-cm diameter. The water level in the
reservoir is 30 m above the centerline of the pipe. The pipe
entrance is sharp-edged, and losses associated with the con-
nection of the pump are negligible. Neglecting the effect of
the kinetic energy correction factor, determine the required
pumping head and the minimum pumping power to maintain
the indicated flow rate.
396
FLUID MECHANICS
8–121 Reconsider Prob. 8–120. Using EES (or other)
software, investigate the effect of the second
pipe diameter on the required pumping head to maintain the
Water
tank
Pump
35 m
20 m
30 m
6 cm
4 cm
FIGURE P8–120
500 m
30 cm
800 m
3 m
3
/s
Oil
A
B
45 cm
FIGURE P8–123
indicated flow rate. Let the diameter vary from 1 to 10 cm in
increments of 1 cm. Tabulate and plot the results.
8–122 Two pipes of identical diameter and material are
connected in parallel. The length of pipe A is twice the length
of pipe B. Assuming the flow is fully turbulent in both pipes
and thus the friction factor is independent of the Reynolds
number and disregarding minor losses, determine the ratio of
the flow rates in the two pipes.
Answer: 0.707
8–123 A pipeline that transports oil at 40°C at a rate
of 3 m
3
/s branches out into two parallel pipes
made of commercial steel that reconnect downstream. Pipe A
is 500 m long and has a diameter of 30 cm while pipe B is
800 m long and has a diameter of 45 cm. The minor losses
are considered to be negligible. Determine the flow rate
through each of the parallel pipes.
60 psig
50 ft
20 gpm
Water
main
FIGURE P8–125E
8–124 Repeat Prob. 8–123 for hot-water flow of a district
heating system at 100°C.
8–125E A water fountain is to be installed at a remote loca-
tion by attaching a cast iron pipe directly to a water main
through which water is flowing at 70°F and 60 psig. The
entrance to the pipe is sharp-edged, and the 50-ft-long piping
system involves three 90° miter bends without vanes, a fully
open gate valve, and an angle valve with a loss coefficient of
5 when fully open. If the system is to provide water at a rate
of 20 gal/min and the elevation difference between the pipe
and the fountain is negligible, determine the minimum diam-
eter of the piping system.
Answer: 0.76 in
cen72367_ch08.qxd 11/4/04 7:14 PM Page 396
CHAPTER 8
397
8–126E Repeat Prob. 8–125E for plastic pipes.
8–127 In a hydroelectric power plant, water at 20°C is sup-
plied to the turbine at a rate of 0.8 m
3
/s through a 200-m-
long, 0.35-m-diameter cast iron pipe. The elevation differ-
ence between the free surface of the reservoir and the turbine
discharge is 70 m, and the combined turbine–generator effi-
ciency is 84 percent. Disregarding the minor losses because
of the large length-to-diameter ratio, determine the electric
power output of this plant.
8–128 In Prob. 8–127, the pipe diameter is tripled in order
to reduce the pipe losses. Determine the percent increase in
the net power output as a result of this modification.
8–129E The drinking water needs of an office are met by
large water bottles. One end of a 0.35-in-diameter, 6-ft-long
plastic hose is inserted into the bottle placed on a high stand,
while the other end with an on/off valve is maintained 3 ft
below the bottom of the bottle. If the water level in the bottle
is 1 ft when it is full, determine how long it will take to fill
an 8-oz glass ( 0.00835 ft
3
) (a) when the bottle is first
opened and (b) when the bottle is almost empty. Take the
total minor loss coefficient, including the on/off valve, to be
2.8 when it is fully open. Assume the water temperature to be
the same as the room temperature of 70°F.
Answers: (a) 2.4 s,
(
b) 2.8 s
8–133 The water at 20°C in a 10-m-diameter, 2-m-high
aboveground swimming pool is to be emptied by unplugging
a 3-cm-diameter, 25-m-long horizontal plastic pipe attached
to the bottom of the pool. Determine the initial rate of dis-
charge of water through the pipe and the time it will take to
empty the swimming pool completely assuming the entrance
to the pipe is well-rounded with negligible loss. Take the fric-
tion factor of the pipe to be 0.022. Using the initial discharge
velocity, check if this is a reasonable value for the friction
factor.
Answers: 1.01 L/s, 86.7 h
3 f
t
1 f
t
6 ft
0.35 in
FIGURE P8–129E
8–130E Reconsider Prob. 8–129E. Using EES (or
other) software, investigate the effect of the
hose diameter on the time required to fill a glass when the
bottle is full. Let the diameter vary from 0.2 to 2 in, in incre-
ments of 0.2 in. Tabulate and plot the results.
8–131E Reconsider Prob. 8–129E. The office worker who
set up the siphoning system purchased a 12-ft-long reel of the
plastic tube and wanted to use the whole thing to avoid cut-
ting it in pieces, thinking that it is the elevation difference
that makes siphoning work, and the length of the tube is not
D
2
D
1
V
1
= 10 m/s
FIGURE P8–132
2 m
Swimming
pool
10 m
25 m
3 cm
FIGURE P8–133
8–134
Reconsider Prob. 8–133. Using EES (or other)
software, investigate the effect of the discharge
pipe diameter on the time required to empty the pool com-
pletely. Let the diameter vary from 1 to 10 cm, in increments
of 1 cm. Tabulate and plot the results.
8–135 Repeat Prob. 8–133 for a sharp-edged entrance to
the pipe with K
L
0.5. Is this “minor loss” truly “minor” or
not?
important. So he used the entire 12-ft-long tube. Assuming
the turns or constrictions in the tube are not significant (being
very optimistic) and the same elevation is maintained, deter-
mine the time it takes to fill a glass of water for both cases.
8–132 A circular water pipe has an abrupt expansion from
diameter D
1
15 cm to D
2
20 cm. The pressure and the
average water velocity in the smaller pipe are P
1
120 kPa
and 10 m/s, respectively, and the flow is turbulent. By apply-
ing the continuity, momentum, and energy equations and dis-
regarding the effects of the kinetic energy and momentum-
flux correction factors, show that the loss coefficient for
sudden expansion is K
L
(1 D
1
2
/D
2
2
)
2
, and calculate K
L
and P
2
for the given case.
cen72367_ch08.qxd 11/4/04 7:14 PM Page 397
8–136 A system that consists of two interconnected cylindri-
cal tanks with D
1
30 cm and D
2
12 cm is to be used to
determine the discharge coefficient of a short D
0
5 mm
diameter orifice. At the beginning (t 0 s), the fluid heights
in the tanks are h
1
50 cm and h
2
15 cm, as shown in Fig.
P8–136. If it takes 170 s for the fluid levels in the two tanks to
equalize and the flow to stop, determine the discharge coeffi-
cient of the orifice. Disregard any other losses associated with
this flow.
398
FLUID MECHANICS
garding entrance effects and velocity heads, obtain a relation
for the variation of fluid depth in the tank with time.
8–138 A student is to determine the kinematic viscosity of
an oil using the system shown in Prob. 8–137. The initial
fluid height in the tank is H 40 cm, the tube diameter is d
6 mm, the tube length is L 0.65 m, and the tank diame-
ter is D 0.63 m. The student observes that it takes 2842 s
for the fluid level in the tank to drop to 36 cm. Find the fluid
viscosity.
Design and Essay Problems
8–139 Electronic boxes such as computers are commonly
cooled by a fan. Write an essay on forced air cooling of elec-
tronic boxes and on the selection of the fan for electronic
devices.
8–140 Design an experiment to measure the viscosity of
liquids using a vertical funnel with a cylindrical reservoir of
height h and a narrow flow section of diameter D and length
L. Making appropriate assumptions, obtain a relation for vis-
cosity in terms of easily measurable quantities such as den-
sity and volume flow rate. Is there a need for the use of a cor-
rection factor?
8–141 A pump is to be selected for a waterfall in a garden.
The water collects in a pond at the bottom, and the elevation
difference between the free surface of the pond and the loca-
tion where the water is discharged is 3 m. The flow rate of
water is to be at least 8 L/s. Select an appropriate motor–
pump unit for this job and identify three manufacturers with
product model numbers and prices. Make a selection and
explain why you selected that particular product. Also esti-
mate the cost of annual power consumption of this unit
assuming continuous operation.
8–142 During a camping trip you notice that water is dis-
charged from a high reservoir to a stream in the valley
through a 30-cm-diameter plastic pipe. The elevation differ-
ence between the free surface of the reservoir and the stream
is 70 m. You conceive the idea of generating power from this
water. Design a power plant that will produce the most power
from this resource. Also, investigate the effect of power gen-
eration on the discharge rate of water. What discharge rate
will maximize the power production?
Orifice
h
1
h
h
2
Tank 2Tank 1
FIGURE P8–136
D
d
L
H
Discharge
tube
FIGURE P8–137
8–137 A highly viscous liquid discharges from a large con-
tainer through a small-diameter tube in laminar flow. Disre-
cen72367_ch08.qxd 11/4/04 7:14 PM Page 398
