B48BB

Process Engineering A
Fluid Mechanics

Topic 3
Frictional Pressure Loss
&
Flow Measurement

June 2014 v1 © Heriot-Watt University 1

 Learning Objectives
After studying this Fluid Mechanics topic you should be able to
complete the following tasks:
• Calculate the Reynolds Number and use it to calculate
friction factors and pressure losses for both laminar and
turbulent flow.
• Distinguish between different friction factors and use
friction factor charts.
• Distinguish between smooth and rough pipes, calculate
relative roughness and use empirical correlations to find
friction factors.
• Calculate pressure loss, for both laminar and turbulent flow,
from friction factor information either obtained from charts
or correlations and convert to “head” loss.

June 2014 v1 © Heriot-Watt University 2

 • Derive expression for the hydraulic mean diameter of non-
circular ducts and other conduit geometries.
• Derive equations for the pressure loss in bends and fittings
and apply these equations to solve practical problems
involving realistic pipework layouts.
• Derive equation for the measurement of flow through pitot
tubes, orifice plates, venturi meters and rotameters.
• Solve problems associated with these different flow
measuring devices.
• Describe other flow measuring devices such as turbine flow
meters and positive displacement meters.

June 2014 v1 © Heriot-Watt University 3

 3.1 Introduction
When designing pipeline systems there is a balance between
capital costs and running costs:
• If the pipe diameter is too large, in relation to the flow
passing through it, then capital costs will be high.
• If pipe diameter is too small, in relation to the flow passing
through it, then running costs will be high.
• Designing a pipeline system involves striking a balance
between these two cases – an effort must be made to
optimise the design and to minimise the total cost.

However, in order to estimate the running costs an engineer

needs to be able to estimate the total pressure change from
one end of a pipeline to the other.

June 2014 v1 © Heriot-Watt University 4

 Consider the situation given below which might be the layout
of ain typical pipe in a chemical plant or refinery:
a chemical plant or refinery:

• The engineer needs to be able to calculate all the pressure

changes as the fluid flows through the pipework.
• The engineer needs to be able to select a suitable pump,
which will work in conjunction with the pipework.
• The engineer needs to be able to estimate the running
costs of the system and meet energy saving targets.
This topic will answer the first objective and the next topic will
answer the last two objectives.
June 2014 v1 © Heriot-Watt University 5
 3.2 Pressure Loss Through Straight Pipe
Pressure loss in a pipeline arises because of a number of
factors, such as straight pipe, bends, valves and other fittings.
• However, the first consideration is the pressure loss due to
the “straight pipe” itself.

3.2.1 Laminar Flow

For laminar flow the following equation was derived
analytically :
 Pf   d 4  128QL
Q     >>> ΔPf 
 L   128  d 4

128QL
 ΔPf 
4d 2 (d 2 / 4)

June 2014 v1 © Heriot-Watt University 6

 However, the volumetric flow rate divided by the pipe cross-
sectional area is the average velocity u (m/s), so that
32uL
ΔPf 
d2
• This is the well-known Poiseuille-Hagen equation and
enables the frictional pressure loss to be found in straight
pipe; however, flow must be laminar, i.e. Re < 2000
• For laminar flow, only the pipe diameter is needed – the
roughness of the pipe is not a factor.
• The Poiseuille-Hagen equation was derived analytically and
no empirical factors are needed to find the frictional
pressure loss.
For turbulent flow no such analytical solution is available and
an empirical approach must be used.
June 2014 v1 © Heriot-Watt University 7
 3.2.2 Turbulent Flow
The relationship below links the frictional pressure loss to the
wall shear stress and pipe parameters. It is perfectly general
and applies equally to:
 Either laminar or turbulent flow.
 Either Newtonian or non-Newtonian fluids.
4 w L
Pf 
d
• Multiply the RHS of this expression, above and below, by
the factor u 2
/ 2 (N/m2), to get

  w   L  u 2
Pf 4 2   
 u / 2   d  2

June 2014 v1 © Heriot-Watt University 8

 • Re-arrange this as follows:

 w   L  u 2
Pf 8 2    ………………………………(A)
 u  d  2
• Experiments by Stanton and Pannell found that the
dimensionless group  w / u  depended on two other
2

groups, also dimensionless, as follows:

 w   du   
 2  function of   and   ……………(B)
 u     d
• Where,  (m) is the absolute roughness and  / d  is the
relative roughness.

Given a chart/correlation for (B) then, once  w / u is found
2

from (B), the frictional pressure loss may be found from (A).

June 2014 v1 © Heriot-Watt University 9

 In many texts, equation (A) is written as follows:
L
   u 2
w
Pf 8   Where >>>  2
d 2 u
• The units of Pf (N/m2) come from the units of  (kg/m3)
and the units of u (m/s) – all other terms are dimensionless.
• The symbol  is used to denote the “  - friction factor”.

Thus, in order to calculate the frictional pressure drop, proceed

as follows:
1. First a “  - friction factor” chart or correlation is
needed; correlations better suit Mathcad or Excel.
2. Then calculate the Reynolds Number Re and  / d .
3. Then find  from the chart or the equation.
4. Finally calculate Pf from equation at the top.

June 2014 v1 © Heriot-Watt University 10

 Care is needed because other friction factors are used. Thus, it
is important to properly identify the friction factor chart, so
that the correct pressure drop equation is used.
• The Fanning friction factor f , also dimensionless, was
defined as follows:
 w 
f 2 2 2 
 u 
• If the Fanning friction factor is used, then the pressure drop
equation must be modified to give the same overall result
 L  u
2
Pf 4 f  
d 2
The procedure is the same and a nearly identical result should
be obtained using either friction factor.
June 2014 v1 © Heriot-Watt University 11
 There is a third friction factor. The Moody friction factor f . It
is also dimensionless and was defined as follows:
 w 

f 8 8 2 
 u 
• If the Moody friction factor is used, then the pressure drop
equation must again be modified to give the same overall
result
 L  u
2
Pf  f  
d 2
• Given consistent charts and equations, all three friction
factors should produce the more or less the same pressure
drop result.
To avoid confusion, a better approach is outlined on next slide.
June 2014 v1 © Heriot-Watt University 12
 Look up whichever friction factor chart is available – the
recommended chart is the “ - friction factor” given in
Coulson & Richardson, Volume 1.
• If different from C & R friction factor convert to a “  -
friction factor” using equation below:
f f
 
2 8
• Then find the frictional pressure loss using only the equation
below
 L  u Use Conversion >>> Pf gh f
2
Pf 8  
d 2
• Alternatively, if the friction head loss h f (m) is needed, then
2
 L u
h f 8  
 d  2g
June 2014 v1 © Heriot-Watt University 13
 3.2.3 Friction Factor Charts
The  - friction factor chart supplied with Coulson &
Richardson can be divided into a number of regions:
1. Re < 2000: i.e. the laminar flow region. It was shown
analytically that the Poiseuille-Hagen equation may be
used to estimate frictional pressure loss.
32uL
ΔP f 
d2
It should be possible to put this in the more general form
used for turbulent flow pressure loss, shown below:
 L  u
2
Pf 8  
d 2
If the Poiseuille-Hagen equation takes this form, then work
out and an expression for laminar flow  - friction factor.
June 2014 v1 © Heriot-Watt University 14
 Re-arrange the Poiseuille-Hagen equation as follows:
32uL  8    L  u 2
ΔP f  >>> ΔP f  8   
d2  du   d  2
In the laminar flow region, the  - friction factor must
clearly be given by
8 Notice this arrangement
 of variables is……8/Re
Re
This is an analytical solution and the friction factor chart is
not really needed (just solve the above equation).
Nevertheless, the laminar friction factor is always plotted.
From the above expression, the  - friction factor depends
only on the Reynolds Number and not on the relative
roughness  / d  - thus there is only one line in this area.

June 2014 v1 © Heriot-Watt University 15

 Summary: frictional pressure loss for any Newtonian fluid
under laminar flow conditions (Re < 2000);
• Use either the Poiseuille-Hagen equation, below, in which
case no friction factor is needed.
32uL
ΔP f 
d2
• Or use the more general equation below (used also for
turbulent flow) in which case a friction factor will be needed.
 L  u
2
Pf 8  
d 2
• And the friction factor (laminar flow) is given by:
8
 This may also be estimated, less
Re accurately, from a friction factor chart.
Very similar answer will be obtained either way.
June 2014 v1 © Heriot-Watt University 16
 2. 2000 < Re < 3000: i.e. critical region corresponding to the
unstable “critical zone” where flow can either be laminar
or turbulent and may indeed switch between the two.
• Note the laminar flow
region.
• Note the critical region.
• Onset of turbulence
occurs Re > 2000
• This is, therefore, a ф
Friction Factor Chart.

Different charts may be identified by examining laminar flow region:

Fanning Moody
 - Friction Factor f - Friction Factor f  -Friction Factor
Laminar Flow
Analytical 8 16 64
 f  f 
Solution: Re Re Re

June 2014 v1 © Heriot-Watt University 17

 3. Re > 3000: this is turbulent flow region where friction
factor depend on both Re and  / d . However, for smooth
pipes the friction factor is independent of  / d.
Blasius developed the following two correlations for
smooth pipes, where the  - friction factor depends only on
the Reynolds Number Re:
A. For Reynolds Number between 2.5 x 103 and 105;
 0.0396 Re  0.25
B. For Reynolds Number between 2.5 x 103 and 107;
  0.5 2.5 ln(Re  0.5 )  0.3
Turbulent flow is usually only reliable after Re > 4000, but
turbulent flow correlations are being used here at lower Re.
This is to maximise Pf where there is uncertainty.
June 2014 v1 © Heriot-Watt University 18
 4. Re > 3000: in this region pipes are not smooth, but the
operating point on the chart is to the left of the dashed line.
Here the friction factor  depends on both Re and  / d .
There are several correlations that may be used but one that
is widely quoted (C & R, Volume 1, 1999) is given by
 0.5 
  2.5 ln(0.27  0.885 Re  1   0.5 )
d
The flow is turbulent, but there is still a laminar sub-layer
adjacent to the pipe wall. If the laminar sub-layer thickness
exceeds absolute roughness, the pipe is considered “smooth”.
If the sub-layer thickness is less than absolute roughness, the
pipe is not smooth and additional form drag is exerted by the
turbulent core on the roughness (protruding through sub-layer).

June 2014 v1 © Heriot-Watt University 19

 5. If Re > 3000 and the point lies somewhere to the right of
the dotted line.

The friction factor  then becomes independent of the Re

and only depends on  / d. For this region a popular
correlation is given by


  0.5 3.2  2.5 ln
d

Notice that the friction factor chart may be used in all regions,
including laminar flow region.

However, when working with Mathcad or Excel correlations

are much more convenient.

June 2014 v1 © Heriot-Watt University 20

 3.2.4 Typical Absolute Roughness Values
Internal roughness on the inside of a pipe can be thought of as
a series of indentations – the size of these indentations
depends on the material of construction:
Pipeline Absolute Roughness
Material
 (mm)

Drawn Tube
- Copper 0.0015
- Aluminium
- PVC

Commercial steel pipe 0.046

Cast iron pipe 0.26

Galvanised iron 0.15

Concrete 0.3 – 3

Find the absolute roughness, based on selected pipe. Calculate

the relative roughness  / d >>>> find the  - friction factor.
June 2014 v1 © Heriot-Watt University 21
 3.2.5 Hydraulic Mean Diameter
For a circular pipe, the inside pipe diameter d is the single
“characteristic dimension” of the pipe:
• However, what happens if the duct is square? – there is
no diameter!
• And, what happens if the fluid is flowing in the annulus
between two concentric pipes? – there are now two
diameters!
• The solution is to define “the hydraulic mean diameter”,
symbol d h , (or equivalent diameter) as follows:
4 Flow cross  sectional area
dh 
Wetted Perimeter
4 AF
 dh 
PW

June 2014 v1 © Heriot-Watt University 22

 For a circular pipe of diameter d the hydraulic mean diameter
dh
is 4 AF 4 d 2 / 4
dh   d
PW d
• As expected, for a circular pipe the hydraulic mean
diameter is equal to the actual diameter.
• For a rectangular shaped duct of height H (m) and
breadth B (m), the hydraulic mean diameter is simply:
4 BH
dh 
2B  H  This concept is also used in Heat
Transfer, where flow might be through
2 BH an annulus or over a tube bundle.
 dh 
B  H 

June 2014 v1 © Heriot-Watt University 23

 Example 3.2

Problem:
Water flows through a perfectly insulated, horizontal, smooth
rectangular duct that discharges to the atmosphere. The duct
has a cross section of 100 x125 mm and a total length of 30 m.

The fluid velocity is 6 m/s. At some distance x from the exit of

the duct, the static head is 3.5 m of H2O.

Determine the gauge pressure of the water at the duct inlet,

the distance x and the temperature rise of the water across
the entire length of duct.

Take  1000 kg/m3,  0.001 N s/m2 and C P 4.2 kJ/kg K.

June 2014 v1 © Heriot-Watt University 24

Example 3.2

Solution:
Start by drawing a sketch of the equipment – this helps to
visualise the problem, after which a solution strategy should
become evident:
solution strategy should become evident:

P1 P2 1 Atm
Px
L 30 m

 u 6 m/s x (m)

hStatic 3.5 m H 2 O

For instance if the frictional pressure drop from “1” to “2” can
be found then, by adding this drop to the exit pressure, the
inlet pressure may be calculated – distance x then follows.
June 2014 v1 © Heriot-Watt University 25
 • Write down the hydraulic mean diameter for this square
duct and solve using the information given:
4 BH 4 0.10.125
dh   0.1111 m
2B  H  20.1  0.125

• Now calculate the Re based on this hydraulic mean

diameter:
d h u 0.1111 6 1000
Re   666700
 1 10 3

Since the duct is “smooth”, and based on the Re calculated

above, choose a suitable correlation and then solve:
 0.0396 Re  0.25 0.0396 666700  0.25

 0.001386 Always check the friction factor chart to

see if the value looks reasonable.

June 2014 v1 © Heriot-Watt University 26

 • Write down the frictional pressure loss equation in terms of
friction factor and solve for pressure drop from “1” to “2”:
 L  u 2
Pf 8  
 dh  2
30 1000 6 2
 Pf 8 0.001386  
0.1111 2
 Pf 53893.1 N/m 2 or 53.893 kN/m2
• This is the pressure drop from “1” to “2”. If pressure at “2” is
known, calculate the absolute pressure at “1”:

P1  P2  Pf

 P1 101.325  53.893 155.22 kN/m 2 absolute

June 2014 v1 © Heriot-Watt University 27

 • However, the gauge pressure at “1” was asked for in the
question:
 P1 53.89 kN/m2 gauge
• At point x the static pressure head is known to be 3.5 m
H2O, convert this to (kN/m2):
Px gh 1000 9.81 3.5
 Px 34.34 kN/m2 gauge
• Calculate the pressure drop per unit length of duct:
Pf 53.89
 1.796 kN/m3
L 30
• Calculate the distance of point x to the end of the duct.
34.34
Lx  19.12 m
1.796
June 2014 v1 © Heriot-Watt University 28
 • Write down the relationship between the energy loss due to
friction and the pressure loss due to friction:
Pf
Ef  ……(J/kg)

• Write down the relationship between the energy loss due to
friction and the change in temperature of the fluid:
E f Δ Internal Energy C P(T2  T1 ) ……(J/kg)
• Equate the two expressions and solve for temperature
difference:
Pf /  53893 / 1000
(T2  T1 )   ……(J/kg)(K kg/J)
CP 4200
The frictional pressure loss ends up
(T2  T1 ) 0.0128 K as heat and fluid is very slightly
hotter at the exit end of the duct.

June 2014 v1 © Heriot-Watt University 29

 3.3 Pressure Loss Through Pipe Fittings
In addition to “straight pipe” fluids also pass through pipeline
components that can either change cross sectional area, or flow
direction – see typical “enlargement” below:

• Such fittings produce eddies that

can dissipate a significant
quantity of energy as “form
drag”.
• Eddy-based losses usually occur in pipeline components such
as abrupt enlargements, abrupt contractions, 90o elbows,
tees, orifice plates, valves (open or partially closed), etc.
Theoretical examination of above enlargement can suggest how
empirical methods may be developed for other pipeline
components – these components are often called “fittings”.
June 2014 v1 © Heriot-Watt University 30
 It is very important to be able to predict the pressure losses
that occur in a pipe system:
• These losses come about in two ways: “skin friction” losses;
and “form friction” losses.
• Skin friction losses depend on the fluid viscosity and arise
because the fluid is continually being sheared. The shear
rate induces a shear stress at the pipe wall.
• This wall shear stress, gives rise to a skin friction force and
this in turn leads to pressure and flowing energy losses.
These losses are associated with straight pipe.
• Form friction losses depend on the shape and geometry of
the pipework. For non-streamline fittings the fluid will
separate from the wall – producing “turbulent eddies”.
This is referred to as “form friction losses” and these losses are
associated with various pipe fittings.
June 2014 v1 © Heriot-Watt University 31
 Thus for any fitting the energy loss E f (J/kg) through the
fitting can be experimentally correlated by the expression:

u12 This equation is suggested by analysis of

E f k a sudden expansion. The constant “k” is
2 then found experimentally for all other
fittings – thus approach is empirical.

• The frictional energy loss may be converted to a frictional

pressure loss as follows:
Pf
u12
 k
 2
The frictional pressure loss caused by
u12 form drag through any fitting can be
 Pf k calculated using this expression if “k” is
2 listed for the fitting under consideration.

June 2014 v1 © Heriot-Watt University 32

 3.3.1 Lost Velocity Heads Method
It has been shown how the energy loss E f (J/kg) through the
fitting can be experimentally correlated by the expression:
u2
E f k
2
• This was converted to a frictional pressure loss on the
previous slide – alternatively it may be converted to
frictional head loss, just divide the above by g
u2
h f k
2g
Thus, the dimensionless loss coefficient k may be thought of as
2
a “number of lost velocity heads” u / 2 g .

June 2014 v1 © Heriot-Watt University 33

 A table of experimentally determined lost velocity heads, for
different pipework fittings is given below:
Fitting Number of lost velocity

Std. 45° elbows

heads
0.35
k • More accurate lost
90° elbow standard radius 0.6 - 0.8 velocity head
90° square elbow 1.5
methods are now
Entry from leg T-piece 1.2
Entry into leg T-piece 1.8 available – see
Sudden Reduction (Tank Outlet) 0.5 notes.
Sudden Expansion (Tank Inlet) 1.0
Unions and Couplings 0.04
Globe valve fully open 6
• However, these
Gate valve fully open 0.15
newer methods are
Gate valve 75% open 1
Gate valve 50% open 4 more technical and
Gate valve 25% open 16 harder to use.
Ball valve (100% open) 0.4
Plug valve open 0.4

June 2014 v1 © Heriot-Watt University 34

 3.3.2 Equivalent Length Method
• Looking at the “straight pipe” pressure loss expression
 L  u
2
Pf 8  
d 2
• Consider a bend or fitting as an “equivalent length” Le of
(m) of straight pipe. For a single “fitting” the pressure loss
across it is given as
 Le  u 2
Pb 8  
 d  2
• Thus, the pressure loss Pb for all bends and fittings, with
total equivalent length  Le (m), is given by

  Le  u 2
Pb 8  
 d  2
 
June 2014 v1 © Heriot-Watt University 35
 The total pressure loss PL of all bends and fittings, plus
straight pipe, is found by adding the above two pressure losses
together as follows:
 L   Le  u 2
PL 8  
 d  2
 
• However the equivalent lengths are actually listed as an
equivalent number of pipe diameters  ( Le / d )

 L  Le  u 2
 PL 8   
 2
 d d 
Thus pipe diameters of straight pipe is added to equivalent
pipe diameters of all bends and fittings to find the total
pressure loss.
June 2014 v1 © Heriot-Watt University 36
Typical values of dimensionless pipe diameters, for various
fittings, are listed in the table below:
Fitting Number of Pipe
Diameters L / d
e
• The equivalent length
approach was
45° elbows 15
90° elbow standard radius 30-40
90° square elbow 75 superseded by more
Entry from leg T-piece
Entry into leg T-piece
60
90
accurate lost velocity
Sudden Reduction (Tank Outlet) 25 head methods.
Sudden Expansion (Tank Inlet) 50
Unions and Couplings 2
Globe valve fully open 450
• However, the
Gate valve fully open 7.5 equivalent length
Gate valve 75% open
Gate valve 50% open
40
200
approach is easier to
Gate valve 25% open 800 use and still produces
Ball valve (100% open)
Plug valve open
18
18
reasonable first-order
estimates.

June 2014 v1 © Heriot-Watt University 37

 Example 3.3

Problem:

A pipeline connecting two tanks contains 3 standard elbows, a

plug valve that is normally fully open and two gate valves that
are normally 75% open.

The line is commercial steel pipe, 30 mm internal diameter.

Length of the pipeline is 25 m. If the fluid has a density 990
kg/m3 and a viscosity 0.85 cP, calculate the overall pressure
drop for a mass flowrate 4000 kg/hr.

June 2014 v1 © Heriot-Watt University 38

 Example 3.3

Solution:
• Calculate the cross-sectional area of the pipe:
 
A  d 2  0.030 0.000707 m 2
4 4
• Knowing the mass flow rate and the density calculate the
volumetric flow rate and the average velocity:
m
4000 3  1 hr 
Q  m /hr  
 990  3600 s 
Q 0.00112 m 3 /s

Q 0.00112
u   1.58 m/s
u 0.000707
June 2014 v1 © Heriot-Watt University 39
 Example 3.3

Solution:
• Now calculate the Reynolds Number, Re

du 0.030 1.58 990

Re   55207
 0.85 10 3

• The relative roughness for this type of pipe can be found:

 0.046 10  3
 0.0015
d 0.030
• Flow is turbulent (Re > 4000) and look up  - friction factor
chart in C & R (Vol. 1)
 0.0025

June 2014 v1 © Heriot-Watt University 40

 Example 3.3, Solution:
• Use the simpler equivalent length method which produces
reasonable first-order estimates and is not as technical as
newer lost velocity head methods – tabulate  ( Le / d ) :
Fitting Type Number No. Pipe Diameters Total Equivalent Pipe
Diameters

Elbow 3 35 3x35=105
Plug valve 1 18 18
Gate valve 2 40 =2x40=80
Tank 1 outlet 1 25 25
Tank 2 inlet 1 50 50

Total
 (L e / d)  278

 L  Le  u 2  25  990 1. 58 2
PL 8   
 8 0.0025   278  
d d  2  0.030  2

 PL 27466 N/m 2 or 27.5 kN/m2

June 2014 v1 © Heriot-Watt University 41

 3.4 Flow Measurement
In many situations, such as the control of process plant, process
parameters need to be monitored to ensure that the plant is
operating at optimum conditions:
• Flow rate control is one such important process parameter.
Before flow can be controlled it must be measured.
• The velocity v (m/s) at any radial position varies from zero at
the wall, to a maximum vmax (m/s) at the centreline of the
pipe – the average velocity is u (m/s).

3.4.1 Pitot Static Tube

• The Pitot tube actually measures the point velocity v at any
point across the diameter of a pipe. If the overall flow rate Q
(m3/s) is required, then a sequence of readings is needed.

June 2014 v1 © Heriot-Watt University 42

 The pitot tube and a simple U-tube manometer are connected
as shown below:
• The LH leg of the manometer connects to the pitot tube.
• The RH leg of the manometer connects to pipe wall:
must fit perpendicularly to the pipe wall:

• The Pitot “impact

Pitot impact
Static pressure
connection
connection” must be
connection
angled to point
Points “1” “2”
Pitot h
directly into the flow.
tube

Simple m
U-tube
manometer

• The “static pressure” tapping is connected to the LH limb of

the U-tube manometer – it must be connected
perpendicular to the pipe wall.
June 2014 v1 © Heriot-Watt University 43
 Consider the fluid streamline that comes to rest at the pitot
impact connection – point “1” is upstream and point “2” is
downstream at the pitot impact connection itself:
• Apply Bernoulli’s equation between points “1” and “2” on
this streamline. The expression below is written in terms of
heads and ignores frictional head loss which is negligible:
P1 1 2 P2 1 2
 u1  Z 1   u2  Z 2
g 2 g g 2 g
• The pitot tube is horizontal and, since fluid is at rest at “2”,
then u2 0 . Bernoulli’s equation reduces to
P1 1 2 P2
 u1 
g 2 g g
The pressure experienced at the “impact connection”, point
“2” is the full upstream “stagnation head” of the streamline.
June 2014 v1 © Heriot-Watt University 44
 Re-arrange the previous equation in favour of v1
2P2  P1  2P
 v1  >>>  v1 
 
• The static pressure connection measures the upstream
static head of the streamline P1 / g (this does not change
substantially over the short distance under consideration).
• The velocity v1 is the free stream velocity (upstream
velocity unaffected by pitot tube.)
• To solve the above equation, the pressure drop between
points “1” and “2” must be measured.
This may be accomplished using a manometer. However,
modern digital differential pressure instruments are often
more convenient to use.

June 2014 v1 © Heriot-Watt University 45

 3.4.2 Orifice Plate Meters
Differential pressure (DP) flow meters were developed many
years ago by Pitot and Venturi. Early in the 20th century,
Weymouth developed the sharp edged orifice meter:
• The orifice meter has become a very popular flow
measuring device.
• The area of the orifice is constant.
• The basic principle of all variable pressure flow meters is
that flowrate may be found by measuring the drop in
pressure between two points “1” and “2”.
• Bernoulli’s equation is then applied between these two
points.
They are sometimes called “constant area-variable pressure
drop” devices.
June 2014 v1 © Heriot-Watt University 46
 The restricting orifice causes the fluid to accelerate and the
velocity through the orifice will then be higher than the
velocity upstream of the plate.
will then be higher than the velocity well upstream of the device.

Point “1” Point “2”

Orifice
Plate “O”

• Point “1” is well upstream of the orifice plate, where the

flow streamlines are undisturbed.
• Point “2” varies depending on the exact design of the plate,
but it is often located just downstream of the plate.
Due to fluid inertia, the fluid continues to contract into a
smaller area just downstream of the plate – this is known as
the “vena contracta” and point “2” is sometimes located here.

June 2014 v1 © Heriot-Watt University 47

 Apply Bernoulli’s equation between points “1” and “2”. The
expression below is written in terms of heads and ignores the
head loss due to friction:
P1 1 2 P2 1 2
 u1  Z 1   u2  Z 2
g 2 g g 2 g
• If the device is horizontal the hydrostatic heads and g will
cancel leading to
P1 1 2 P2 1 2
 u1   u 2
 2  2 Energy is changed from one
energy storage form (static
1 2 1 2 P1 P2 head) into another
 u 2  u1   (dynamic head).
2 2  
• This simply states that an increase in kinetic energy (J/kg)
must be accompanied by a drop in pressure energy (J/kg).
June 2014 v1 © Heriot-Watt University 48
 Thus, it is possible to write:
2
2
u u 
2
2
1 P1  P2 

• Apply the continuity equation between point “1” and point
“2”, assuming constant density, leads to
2
A1  d1 
u 2 u1 u1  
A2  d2 
• Substitute the latter into the former leads to
2
A 2
u1 2  u1  P1  P2 
2 1 2

A2 
2P1  P2  P is directly proportional to
 u12  >>>
 A12  “average velocity squared”.
  2  1
 A2 

June 2014 v1 © Heriot-Watt University 49

 The volumetric flow rate Q (m3/s) is the average velocity u
(m/s) multiplied by the cross-sectional area A (m2) – all at
point “1”:
2P1  P2 
u1 A1  A1
 A12 
  2  1
 A2 

2P1  P2 
 Qideal  A1
 A12 
  2  1

 A2 

• However the vena-contracta area A2 is less than the orifice

area AO . Apply the coefficient of contraction C c  A2 / Ao .

June 2014 v1 © Heriot-Watt University 50

 2P1  P2 
 Qideal  A1
 A12 
  2 2  1
 C c Ao 
• This is the ideal volumetric flow rate, where there are no
energy losses. In real systems there will always be some
energy loss due to wall friction and form drag.
• To correct for this, introduce the “discharge coefficient” Cd
which corrects for energy losses, the coefficient of
contraction and the KE correction term:
2P1  P2  The Actual volumetric flowrate is
Q C d A1 directly proportional to the square
 A12  root of the overall pressure drop,
  2  1
Cd is needed to correct for non-
 Ao 
ideal flow.
June 2014 v1 © Heriot-Watt University 51
 The discharge coefficient depends on the following factors: the
type of tapping; the ratio of the orifice diameter to the inside
pipe diameter; and the Re.
• A more detailed drawing of a “sharp edge orifice plate” is
since the discharge coefficient will depend on this selection:
shown below:

Notice that a variety of different tapping points are possible.

Each set of tappings is associated with its own “ Cd factor”.

June 2014 v1 © Heriot-Watt University 52

 The relationship between the orifice plate and the vena-
contracta is shown below – turbulent eddies result in a
permanent pressure loss due to form drag

• The mass flow rate m (kg/s) is the volumetric flow rate

Q (m3/s) multiplied by the fluid density  (kg/m3)

2  P1  P2  • Discharge coefficients are typically

m C d A1
 A12  0.65 – 0.70 . See notes for further
 2  1 details and diagram.
A 
 o  • When Re > 104 , then C d 0.62

June 2014 v1 © Heriot-Watt University 53

 3.4.3 Venturi Meters
The principle is the same as before – constrict the flow stream
so that the volumetric flow rate is proportional to the square
root of the pressure drop between two points.
Point “2” • Venturi meters are expensive
to buy and install.
• However, the pressure loss is
lower.
Point “1” “Throat”
• There are a series of two pressure tappings around the
circumference:
• Point “1” is upstream of the “throat” (minimum XSA).
• Point “2” is just beyond the “throat” (vena-contracta).
• The ring of holes at points “1” and “2” ensure that static
pressure at each location is a good average.
June 2014 v1 © Heriot-Watt University 54
 The point of lowest pressure occurs not at the throat itself, but
at a short distance slightly downstream from the throat. This
point is known as the “vena-contracta”.
• The derivation is the same as for the orifice plate
2P1  P2 
Q Cd A1
 A12 
  2  1
 A2 
• In terms of mass flow rate this again becomes
2  P1  P2 
m C d A1
 A12 
 2  1
A 
 2 
The discharge coefficient for a venturi meter is high (around
0.98) and this is due to the streamlined design. A2 is the area
of the throat.
June 2014 v1 © Heriot-Watt University 55
 3.4.4 Linear Meters
With orifice plates and venturi meters the flow is proportional
to the square root of the pressure drop.
In the case of linear meters, the flow varies linearly with the
measurement. Meters of this type include the following:
• Variable area meters – rotameters.
• Turbine meters.
• Displacement meters.
Typically such meters have a large flow range (often around
one-tenth of full flow, to full flow itself) – this feature is called
the “rangeability” of the meter.

June 2014 v1 © Heriot-Watt University 56

 Variable Area Meters
Variable area meters, usually called rotameters, are linear; the
flow rate is linearly dependent on the “float” position as
measured against a vertical scale:
• The upward flow of fluid exerts a drag
on a moveable “float” .
• The float is made of dense material and
does not really “float” (like wood).
• The larger the flow, the larger the upward
drag exerted by the fluid on the” float”.
This upward drag suspends the “float”.

• The float spins slowly inside a vertical tube. The tube gets
wider as the float rises higher up the scale.
• The float is at equilibrium when the upward drag force is
equal to the downward gravitational force.
June 2014 v1 © Heriot-Watt University 57
 By tapering the tube, higher flows and drag forces are needed
to lift the “float” higher up the scale.
• Thus rotameters are known as “constant pressure drop-
variable area” devices.
• They are “linear” because as flow doubles the position of
the float against vertical scale doubles.

With the rotameter shown on the left, the float is

visible through the variable area tube.
Most of these devices are, therefore, manufactured
from glass or a clear plastic; this restricts their use
to low-pressure applications.

For high pressures, metal tubes are used and the float position
is linked to an external scale via a magnetic coupling.
June 2014 v1 © Heriot-Watt University 58
 The rotameter can be imagined as a variable area orifice: the
pressure drop is kept constant, while the orifice area increases
linearly with flow;
• Thus, the orifice plate expression may be applied equally to
a rotameter:

2  P 
m C d A1
 A12 
 2  1
A 
 2 

• Where,
A1  Tube XSA at point indicated by float (m2)
A2  Float-to-tube XSA (m2)
P  Constant pressure drop across the float (N/m2)

June 2014 v1 © Heriot-Watt University 59

 At equilibrium a force balance across the float leads to the
following expression:
V f  f   g
P 
Af
• Where,
V f  Volume of the float (m3)
 f  Density of the float (kg/m3)
  Density of the fluid (kg/m3)
A f  The maximum cross sectional area of the float (m2)

• Substitute the latter expression into the former, leads to

2 gV f  f   
m C d A1
 A12 
A f  2  1

 A2 
June 2014 v1 © Heriot-Watt University 60
 At the end of the final lecture class of week 7 there will be time to derive
expressions for the effect of fluid density on a rotameter reading – take
notes from the board and insert them here.

June 2014 v1 © Heriot-Watt University 61

 Rotameter density correction to be inserted here.

June 2014 v1 © Heriot-Watt University 62

 Rotameter density correction to be inserted here.

June 2014 v1 © Heriot-Watt University 63

 Turbine Flow Meters
As the fluid passes through a turbine meter it spins a rotor
which is fixed inside the casing:
• Turbine flow meters can
measure the flow rate of
both liquids and gases.
• The volume flow rate is
proportional to the speed
of the turbine rotor
• A magnetic coupling is used to measure the speed, so that
no mechanical connection is needed between the rotor and
the sensor.
Thus, no seals are required and no corrosive or toxic fluids can
escape – this is a distinct advantage.
June 2014 v1 © Heriot-Watt University 64
However, the rotor is easily damaged by solids and so turbine
flow meters can only be used in applications where the fluid is
clean:
• Accuracies are typically
0.5% of full scale.

• Avoid viscous liquids

which disrupt normal
shear stress relationship.

• They are sensitive to location and require 10 clear pipe

diameters upstream and 5 clear pipe diameters downstream,
to achieve headline accuracy.
Turbine flow meters are relatively expensive to buy, but very
convenient to use once installed.

June 2014 v1 © Heriot-Watt University 65

 Positive Displacement (PD) Meters
Spaces within the PD rotor (the meter may also have a
reciprocating action) fill and discharge in a continuous manner.
Fluid pressure drop drives the displacement rotor:
• Thus, for every rotation of the rotor a known quantity of
fluid must be displaced.
• The fluid flow rate will then be proportional to the speed of
rotor and all that is needed is a mechanism to measure this
speed…..some sort of speed “pick-up” sensor.
• Usually a magnetic coupling device is used and, with no
mechanical linkage, this eliminates the need for seals – the
added benefit is that fluid cannot escape.
PD flow meters may be used with viscous, dirty, corrosive
fluids and require no clear pipe runs upstream or downstream.
June 2014 v1 © Heriot-Watt University 66
 Example 3.4

Problem:
Several standard Chemical Engineering textbooks (Coulson and
Richardson, Vol. 1, 1999) and design guides (Coulson and
Richardson, vol. 6, 1993) use the following expression for the
mass flow rate through an orifice plate.
2  P1  P2 
m C d Ao
 Ao2 
1  2 
 A1 

Where,
Ao  Cross sectional area of the orifice opening (m2)

Starting with Bernoulli’s equation derive the above expression.

June 2014 v1 © Heriot-Watt University 67

 Example 3.4

Solution:
Write down Bernoulli’s equation, but cancel out the head loss
due to friction (this is corrected later by Cd). Also cancel out
the static heads, using assumption that meter is horizontal:
2
u2  u1  P1  P2 
2 2


• Apply the continuity equation between point “1” and point
“2” Assuming constant density, eliminate u1 in favour of u 2
2
A2  d2 
u1 u 2 u 2  
A1  d1 

June 2014 v1 © Heriot-Watt University 68

 Example 3.4

Solution:
• Substitute the latter equation into the former and solve for u2:
2
A 2
u 22  u 22 22  P1  P2 
A1 

2
P1  P2 

 u 22 
 A22 
1  2 
 A 
 1 

2P1  P2 
 u2 
 A22 
  1  2 
 A1 

June 2014 v1 © Heriot-Watt University 69

 Example 3.4, Solution:
• The volumetric flow rate Q (m3/s) is the average velocity
u2 (m/s) multiplied by the cross-sectional area A2 (m2) at
point “2” (the vena-contracta)……write this down:
2P1  P2 
 Qideal  A2
 A22 
  1  2 
 A1 
• However the vena-contracta area A2 is less than the orifice
area Ao . Apply the coefficient of contraction Cc  A2 / Ao to
correct above expression:
2P1  P2 
 Qideal C c Ao
 C c2 Ao2 
  1  2


 A 1 

June 2014 v1 © Heriot-Watt University 70

 • This is ideal flow with no energy losses. In real systems
there will always be some energy loss due to wall
friction and form drag.
• The “discharge coefficient” Cd allows for energy losses and
other corrections. Write down the final form of the actual
volume and mass flow rate expressions:
2P1  P2 
Q Cd Ao
 Ao2 
  1  2 
 A1 
Or,

2  P1  P2 
m C d Ao
 Ao2 
1  2 
 A1 

June 2014 v1 © Heriot-Watt University 71