TOPIC T4: PUMPS AND TURBINES AUTUMN 2022

Objectives

(1) Understand the role of pumps and turbines as energy-conversion devices and use,
appropriately, the terms head, power and efficiency.
(2) Be aware of the main types of pumps and turbines and the distinction between impulse and
reaction turbines and between radial, axial and mixed-flow devices.
(3) Match pump characteristics and system characteristics to determine the duty point.
(4) Calculate characteristics for pumps in series and parallel and use the hydraulic scaling
laws to calculate pump characteristics at different speeds.
(5) Select the type of pump or turbine on the basis of specific speed.
(6) Understand the mechanics of a centrifugal pump and an impulse turbine.
(7) Recognise the problem of cavitation and how it can be avoided.

1. Energy conversion
1.1 Energy transfer in pumps and turbines
1.2 Power
1.3 Efficiency

2. Types of pumps and turbines

2.1 Impulse and reaction turbines
2.2 Positive-displacement and dynamic pumps
2.3 Radial, axial and mixed-flow devices
2.4 Common types of turbine

3. Pump and system characteristics

3.1 Pump characteristics
3.2 System characteristics
3.3 Finding the duty point
3.4 Pumps in parallel and in series

4. Hydraulic scaling
4.1 Dimensional analysis
4.2 Change of speed
4.3 Specific speed

5. Mechanics of rotodynamic devices

5.1 Centrifugal pump
5.2 Pelton wheel

6. Cavitation

Hydraulics 2 T4-1 David Apsley

1. Energy Conversion
total head

1.1 Energy Transfer in Pumps and Turbines

pipeline
PUMP
Pumps and turbines are energy conversion devices:

pumps turn electrical or mechanical energy into fluid energy; total head

turbines turn fluid energy into electrical or mechanical energy.

pipeline

TURBINE
The energy per unit weight is the total head, 𝐻:

𝑝 𝑉2
𝐻= +𝑧+
𝜌𝑔 2𝑔

The first two terms on the RHS comprise the piezometric head. The last term is the dynamic
head.

1.2 Power

Power = rate of conversion of energy.

If a mass 𝑚 is raised through a height 𝐻 it gains energy 𝑚𝑔𝐻. If it does so in time 𝑡 then the
rate of conversion is 𝑚𝑔𝐻/𝑡.

For a fluid in motion the mass flow rate (𝑚/𝑡) is 𝜌𝑄. The rate of conversion to or from fluid
energy when the total head is changed by H is, therefore, 𝜌𝑄 × 𝑔𝐻, or
power = 𝜌𝑔𝑄𝐻

1.3 Efficiency

Efficiency, 𝜂, is given by
(power)out
𝜂=
(power)in
where power refers to the rate of doing “useful” work (i.e. what the device was intended to do.)

(power)𝑜𝑢𝑡
For turbines: 𝜂=
𝜌𝑔𝑄𝐻

𝜌𝑔𝑄𝐻
For pumps: 𝜂=
(power)𝑖𝑛

Hydraulics 2 T4-2 David Apsley

 Example.
A pump lifts water from a large tank at a rate of 30 L s–1. If the input power is 10 kW and
the pump is operating at an efficiency of 40%, find:
(a) the head developed across the pump;
(b) the maximum height to which it can raise water if the delivery pipe is vertical, with
diameter 100 mm and friction factor 𝜆 = 0.015.

Answer: (a) 13.6 m; (b) 12.2 m

Hydraulics 2 T4-3 David Apsley

2. Types of Pumps and Turbines

2.1 Impulse and reaction turbines

In a pump or turbine a change in fluid head

𝑝 𝑉2
𝐻= +𝑧+
𝜌𝑔 2𝑔
may be brought about by a change in pressure or velocity or both.
• An impulse turbine (e.g. Pelton wheel; water wheel) is one where the change in head is
brought about primarily by a change in velocity. This usually involves unconfined free
jets of water (at atmospheric pressure) impinging on moving vanes.
• A reaction turbine (e.g. Francis turbine; Kaplan turbine; windmill) is one where the
change in head is brought about primarily by a change in pressure.

2.2 Positive-Displacement and Dynamic pumps

Positive-displacement pumps operate by a change in volume; energy conversion is intermittent.

Examples in the human body include the heart (diaphragm pump) and the intestines (peristaltic
pump). In a reciprocating pump (e.g. a bicycle pump) fluid is sucked in on one part of the cycle
and expelled (at higher pressure) in another.

In dynamic pumps there is no change in volume and energy conversion is continuous. Most
pumps are rotodynamic devices where fluid energy is exchanged with the mechanical energy
of a rotating element (called a runner in turbines and an impeller in pumps), with a further
conversion to or from electrical energy.

This course focuses entirely on rotary devices.

Note that, for gases, pumps are usually referred to as fans (for low pressures), blowers or
compressors (for high pressures).

2.3 Radial, Axial and Mixed-Flow Devices

The terms radial and axial refer to the change in direction of flow through a rotodynamic device
(pump or turbine):

Radial Axial Mixed

Hydraulics 2 T4-4 David Apsley

 "eye" (intake)

impeller vane
In a centrifugal pump flow enters along the axis
and is expelled radially. (The reverse is true for a
turbine.)

volute

rotation
flow
An axial-flow pump is like a propeller; the
direction of the flow is unchanged after passing
through the device.

guide vanes

blades
outlet

A mixed-flow device is a hybrid device, used

inlet
for intermediate heads. hub

In many cases – notably in pumped-storage power stations – a device can be run as either a
pump or a turbine.
Inward-flow reaction turbine  centrifugal pump (high head / low discharge)
(e.g. Francis turbine)

Propeller turbine  axial-flow pump (low head / high discharge)

(e.g. Kaplan turbine; windmill)

Hydraulics 2 T4-5 David Apsley

2.4 Common Types of Turbine

Pelton wheels are impulse turbines used in hydroelectric

plant where there is a very high head of water.
Typically, 1 to 6 high-velocity jets of water impinge on
buckets mounted around the circumference of a runner.
A modern variant, which can handle a greater water spear valve

flow and is popular in microhydropower is called a bucket

Turgo turbine. jet

Francis turbines are used in many large hydropower

projects (e.g the Hoover Dam), with an efficiency in
excess of 90%. Such moderate- to high-head turbines are
also used in pumped-storage power stations (e.g.
Dinorwig and Ffestiniog in Wales; Foyers in Scotland),
which pump water uphill during periods of low energy
demand and then run the system in reverse to generate
power during the day. This smooths the power demands
on fossil-fuelled and nuclear power stations which are not easily brought in and out of
operation. Francis turbines are like centrifugal pumps in reverse.

Kaplan turbines are axial-flow (propeller) turbines. In the

Kaplan design the blade angles are adjustable to ensure
efficient operation for a range of discharges.

Wells turbines were specifically developed for wave-energy applications. They have the
property that they rotate in the same direction irrespective of the flow direction.

Bulb generators are large-diameter variants of the Kaplan propeller turbine, which are suitable
for the low-head, high-discharge applications in tidal barrages (e.g. La Rance in France). Flow
passes around the bulb, which contains the electrical generator.

The Archimedes screw has been used since ancient

times to raise water. It is widely used in water
treatment plants because it can accommodate
submerged debris. Recently, several devices have
been installed beside weirs in the north of England
to run in reverse as a turbine and generate power.
This one is a mile from my home!

Hydraulics 2 T4-6 David Apsley

3. Pump and System Characteristics

3.1 Pump characteristics

Pump characteristics are the head, 𝐻, input power, 𝐼, and efficiency, 𝜂, as functions of
discharge, 𝑄. The most important is the 𝐻 vs 𝑄 relationship. Typical shapes of these
characteristics are sketched below for centrifugal and axial-flow pumps.

 
I

I
H H

Q Q
centrifugal pump axial-flow pump

Many pumps are variable-speed devices. Given pump characteristics at one rotation rate, 𝑁,
those at different rotation rates may be determined using hydraulic scaling laws (Section 4).

Ideally, one would like to operate the pump:

• as close as possible to the design point (point of maximum efficiency);
• in a region where the 𝐻 vs 𝑄 relationship is steep; (otherwise there are significant
fluctuations in discharge for small changes in head).

3.2 System characteristics

In general the pump has to supply enough energy to:

• lift water through a certain height – the static lift, ℎ𝑠 ;
• overcome losses dependent on the discharge, 𝑄.
Thus the system head or system characteristic is
𝐻 = ℎ𝑠 + ℎlosses

Typically, losses (whether frictional or due to pipe fittings) are proportional to 𝑄 2 , so that the
system characteristic is often quadratic:
𝐻 = ℎ𝑠 + 𝛼𝑄 2

The static lift is often decomposed delivery reservoir

into the rise from sump to the level of
the pump (the suction head, ℎ𝑠1 ) and
that between the pump and the delivery main
delivery point (ℎ𝑠2 ). The first of these static lift
hs
is limited by the maximum suction suction main
height (approximately 10 m, Pump
suction head
corresponding to 1 atmosphere) and
will be discussed later in the context Sump
of cavitation.

Hydraulics 2 T4-7 David Apsley

3.3 Finding the Duty Point

The pump operates at a duty point where the head supplied by the pump precisely matches the
head requirements of the system at the same discharge; i.e. where the pump and system
characteristics intersect.

H
system
characteristic

duty
point
hs pump
characteristic

Example.
A water pump was tested at a rotation rate of 1500 rpm. The following data was obtained.
(𝑄 is quantity of flow; 𝐻 is head of water; 𝜂 is efficiency).

𝑄 (L s–1) 0 10 20 30 40 50
𝐻 (m) 10.0 10.5 10.0 8.5 6.0 2.5
η 0.0 0.40 0.64 0.72 0.64 0.40

It is proposed to use this pump to draw water from an open sump to an elevation 5.5 m above.
The delivery pipe is 20.0 m long and 100 mm diameter and has a friction factor of 0.005.

If operating at 1500 rpm, find:

(a) the maximum discharge that the pump can provide;
(b) the pump efficiency at this discharge;
(c) the input power required.

Answer: (a) 37.7 L s–1; (b) 0.673; (c) 3.67 kW

In practice, it is desirable to run the pump at a speed where the duty point is close to that of
maximum efficiency. To do this we need to determine how the pump characteristic varies with
rotation rate, 𝑁 – see later.

Hydraulics 2 T4-8 David Apsley

3.4 Pumps in Parallel and in Series

Pumps in Parallel

For the combined pump complex: Q1

Q
– same head: 𝐻
– total discharge: 𝑄 = 𝑄1 + 𝑄2 Q2

Advantages of pumps in parallel are: H

• high capacity: permits a large total two pumps in parallel
discharge;
• flexibility: pumps can be brought
in and out of service if the required single pump
discharge varies;
• redundancy: pumping can continue
double the flow
if one is not operating due to failure
or planned maintenance. Q

Pumps in Series

For the combined pump complex:

– same discharge: 𝑄 Q H1 H2
– total head: 𝐻 = 𝐻1 + 𝐻2

Pumps in series may be used to generate high H

overall head, or to provide regular “boosts” along two pumps in series long
pipelines without large pressures at any particular
point.

double
the head

single pump

Hydraulics 2 T4-9 David Apsley

 Example.
A rotodynamic pump, having the characteristics tabulated below, delivers water from a river
at elevation 102 m to a reservoir with a water level of 135 m, through a pipe of length 1 km
and diameter 350 mm. The friction factor of the pipe may be taken as 𝜆 = 0.035 and minor
losses from valves and fittings can be described by a loss coefficient 𝐾 = 9.

𝑄 (m3 s–1 ) 0 0.05 0.10 0.15 0.20

𝐻 (m) 60 58 52 41 25
𝜂 (%) 0 44 65 64 48

(a) Calculate the discharge and head in the pipeline (at the duty point).

If the discharge is to be increased by the installation of a second identical pump:

(b) determine the unregulated discharge and head produced by connecting the pump:
(i) in parallel;
(ii) in series;

(c) determine the power demand at the duty point in the case of parallel operation;

(d) in the case of parallel operation, if the total flow is throttled by a valve to
0.12 m3 s −1 , calculate the head lost across the valve.

Answer: (a) 0.137 m3 s–1 and 44.3 m;

3 –1 3 –1
(b) (i) 0.184 m s and 53.3 m; (ii) 0.193 m s and 55.4 m; (c) 152 kW; (d) 15.5 m.

Hydraulics 2 T4-10 David Apsley

4 Hydraulic Scaling

4.1 Dimensional Analysis

Provided that the mechanical efficiency is the same, the performance of a particular
geometrically-similar family of pumps or turbines (“homologous series”) may be expected to
depend on:
discharge 𝑄 [L3 T −1 ]

pressure change 𝜌𝑔𝐻 [ML−1 T −2 ]

power 𝑃 [ML2 T −3 ] (input for pumps; output for turbines)

rotor diameter 𝐷 [L]

rotation rate 𝑁 [T −1 ]

fluid density 𝜌 [ML−3 ]

fluid viscosity 𝜇 [ML−1 T −1 ]

(Rotor diameter may be replaced by any characteristic length, since geometric similarity
implies that length ratios remain constant. Rotation rate is typically expressed in either rad s–1
or rpm.)

Since there are 7 variables and 3 independent dimensions, Buckingham’s Pi Theorem yields a
relationship between 4 independent groups, which may be rearranged as (exercise):
𝑄 𝑔𝐻 𝑃 𝜌𝑁𝐷2
Π1 = , Π2 = , Π3 = , Π4 = = Re
𝑁𝐷3 𝑁 2 𝐷2 𝜌𝑁 3 𝐷5 𝜇

For fully-turbulent flow the dependence on molecular viscosity, 𝜇, and hence the Reynolds
number, Π4 , vanishes. Then, for geometrically-similar pumps with different sizes, 𝐷, and
rotation rates, 𝑁:
𝑄 𝑄 𝑔𝐻 𝑔𝐻 𝑃 𝑃
( 3) = ( 3) , ( ) = ( ) , ( 3 5) = ( 3 5)
𝑁𝐷 1 𝑁𝐷 2 𝑁 2 𝐷2 1 𝑁 2 𝐷2 2 𝜌𝑁 𝐷 1 𝜌𝑁 𝐷 2

For pumps (input power 𝑃, output power 𝜌𝑔𝑄𝐻), any one of Π1 , Π2 , Π3 may be replaced by
Π1 Π2 𝜌𝑔𝑄𝐻
= = 𝜂  (efficiency)
Π3 𝑃
The reciprocal of this would be used for turbines.

Hydraulics 2 T4-11 David Apsley

 Example.
A ¼-scale model centrifugal pump is tested under a head of 7.5 m at a speed of 500 rpm. It
was found that 7.5 kW was needed to drive the model. Assuming similar mechanical
efficiencies, calculate:
(a) the speed and power required by the prototype when pumping against a head of 44 m;
(b) the ratio of the discharge in the model to that in the prototype.

Answer: (a) 303 rpm and 1710 kW; (b) 𝑄𝑚 /𝑄𝑝 = 0.0258

Hydraulics 2 T4-12 David Apsley

4.2 Change of Speed

For the same pump (i.e. same 𝐷) operating at different speeds 𝑁1 and 𝑁2 the constancy of the
dimensionless groups
𝑄 𝑔𝐻 𝑃
, , , 𝜂
𝑁𝐷3 𝑁 2 𝐷2 𝜌𝑁 3 𝐷5
gives
𝑄 ∝ 𝑁, 𝐻 ∝ 𝑁2, 𝑃 ∝ 𝑁3 , 𝜂 constant
(As an aid to memory, this might be expected, since 𝑄 ∝ velocity, whilst 𝐻 ∝ energy ∝
velocity 2 ).

These are called the hydraulic scaling laws or affinity laws.

Speed-scaling laws for a single pump:

𝑄2 𝑁2 𝐻2 𝑁2 2 𝑃2 𝑁2 3
= , =( ) , =( ) , 𝜂1 = 𝜂2
𝑄1 𝑁1 𝐻1 𝑁1 𝑃1 𝑁1

Given pump characteristics at one speed one can use the hydraulic scaling laws to deduce
characteristics at a different speed.

4.2.1 Finding the Duty Point at a New Pump Speed

Scale each (𝑄, 𝐻) pair on the original characteristic at H

speed 𝑁1 to get the new characteristic at speed 𝑁2 ; i.e. system characteristic
N2
2
𝑁2 𝑁2
𝑄2 = ( ) 𝑄1 , 𝐻2 = ( ) 𝐻1
𝑁1 𝑁1
N1
new duty point
Where this scaled characteristic intercepts the system
curve gives the duty point at speed 𝑁2 .

hs

Hydraulics 2 T4-13 David Apsley

4.2.2 Finding the Pump Speed For a Given Duty Point (Harder)

To find the pump speed for a given

H
N2 system characteristic
discharge or head plot a hydraulic-
scaling curve back from the required
duty point (𝑄2 , 𝐻2 ) on the system
curve, at unknown speed 𝑁2 : N1
new duty point (Q2, H2)
𝐻 𝑄 2
=( )
𝐻2 𝑄2
(Q1, H1)
Very important: the hydraulic hs
scaling curve is not the same as the scaling
curve
system curve, although, since they
are both quadratic, they can be
quite close. Q

Where the hydraulic scaling curve cuts the original characteristic gives a scaled duty point
(𝑄1 , 𝐻1 ) and thence the ratio of pump speeds from either the ratio of discharges or the ratio of
heads:
𝑁2 𝑄2 𝑁2 2 𝐻2
= or ( ) =
𝑁1 𝑄1 𝑁1 𝐻1

Example.
Water from a well is pumped by a centrifugal pump which delivers water to a reservoir in
which the water level is 15.0 m above that in the sump. When the pump speed is 1200 rpm
its pipework has the following characteristics:

Pipework characteristics:
Discharge (L s–1): 20 30 40 50 60
Head loss in pipework (m): 1.38 3.11 5.52 8.63 12.40

Pump characteristics:
Discharge (L s–1): 0 10 20 30 40
Head (m): 22.0 21.5 20.4 19.0 17.4

A variable-speed motor drives the pump.

(a) Plot the graphs of the system and pump characteristics and determine the discharge
at a speed of 1200 rpm.

(b) Find the pump speed in rpm if the discharge is increased to 40 L s–1.

Answer: (a) 32.4 L s–1; (b) 1290 rpm

Hydraulics 2 T4-14 David Apsley

 Example. (Exam 2020)
A variable-speed pump draws water from a reservoir to an elevated tank. The difference in
water levels between the reservoir and the tank is 10 m. The pipe between them has length
𝐿 = 150 m, diameter 𝐷 = 150 mm and friction factor 𝜆 = 0.02. Minor losses can be
lumped into an overall minor loss coefficient, 𝐾, which is unknown. The characteristics of
the pump at the operational speed are given in the table below.

Pump characteristics at 2400 rpm

Discharge (L s–1) 16 26 36 47 57
Head (m) 39.0 36.4 31.3 22.9 12.1
Efficiency (%) 50.8 65.9 70.0 60.5 38.0

(a) Determine the head loss due to friction as a function of discharge, giving numerical
values of the function coefficients and stating the units used for head and discharge.

The discharge at the duty point is 46 L s–1.

(b) Find the pump head and power consumption at the duty point.

(c) Determine the overall minor loss coefficient, 𝐾.

After a rearrangement of facilities, the elevated tank is raised by 15 m and the pipe
lengthened by 70 m. Through careful engineering, minor losses have been significantly
reduced and can be assumed to be negligible (𝐾 ≈ 0).

(d) If the same discharge is to be maintained, find the new rotation speed of the pump.

Answer: (a) ℎ𝑓 = 3264𝑄 2 (ℎ𝑓 in m, 𝑄 in m3 s–1); (b) 23.8 m, 17.3 kW; (c) 20.0;
(d) 2693 rpm.

Solution.
ℎ𝑠 = 10 m
𝐿 = 150 m
𝐷 = 0.15 m
𝜆 = 0.02
𝐾 =?

(a) The head loss due to friction is

𝐿 𝑉2 𝑄
ℎ𝑓 =𝜆 ( ) where 𝑉=
𝐷 2𝑔 π𝐷2 /4
8𝜆𝐿
= 2 5 𝑄2
π 𝑔𝐷
With the data above, this gives
ℎ𝑓 = 3264𝑄 2 (ℎ𝑓 in m, 𝑄 in m3 s−1 )
or
ℎ𝑓 = 0.003264𝑄 2 (ℎ𝑓 in m, 𝑄 in L s−1 )

Hydraulics 2 T4-15 David Apsley

 (b) The discharge 𝑄 = 46 L s −1 (= 0.046 m3 s−1 ). Draw graphs of 𝐻 vs 𝑄 and 𝜂 vs 𝑄 to
determine pump head and efficiency at the duty point.

At the duty point,

𝐻 = 23.8 m
𝜂 = 62.0% (= 0.62)
The power consumption is then

Hydraulics 2 T4-16 David Apsley

 output power
input power =
efficiency
𝜌𝑔𝑄𝐻
=
𝜂
1000 × 9.81 × 0.0460 × 23.8
=
0.62
= 17320 W

Answer: 𝐻 = 23.8 m; 𝑃 = 17.3 kW.

(c) At the duty point, pump head equals the system requirement (static lift + frictional
losses + minor losses):
𝑉2 𝑄
𝐻 = ℎ𝑠 + ℎ𝑓 + 𝐾 ( ) where 𝑉=
2𝑔 π𝐷2 /4

 23.8 = 10 + 6.907 + 0.3454𝐾

 𝐾 = 19.96

Answer: 20.0.

(d) In the new arrangement,

ℎ𝑠 = 25 m
𝐿 = 220 m
𝐷 = 0.15 m
𝜆 = 0.02
𝐾=0

The system head is

𝐿 𝑉2 𝑄
𝐻sys = ℎ𝑠 + 𝜆 ( ) where 𝑉=
𝐷 2𝑔 π𝐷2 /4
8𝜆𝐿
= ℎ𝑠 + 2 5 𝑄 2
π 𝑔𝐷
With the data above and 𝑄 = 0.046 m3 s −1 , this gives
𝐻sys = 35.1 m
This must match the pump head. Hence, at an unknown speed 𝑁2 the pump parameters are
𝑄2 = 46.0 L s−1
𝐻2 = 35.1 m
Plot a speed-scaling curve through this point until it reaches the characteristic for the original
speed 𝑁1 = 2400 rpm. This scaling curve is

Hydraulics 2 T4-17 David Apsley

 𝐻 𝑁 2 𝑄 2
= ( ) = ( )
𝐻2 𝑁2 𝑄2
i.e., for 𝐻 in m and 𝑄 in L s−1,
𝐻 𝑁 2 𝑄 2
=( ) =( )
35.1 𝑁2 46.0
or
𝐻 = 0.0166𝑄 2
This is illustrated below.

The scaling curve meets the 𝑁1 characteristic at (𝑄1 , 𝐻1 ) = (41.0,27.9) and, hence, using
the ratio of discharges (say),
𝑁2 𝑄2
=
𝑁1 𝑄1

𝑁2 46.0
 =
2400 41.0

 𝑁2 = 2693 rpm

Answer: 2693 rpm.

Hydraulics 2 T4-18 David Apsley

4.3 Specific speed

The specific speed (or type number) is a guide to the type of pump or turbine required for a
particular role.

4.3.1 Specific Speed for Pumps

The specific speed, 𝑁𝑠 , is the rotational speed needed to discharge 1 unit of flow against 1 unit
of head. (For what “unit” means in this instance, see below.)

For a given pump, the hydraulic scaling laws give

𝑄 𝑔𝐻
Π1 ≡ = constant , Π 2 = = constant
𝑁𝐷3 𝑁 2 𝐷2
Eliminating 𝐷, and choosing an exponent that will make the combination proportional to 𝑁:
1/4
Π12 𝑄1/2 𝑁
( 3) =
Π2 (𝑔𝐻)3/4
or, since 𝑔 is constant, then at any given (e.g. maximum) efficiency:
𝑄1/2 𝑁
= (dimensional) constant
𝐻 3/4

This constant is the specific speed, 𝑁𝑠 , occurring when 𝑄 and 𝐻 in specified units (see below)
are numerically equal to 1.0:
Specific speed (pump):

𝑄1/2 𝑁
𝑁𝑆 = 3/4
𝐻

Notes.
• The specific speed is a single value calculated at the “normal” operating point (usually
𝑄 and 𝐻 at the maximum efficiency point for the anticipated rotation rate 𝑁).

• With the commonest definition (in the UK and Europe), 𝑁 is in rpm, 𝑄 in m3 s–1, 𝐻 in
m, but this is far from universal, so be careful.

• In principle, the units of 𝑁𝑠 are the same as those of 𝑁, which doesn’t look correct from
the definition but only because that has been shortened from
(1 m3  s −1 )1/2 × 𝑁𝑠 𝑄1/2 𝑁
= 3/4
(1 m)3/4 𝐻

• Because of the omission of 𝑔 the definition of 𝑁𝑠 depends on the units of 𝑄 and 𝐻. A

less-common (but, IMHO, more mathematically correct) quantity is the dimensionless
specific speed 𝐾𝑛 given by

Hydraulics 2 T4-19 David Apsley

 𝑄1/2 𝑁
𝐾𝑛 =
(𝑔𝐻)3/4
If the time units are consistent then 𝐾𝑛 has the same angular units as 𝑁 (rev or rad).

• High specific speed ↔ large discharge / small head (axial-flow device).

Low specific speed ↔ small discharge / large head (centrifugal device).
Approximate ranges of 𝑁𝑠 are (from Hamill, 2011):
Type 𝑁𝑠 (rpm)
Radial (centrifugal) 10 – 70 large head
Mixed flow 70 – 170
Axial > 110 small head

Example.
A pump is needed to operate at 3000 rpm (i.e. 50 Hz) with a head of 6 m and a discharge of
0.2 m3 s–1. By calculating the specific speed, determine what sort of pump is required.

Answer: 𝑁𝑠 = 350 rpm; axial-flow pump

4.3.2 Specific Speed for Turbines

For turbines the output power, 𝑃, is more important than the discharge, 𝑄. The relevant
dimensionless groups are
𝑔𝐻 𝑃
Π2 ≡ , Π3 ≡
𝑁 2 𝐷2 𝜌𝑁 3 𝐷5
Eliminating 𝐷,
1/4
Π23 (𝑃/𝜌)1/2 𝑁
( 5) =
Π2 (𝑔𝐻)5/4
or, since 𝜌 and 𝑔 are usually taken as constant (there does seem to be a presumption that
turbines are always operating in fresh water) then at any given efficiency:
𝑃1/2 𝑁
= (dimensional) constant
𝐻 5/4

The specific speed of a turbine, 𝑁𝑠 , is the rotational speed needed to develop 1 unit of power
for a head of 1 unit. (For what “unit” means in this instance, see below.)

Specific speed (turbine):

𝑃1/2 𝑁
𝑁𝑆 =
𝐻 5/4

Hydraulics 2 T4-20 David Apsley

Notes.
• With the commonest definition (in the UK and Europe), 𝑁 is in rpm, 𝑃 in kW (note),
𝐻 in m, but, again, this is not a universal convention. As with pumps, the units of 𝑁𝑠
are the same as those of 𝑁.

• As with pumps, a less commonly used, but mathematically more acceptable, quantity
is the dimensionless specific speed 𝐾𝑛 , which retains the 𝜌 and 𝑔 dependence:
(𝑃/𝜌)1/2 𝑁
𝐾𝑛 =
(𝑔𝐻)5/4
𝐾𝑛 has the angular units of 𝑁 (revs or radians) – see Massey (2011).

• High specific speed ↔ small head (axial-flow device)

Low specific speed ↔ large head (centrifugal or impulse device).

Approximate ranges are (from Hamill, 2011):

Type 𝑁𝑠 (rpm)
Pelton wheel (impulse) 12 – 60 very large head
Francis turbine (radial-flow) 60 – 500 large head
Kaplan turbine (axial-flow) 280 – 800 small head

Hydraulics 2 T4-21 David Apsley

5. Mechanics of Rotodynamic Devices
u2=r2
5.1 Centrifugal Pump
w2
Fluid enters at the eye of the impeller and flows outward. As it 2 w1
does so it picks up the tangential velocity of the impeller vanes
(or blades) which increases linearly with radius (𝑢 = 𝑟𝜔). At
u1=r1
exit the fluid is expelled nearly tangentially at high velocity.
Kinetic energy is subsequently converted to pressure energy in
the expanding volute.

The analysis makes use of rotational dynamics:
power = torque × angular velocity
torque = rate of change of angular momentum
where angular momentum is given by “tangential component of momentum  radius”.

The absolute velocity of the fluid is the vector sum of:

impeller velocity (tangential)
+
velocity relative to the impeller (parallel to the vanes)

Write:
u for the impeller velocity (𝑢 = 𝑟𝜔)
w for the fluid velocity relative to the impeller
v=u+w for the absolute velocity

The radial component of absolute velocity is determined primarily by the flow rate:
𝑄
𝑣𝑟 =
𝐴 resultant, v
where 𝐴 is the effective outlet area. The tangential part u = r
(also called the whirl velocity) is a combination of impeller
speed (𝑢 = 𝑟𝜔) and tangential component relative to the
vanes: vane
w 
𝑣𝑡 = 𝑢 − 𝑤 cos 𝛽
Only 𝑣𝑡 contributes to the angular momentum.
v
With subscripts 1 and 2 denoting inlet and outlet
respectively,
vt
torque,  𝑇 = 𝜌𝑄(𝑣𝑡2 𝑟2 − 𝑣𝑡1 𝑟1 ) vr
power = 𝑇𝜔 = 𝜌𝑄(𝑣𝑡2 𝑟2 𝜔 − 𝑣𝑡1 𝑟1 𝜔)

But head 𝐻 = power/(𝜌𝑔𝑄) , whilst 𝑟𝜔 = 𝑢.

Hence, we have:

Hydraulics 2 T4-22 David Apsley

Euler’s turbomachinery equation:

1
𝐻= (𝑣 𝑢 − 𝑣𝑡1 𝑢1 )
𝑔 𝑡2 2

The pump is usually designed so that the initial angular momentum is small; i.e. 𝑣𝑡1 ≈ 0. Then
1
𝐻= 𝑣 𝑢
𝑔 𝑡2 2

Effect of Blade Angle

Because in the frame of the impeller the fluid leaves the blades in a direction parallel to their
surface, forward-facing blades would be expected to increase the whirl velocity 𝑣𝑡 whilst
backward-facing blades would diminish it.

Radial and tangential components of velocity:

𝑣𝑟 = 𝑤 sin 𝛽 , 𝑣𝑡 = 𝑢 − 𝑤 cos 𝛽
Eliminating 𝑤:
𝑣𝑡 = 𝑢 − 𝑣𝑟 cot 𝛽 H
Hence, if inlet whirl can be ignored,
𝑢2
𝐻 = (𝑢2 − 𝑣𝑟2 cot 𝛽)
𝑔 β  90 (forward-facing blades)

β = 90
Since 𝑣𝑟 is also 𝑄/𝐴 (where 𝐴 is exit area of the impeller):
β  90 (backward-facing blades)
𝑟2 𝜔 𝑄
𝐻= (𝑟2 𝜔 − cot 𝛽)
𝑔 𝐴 Q
This is of the form 𝐻 = 𝑎 − 𝑏𝑄, where
𝐻 initially decreases with 𝑄 for backward-facing blades (𝛽 < 90°; cot 𝛽 > 0)
𝐻 initially increases with 𝑄 for forward-facing blades (𝛽 > 90°; cot 𝛽 < 0)
This gives rise to the pump characteristics shown. Backward-facing blades are usually
preferred because, although forward-facing blades might be expected to increase whirl velocity
and hence output head, the shape of the characteristic is such that small changes in head cause
large changes in discharge and the pump tends to “hunt” for its operating point (pump surge).

Non-Ideal Behaviour

The above is a very ideal analysis. There are many sources of losses. These include:
• leakage back from the high-pressure volute to the low-pressure impeller eye;
• frictional losses;
• “shock” or flow-separation losses at entry;
• non-uniform flow at inlet and outlet of the impeller;
• cavitation (when the inlet pressure is small).

Hydraulics 2 T4-23 David Apsley

 Example.
A centrifugal pump is required to provide a head of 40 m. The impeller has outlet diameter
0.5 m and inlet diameter 0.25 m and rotates at 1500 rpm. The flow approaches the impeller
radially at 10 m s–1 and the radial velocity falls off as the reciprocal of the radius. Calculate
the required vane angle at the outlet of the impeller.

Answer: 9.69°

5.2 Pelton Wheel

A Pelton wheel is the most common type of impulse turbine. One bucket
or more jets of water impinge on buckets arranged around a v
turbine runner. The deflection of water changes its momentum jet u
and imparts a force to rotate the runner.

The power (per jet) 𝑃 is given by:

power = force (on bucket) × velocity (of bucket)
Force 𝐹 on the bucket is equal and opposite to that on the jet; by the momentum principle:
force (on fluid) = mass flux × change in velocity

Because the absolute velocity of water leaving the bucket is the vector resultant of the runner
velocity (𝑢) and the velocity relative to the bucket, the change in velocity is most easily
established in the frame of reference of the moving
bucket. k(v-u)

Assuming that the relative velocity leaving the v-u 

buckets is 𝑘 times the relative velocity of approach,
𝑣 − 𝑢 (where 𝑘 is slightly less than 1 due to friction):
change in 𝑥 − velocity = −𝑘(𝑣 − 𝑢) cos(180° − 𝜃) − (𝑣 − 𝑢)
= −(𝑣 − 𝑢)(1 − 𝑘 cos 𝜃)
where 𝜃 is the total angle turned (here, greater than 90°). The maximum force would be
obtained if the flow was turned through 180°, but the necessity of deflecting it clear of the next
bucket means that 𝜃 is typically about 165°.

From the momentum principle:

−𝐹 = −𝜌𝑄(𝑣 − 𝑢)(1 − 𝑘 cos 𝜃)
The power transferred in each jet is then
𝑃 = 𝐹𝑢 = 𝜌𝑄(𝑣 − 𝑢)𝑢(1 − 𝑘 cos 𝜃)

The velocity part of the power may be written

1 2 1
(𝑣 − 𝑢)𝑢 = 𝑣 − ( 𝑣 − 𝑢)2
4 2

Hydraulics 2 T4-24 David Apsley

Hence, for a given jet (𝑄 and 𝑣), the power is a maximum when the runner speed 𝑢 is such that
1
𝑢 = 2 𝑣, or the runner speed is half the jet speed. (At this point the absolute velocity leaving
the runner at 180° would be 0 if k = 1, corresponding to a case where all the kinetic energy of
the fluid is transferred to the runner.) In practice, the runner speed 𝑢 is often fixed by the need
to synchronise the generator to the electricity grid, so it is usually the jet which is controlled
(by a spear valve). Because of other losses the speed ratio is usually slightly less than ½, a
typical value being 0.46.

The jet velocity is given by Bernoulli’s equation, with a correction for non-ideal flow:
𝑣 = 𝑐𝑣 √2𝑔𝐻
where 𝐻 is the head upstream of the nozzle (= gross head minus any losses in the pipeline) and
𝑐𝑣 is an orifice coefficient with typical values in the range 0.97– 0.99.

Example.
In a Pelton wheel, 6 jets of water, each with a diameter of 75 mm and carrying a discharge
of 0.15 m3 s–1 impinge on buckets arranged around a 1.5 m diameter Pelton wheel rotating
at 180 rpm. The water is turned through 165° by each bucket and leaves with 90% of the
original relative velocity. Neglecting mechanical and electrical losses within the turbine,
calculate the power output.
Answer: 471 kW

Hydraulics 2 T4-25 David Apsley

6. Cavitation

Cavitation is the formation, growth and rapid collapse of vapour bubbles in flowing liquids.

Bubbles form at low (sub-atmospheric) pressures when the absolute pressure drops to the
vapour pressure and the liquid spontaneously boils. (Bubbles may also arise from dissolved
gases coming out of solution.) When the bubbles are subsequently swept into higher-pressure
regions they collapse very rapidly, with large radial velocities and enormous transient
pressures, potentially leading to surface damage.

Cavitation may cause performance loss, vibration, noise, surface pitting and, occasionally,
major structural damage. Besides the inlet to pumps, the phenomenon is prevalent in marine-
current turbines, ship and submarine propellers and on reservoir spillways.

The best way of preventing cavitation in a pump is to ensure that the inlet (suction) pressure is
not too low. The net positive suction head (NPSH) is the difference between the (stagnation)
pressure head and that corresponding to the vapour pressure/cavitation pressure:
1
(𝑝 + 𝜌𝑉 2 ) − 𝑝cav
NPSH = 2
𝜌𝑔
It is, in length units, the margin by which the stagnation pressure (which would arise when the
fluid is brought to rest) exceeds that at which cavitation may occur.

The net positive suction head must be kept well above zero to allow for further pressure loss in
the impeller. A key measure is the available net positive suction head, NPSHa, which is the
NPSH at pump inlet. This should be compared with the required net positive suction head,
NPSHr, which is a manufacturer-specified quantity with a margin of safety to prevent
cavitation, including an allowance for further pressure loss within the pump and any dissolved
gases.
pump inlet

patm zinlet
The inlet pressure may be determined from Bernoulli’s
equation, measuring 𝑧 relative to the level in the sump:
sump
𝐻pump inlet = 𝐻sump − head loss

𝑝inlet 𝑉 2 𝑝atm
 + 𝑧inlet + = − ℎ𝑓
𝜌𝑔 2𝑔 𝜌𝑔
Hence,
1
(𝑝 + 2 𝜌𝑉 2 ) 𝑝atm
inlet
= − 𝑧inlet − ℎ𝑓
𝜌𝑔 𝜌𝑔
whence
𝑝atm − 𝑝cav
NPSHa = − 𝑧inlet − ℎ𝑓
𝜌𝑔

Hydraulics 2 T4-26 David Apsley

To avoid cavitation one should aim to keep NPSHa as large as possible by:
• keeping 𝑧inlet small or, better still, negative (i.e. below the level of water in the sump;
• keeping ℎ𝑓 small (short, large-diameter pipes).

The first also assists in pump priming.

Hydraulics 2 T4-27 David Apsley